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 * target.h (vec_perm_indices): New typedef.
6 (auto_vec_perm_indices): Likewise.
7 * optabs-query.h: Include target.h
8 (can_vec_perm_p): Take a vec_perm_indices *.
9 * optabs-query.c (can_vec_perm_p): Likewise.
10 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12 * tree-vect-generic.c (lower_vec_perm): Likewise.
13 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14 (vect_grouped_load_supported): Likewise.
15 (vect_shift_permute_load_chain): Likewise.
16 (vect_permute_store_chain): Use auto_vec_perm_indices.
17 (vect_permute_load_chain): Likewise.
18 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
19 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
20 Update uses of can_vec_perm_p.
21 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
22 mode with a number of elements. Take a vec_perm_indices *.
23 (vect_create_epilog_for_reduction): Update accordingly.
24 Use auto_vec_perm_indices.
25 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
26 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
27 (vect_transform_slp_perm_load): Likewise.
28 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
29 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
30 (vect_gen_perm_mask_checked): Likewise.
31 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
32 (vect_gen_perm_mask_checked): Likewise.
33 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
34 (vectorizable_store): Likewise.
35 (vectorizable_load): Likewise.
36 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
37 (vectorizable_bswap): Likewise.
39 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
40 Alan Hayward <alan.hayward@arm.com>
41 David Sherwood <david.sherwood@arm.com>
43 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
44 * tree.c (build_vector): Likewise.
45 (build_vector_from_ctor): Update accordingly.
46 (build_vector_from_val): Likewise.
47 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
48 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
49 * tree-vect-generic.c (add_rshift): Likewise.
50 (expand_vector_divmod): Likewise.
51 (optimize_vector_constructor): Likewise.
52 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
53 (vect_transform_slp_perm_load): Likewise.
54 (vect_schedule_slp_instance): Likewise.
55 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
56 (vectorizable_call): Likewise.
57 (vect_gen_perm_mask_any): Likewise. Add elements in order.
58 * expmed.c (make_tree): Likewise.
59 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
60 a vector passed to build_vector.
61 (fold_convert_const): Likewise.
62 (exact_inverse): Likewise.
63 (fold_ternary_loc): Likewise.
64 (fold_relational_const): Likewise.
65 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
66 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
67 (const_unop): Likewise. Store the reduction accumulator in a
68 variable rather than an array.
69 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
70 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
71 the new vector, rather than constructing it after the input arrays.
72 (native_interpret_vector): Use auto_vec<tree> when building
73 a vector passed to build_vector. Add elements in order.
74 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
75 auto_vec<tree> when building a vector passed to build_vector.
76 (vect_create_epilog_for_reduction): Likewise.
77 (vectorizable_induction): Likewise.
78 (get_initial_def_for_reduction): Likewise. Fix indentation of
80 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
82 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
83 passed to build_vector.
85 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
86 Alan Hayward <alan.hayward@arm.com>
87 David Sherwood <david.sherwood@arm.com>
89 * tree-core.h (tree_base::u): Add an "nelts" field.
90 (tree_vector): Use VECTOR_CST_NELTS as the length.
91 * tree.c (tree_size): Likewise.
92 (make_vector): Initialize VECTOR_CST_NELTS.
93 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
94 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
96 * expr.c (const_vector_mask_from_tree): Consistently use "units"
97 as the number of units, setting it from VECTOR_CST_NELTS.
98 (const_vector_from_tree): Likewise.
99 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
100 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
101 (fold_negate_expr_1): Likewise.
102 (fold_convert_const): Likewise.
103 (const_binop): Likewise. Differentiate the number of output and
105 (const_unop): Likewise.
106 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
107 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
108 in cases that did the opposite.
110 2017-09-14 Richard Biener <rguenther@suse.de>
112 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
115 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
117 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
119 2017-09-14 Jakub Jelinek <jakub@redhat.com>
122 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
123 if and where to split a bb, except for splitting before debug insn
124 sequences followed by non-label real insn. Delete debug insns
125 in between basic blocks.
127 * combine.c (make_compound_operation_int): Formatting fixes.
129 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
130 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
131 * config/netbsd.h (LINK_EH_SPEC): Likewise.
132 * config/sol2.h (LINK_EH_SPEC): Likewise.
133 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
134 * config/s390/linux.h (LINK_SPEC): Likewise.
135 * config/freebsd.h (LINK_EH_SPEC): Likewise.
136 * config/openbsd.h (LINK_EH_SPEC): Likewise.
137 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
138 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
139 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
140 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
141 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
142 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
144 2017-09-13 Jakub Jelinek <jakub@redhat.com>
146 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
148 (ENDFILE_LINUX_SPEC): Likewise.
149 (LINK_EH_SPEC): Likewise.
150 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
151 (LINK_OS_LINUX_SPEC32): Likewise.
152 (LINK_OS_LINUX_SPEC64): Likewise.
153 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
154 (LINK_OS_LINUX_SPEC): Likewise.
156 2017-09-13 Martin Liska <mliska@suse.cz>
159 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
160 CASE_HIGH is NULL_TREE.
162 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * target.def (secondary_memory_needed): New hook.
167 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
168 instead of SECONDARY_MEMORY_NEEDED.
169 (secondary_memory_needed_mode): Likewise.
170 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
171 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
172 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
173 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
174 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
175 * doc/tm.texi: Regenerate.
176 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
177 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
178 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
179 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
180 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
181 * config/i386/i386.c (inline_secondary_memory_needed): Put the
182 mode argument first and change the reg_class arguments to reg_class_t.
183 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
184 Make static. Update the call to inline_secondary_memory_needed.
185 (ix86_register_move_cost): Update the call to
186 inline_secondary_memory_needed.
187 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
188 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
190 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
191 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
193 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
194 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
195 * config/mips/mips.c (mips_secondary_memory_needed): Make static
196 and match hook interface. Add comment from mips.h.
197 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
198 * config/mmix/mmix.md (truncdfsf2): Refer to
199 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
201 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
202 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
203 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
204 (pa_secondary_memory_needed): New function.
205 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
206 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
207 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
208 (pdp11_secondary_memory_needed): Make static and match hook interface.
209 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
210 * config/powerpcspe/powerpcspe-protos.h
211 (rs6000_secondary_memory_needed_ptr): Delete.
212 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
214 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
215 (rs6000_option_override_internal): Assign to
216 targetm.secondary_memory_needed rather than
217 rs6000_secondary_memory_needed_ptr.
218 (rs6000_secondary_memory_needed): Match hook interface.
219 (rs6000_debug_secondary_memory_needed): Likewise.
220 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
221 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
222 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
223 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
224 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
225 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
227 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
228 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
229 (rs6000_option_override_internal): Assign to
230 targetm.secondary_memory_needed rather than
231 rs6000_secondary_memory_needed_ptr.
232 (rs6000_secondary_memory_needed): Match hook interface.
233 (rs6000_debug_secondary_memory_needed): Likewise.
234 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
235 * config/s390/s390.c (s390_secondary_memory_needed): New function.
236 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
237 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
238 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
239 (sparc_secondary_memory_needed): New function.
240 * lra-constraints.c (check_and_process_move): Refer to
241 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
243 (curr_insn_transform): Likewise.
244 (process_alt_operands): Use targetm.secondary_memory_needed
245 instead of TARGET_SECONDARY_MEMORY_NEEDED.
246 (check_secondary_memory_needed_p): Likewise.
247 (choose_split_class): Likewise.
248 * reload.c: Unconditionally include code that was previously
249 conditional on SECONDARY_MEMORY_NEEDED.
250 (push_secondary_reload): Use targetm.secondary_memory_needed
251 instead of TARGET_SECONDARY_MEMORY_NEEDED.
252 (push_reload): Likewise.
253 * reload1.c: Unconditionally include code that was previously
254 conditional on SECONDARY_MEMORY_NEEDED.
255 (choose_reload_regs): Use targetm.secondary_memory_needed
256 instead of TARGET_SECONDARY_MEMORY_NEEDED.
257 (gen_reload): Likewise.
258 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
260 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
261 Alan Hayward <alan.hayward@arm.com>
262 David Sherwood <david.sherwood@arm.com>
264 * target.def (secondary_memory_needed_mode): New hook:
265 * targhooks.c (default_secondary_memory_needed_mode): Declare.
266 * targhooks.h (default_secondary_memory_needed_mode): New function.
267 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
268 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
269 * doc/tm.texi: Regenerate.
270 * lra-constraints.c (check_and_process_move): Use
271 targetm.secondary_memory_needed_mode instead of
272 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
273 (curr_insn_transform): Likewise.
274 * reload.c (get_secondary_mem): Likewise.
275 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
276 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
278 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
279 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
280 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
281 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
282 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
284 * config/powerpcspe/powerpcspe-protos.h
285 (rs6000_secondary_memory_needed_mode): Delete.
286 * config/powerpcspe/powerpcspe.c
287 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
288 (rs6000_secondary_memory_needed_mode): Make static.
289 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
290 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
292 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
294 (rs6000_secondary_memory_needed_mode): Make static.
295 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
296 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
297 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
298 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
299 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
301 (sparc_secondary_memory_needed_mode): New function.
302 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
304 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
306 * config/aarch64/constraints.md (Umq): New constraint.
307 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
309 (mov<mode>): Update condition.
311 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
313 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
314 when bitposition is the same.
316 2017-09-13 Richard Biener <rguenther@suse.de>
318 * dwarf2out.c (output_die_symbol): Remove.
319 (output_die): Do not output a DIEs symbol.
321 2017-09-13 Richard Biener <rguenther@suse.de>
324 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
325 default-def to avoid breaking iterator update with the weird
326 interaction with cgraph_update_edges_for_call_stmt_node.
328 2017-09-13 Richard Biener <rguenther@suse.de>
330 * tree-cfg.c (verify_gimple_assign_binary): Add verification
331 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
332 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
333 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
335 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
337 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
338 Disable pc relative literal load irrespective of
339 TARGET_FIX_ERR_A53_84341 for default.
341 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
343 * config/sparc/sparc.c (output_return): Output the source location of
344 the insn in the delay slot, if any.
345 (output_sibcall): Likewise.
347 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
350 * common.opt (-static-pie): New alias.
351 (shared): Negate static-pie.
352 (-no-pie): Update help text.
354 (static-pie): New option.
355 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
357 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
358 (LINK_EH_SPEC): Likewise.
359 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
360 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
361 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
362 * gcc.c (LINK_COMMAND_SPEC): Likewise.
363 (init_gcc_specs): Likewise.
364 (init_spec): Likewise.
365 (display_help): Update help message for -pie.
366 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
369 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
371 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
372 (movdi_aarch64): Likewise.
373 (movti_aarch64): Likewise.
375 2017-09-12 Simon Wright <simon@pushface.org>
378 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
379 calculation of the minor version, always output as 0.
381 2017-09-12 Jakub Jelinek <jakub@redhat.com>
384 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
385 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
386 on it early, rather than manual conversion late. For
387 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
388 instead of performing manual conversion.
390 2017-09-12 Carl Love <cel@us.ibm.com>
392 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
393 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
395 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
396 VMULOSW): Add definitions.
397 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
398 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
399 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
400 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
401 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
403 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
405 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
407 (movti_aarch64): Likewise.
408 (movdf_aarch64): Likewise.
409 (movtf_aarch64): Likewise.
410 (load_pairdi): Likewise.
411 (store_pairdi): Likewise.
412 (load_pairdf): Likewise.
413 (store_pairdf): Likewise.
414 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
415 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
416 (ldr_got_small_<mode>): Likewise.
417 (ldr_got_small_28k_<mode>): Likewise.
418 (ldr_got_tiny): Likewise.
419 * config/aarch64/iterators.md (ldst_sz): New.
420 (ldpstp_sz): Likewise.
421 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
423 (thunderx_load): Split load_8 to load_16.
424 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
426 (thunderx2t99_storepair_basic): Split store_8 to store_16.
427 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
428 (xgene1_store_pair): Split store_8 to store_16.
429 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
430 (falkor_st_0_st_sd): Split store_8 to store_16.
432 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
434 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
435 and store1/2/3/4 to store_4/8/12/16.
436 * config/aarch64/aarch64.md: Update for rename.
437 * config/arm/arm.md: Likewise.: Likewise.
438 * config/arm/arm.c: Likewise.
439 * config/arm/thumb1.md: Likewise.
440 * config/arm/thumb2.md: Likewise.
441 * config/arm/vfp.md: Likewise.
442 * config/arm/arm-generic.md: Likewise.
443 * config/arm/arm1020e.md: Likewise.
444 * config/arm/arm1026ejs.md: Likewise.
445 * config/arm/arm1136jfs.md: Likewise.
446 * config/arm/arm926ejs.md: Likewise.
447 * config/arm/cortex-a15.md: Likewise.
448 * config/arm/cortex-a17.md: Likewise.
449 * config/arm/cortex-a5.md: Likewise.
450 * config/arm/cortex-a53.md: Likewise.
451 * config/arm/cortex-a57.md: Likewise.
452 * config/arm/cortex-a7.md: Likewise.
453 * config/arm/cortex-a8.md: Likewise.
454 * config/arm/cortex-a9.md: Likewise.
455 * config/arm/cortex-m4.md: Likewise.
456 * config/arm/cortex-m7.md: Likewise.
457 * config/arm/cortex-r4.md: Likewise.
458 * config/arm/exynos-m1.md: Likewise.
459 * config/arm/fa526.md: Likewise.
460 * config/arm/fa606te.md: Likewise.
461 * config/arm/fa626te.md: Likewise.
462 * config/arm/fa726te.md: Likewise.
463 * config/arm/fmp626.md: Likewise.
464 * config/arm/iwmmxt.md: Likewise.
465 * config/arm/ldmstm.md: Likewise.
466 * config/arm/marvell-pj4.md: Likewise.
467 * config/arm/xgene1.md: Likewise.
468 * config/aarch64/thunderx.md: Likewise.
469 * config/aarch64/thunderx2t99.md: Likewise.
470 * config/aarch64/falkor.md: Likewise.
472 2017-09-12 Martin Liska <mliska@suse.cz>
474 * attribs.c (private_lookup_attribute): New function.
475 * attribs.h (private_lookup_attribute): Declared here.
476 (lookup_attribute): Called from this place.
478 2017-09-12 Richard Biener <rguenther@suse.de>
480 PR tree-optimization/82157
481 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
482 stmts with side-effects.
484 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
485 Alan Hayward <alan.hayward@arm.com>
486 David Sherwood <david.sherwood@arm.com>
488 * target.def (hard_regno_nregs): New hook.
489 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
490 * targhooks.h (default_hard_regno_nregs): Declare.
491 * targhooks.c (default_hard_regno_nregs): New function.
492 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
493 (TARGET_HARD_REGNO_NREGS): ...this hook.
494 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
495 (CLASS_MAX_NREGS): Likewise.
496 * doc/tm.texi: Regenerate.
497 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
498 instead of HARD_REGNO_NREGS.
499 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
500 HARD_REGNO_NREGS in the comment.
501 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
502 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
503 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
504 Return an unsigned int.
505 (TARGET_HARD_REGNO_NREGS): Redefine.
506 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
507 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
508 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
509 (arc_hard_regno_nregs): New function.
510 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
511 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
512 (arm_hard_regno_nregs): New function.
513 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
514 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
515 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
516 (TARGET_HARD_REGNO_NREGS): Redefine.
517 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
518 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
519 (HARD_REGNO_NREGS): Delete.
520 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
521 (cr16_hard_regno_nregs): New function.
522 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
523 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
524 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
525 (cris_hard_regno_nregs): New function.
526 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
527 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
528 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
529 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
530 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
531 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
532 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
533 (frv_hard_regno_nregs): Make static. Take and return an
535 (frv_class_max_nregs): Remove outdated copy of documentation.
536 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
537 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
538 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
539 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
540 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
541 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
542 (TARGET_HARD_REGNO_NREGS): Redefine.
543 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
544 (CLASS_MAX_NREGS): Update comment.
545 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
546 (ia64_hard_regno_nregs): New function.
547 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
548 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
549 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
550 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
551 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
553 (m32c_hard_regno_nregs): Likewise. Make static.
554 (TARGET_HARD_REGNO_NREGS): Redefine.
555 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
556 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
557 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
558 (m68k_hard_regno_nregs): New function.
559 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
560 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
561 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
562 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
563 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
564 Take and return an unsigned int.
565 (TARGET_HARD_REGNO_NREGS): Redefine.
566 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
567 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
568 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
569 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
570 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
571 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
572 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
573 (msp430_hard_regno_nregs): Make static. Take and return an
575 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
576 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
577 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
578 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
579 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
580 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
581 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
582 (TARGET_HARD_REGNO_NREGS): Redefine.
583 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
584 (PA_HARD_REGNO_NREGS): ...this.
585 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
586 (PA_HARD_REGNO_NREGS): ...this.
587 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
588 (pa_hard_regno_nregs): New function.
589 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
590 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
591 (pdp11_hard_regno_nregs): New function.
592 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
593 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
594 (rs6000_hard_regno_nregs_hook): New function.
595 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
596 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
597 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
598 Take and return an unsigned int. Move earlier in file.
599 (TARGET_HARD_REGNO_NREGS): Redefine.
600 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
601 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
602 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
603 (rl78_hard_regno_nregs): Make static. Take and return an
605 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
606 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
607 (rs6000_hard_regno_nregs_hook): New function.
608 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
609 * config/rx/rx.c (rx_hard_regno_nregs): New function.
610 (TARGET_HARD_REGNO_NREGS): Redefine.
611 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
612 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
613 instead of HARD_REGNO_NREGS.
614 (s390_hard_regno_nregs): New function.
615 (s390_hard_regno_mode_ok): Add comment from s390.h.
616 (TARGET_HARD_REGNO_NREGS): Redefine.
617 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
618 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
619 (sh_hard_regno_nregs): New function.
620 (sh_pass_in_reg_p): Use it.
621 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
622 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
623 (sparc_hard_regno_nregs): New function.
624 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
625 * config/spu/spu.c (spu_hard_regno_nregs): New function.
626 (spu_function_arg_advance): Use it, supplying a valid register number.
627 (TARGET_HARD_REGNO_NREGS): Redefine.
628 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
629 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
630 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
631 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
632 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
633 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
634 (CLASS_MAX_NREGS): Remove copy of old documentation.
635 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
636 (visium_hard_regno_nregs): New function.
637 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
638 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
639 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
640 xtensa_hard_regno_nregs): New function.
641 * system.h (HARD_REGNO_NREGS): Poison.
643 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
645 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
646 hard_regno_nregs instead of HARD_REGNO_NREGS.
647 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
648 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
649 (c6x_expand_epilogue): Likewise.
650 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
651 (frv_read_iacc_argument): Likewise.
652 * config/sh/sh.c: Include regs.h.
653 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
654 (regs_used): Likewise.
655 (output_stack_adjust): Likewise.
656 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
657 * expmed.c: Include regs.h.
658 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
659 * ree.c: Include regs.h.
660 (combine_reaching_defs): Use hard_regno_nregs instead of
662 (add_removable_extension): Likewise.
664 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
666 * regs.h (hard_regno_nregs): Turn into a function.
667 (end_hard_regno): Update accordingly.
668 * caller-save.c (setup_save_areas): Likewise.
669 (save_call_clobbered_regs): Likewise.
670 (replace_reg_with_saved_mem): Likewise.
671 (insert_restore): Likewise.
672 (insert_save): Likewise.
673 * combine.c (can_change_dest_mode): Likewise.
674 (move_deaths): Likewise.
675 (distribute_notes): Likewise.
676 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
677 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
678 (rs6000_split_multireg_move): Likewise.
679 (rs6000_register_move_cost): Likewise.
680 (rs6000_memory_move_cost): Likewise.
681 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
682 (rs6000_split_multireg_move): Likewise.
683 (rs6000_register_move_cost): Likewise.
684 (rs6000_memory_move_cost): Likewise.
685 * cselib.c (cselib_reset_table): Likewise.
686 (cselib_lookup_1): Likewise.
687 * emit-rtl.c (set_mode_and_regno): Likewise.
688 * function.c (aggregate_value_p): Likewise.
689 * ira-color.c (setup_profitable_hard_regs): Likewise.
690 (check_hard_reg_p): Likewise.
691 (calculate_saved_nregs): Likewise.
692 (assign_hard_reg): Likewise.
693 (improve_allocation): Likewise.
694 (calculate_spill_cost): Likewise.
695 * ira-emit.c (modify_move_list): Likewise.
696 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
697 (ira_hard_reg_in_set_p): Likewise.
698 * ira.c (setup_reg_mode_hard_regset): Likewise.
699 (clarify_prohibited_class_mode_regs): Likewise.
700 (check_allocation): Likewise.
701 * lra-assigns.c (find_hard_regno_for_1): Likewise.
702 (lra_setup_reg_renumber): Likewise.
703 (setup_try_hard_regno_pseudos): Likewise.
704 (spill_for): Likewise.
705 (assign_hard_regno): Likewise.
706 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
707 * lra-constraints.c (in_class_p): Likewise.
708 (lra_constraint_offset): Likewise.
709 (simplify_operand_subreg): Likewise.
710 (lra_constraints): Likewise.
711 (split_reg): Likewise.
712 (split_if_necessary): Likewise.
713 (invariant_p): Likewise.
714 (inherit_in_ebb): Likewise.
715 * lra-lives.c (process_bb_lives): Likewise.
716 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
717 (get_hard_regs): Likewise.
718 (do_remat): Likewise.
719 * lra-spills.c (assign_spill_hard_regs): Likewise.
720 * mode-switching.c (create_pre_exit): Likewise.
721 * postreload.c (reload_combine_recognize_pattern): Likewise.
722 * recog.c (peep2_find_free_register): Likewise.
723 * regcprop.c (kill_value_regno): Likewise.
724 (set_value_regno): Likewise.
725 (copy_value): Likewise.
726 (maybe_mode_change): Likewise.
727 (find_oldest_value_reg): Likewise.
728 (copyprop_hardreg_forward_1): Likewise.
729 * regrename.c (check_new_reg_p): Likewise.
730 (regrename_do_replace): Likewise.
731 * reload.c (push_reload): Likewise.
732 (combine_reloads): Likewise.
733 (find_dummy_reload): Likewise.
734 (operands_match_p): Likewise.
735 (find_reloads): Likewise.
736 (find_equiv_reg): Likewise.
737 (reload_adjust_reg_for_mode): Likewise.
738 * reload1.c (count_pseudo): Likewise.
739 (count_spilled_pseudo): Likewise.
740 (find_reg): Likewise.
741 (clear_reload_reg_in_use): Likewise.
742 (free_for_value_p): Likewise.
743 (allocate_reload_reg): Likewise.
744 (choose_reload_regs): Likewise.
745 (reload_adjust_reg_for_temp): Likewise.
746 (emit_reload_insns): Likewise.
747 (delete_output_reload): Likewise.
748 * rtlanal.c (subreg_get_info): Likewise.
749 * sched-deps.c (sched_analyze_reg): Likewise.
750 * sel-sched.c (init_regs_for_mode): Likewise.
751 (mark_unavailable_hard_regs): Likewise.
752 (choose_best_reg_1): Likewise.
753 (verify_target_availability): Likewise.
754 * valtrack.c (dead_debug_insert_temp): Likewise.
755 * var-tracking.c (track_loc_p): Likewise.
756 (emit_note_insn_var_location): Likewise.
757 * varasm.c (make_decl_rtl): Likewise.
758 * reginfo.c (choose_hard_reg_mode): Likewise.
759 (init_reg_modes_target): Refer directly to
760 this_target_regs->x_hard_regno_nregs.
762 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
764 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
765 instead of hard_regno_nregs.
767 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
769 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
770 end_hard_regno instead of hard_regno_nregs.
771 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
772 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
773 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
774 * ira-color.c (improve_allocation): Likewise.
775 * lra-assigns.c (find_hard_regno_for_1): Likewise.
776 * lra-lives.c (mark_regno_live): Likewise.
777 (mark_regno_dead): Likewise.
778 * lra-remat.c (operand_to_remat): Likewise.
779 * lra.c (collect_non_operand_hard_regs): Likewise.
780 * postreload.c (reload_combine_note_store): Likewise.
781 (move2add_valid_value_p): Likewise.
782 * reload.c (regno_clobbered_p): Likewise.
784 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
786 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
788 * config/v850/v850.c (v850_reorg): Likewise.
789 * reload.c (refers_to_regno_for_reload_p): Likewise.
790 (find_equiv_reg): Likewise.
791 * reload1.c (reload_reg_reaches_end_p): Likewise.
793 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
795 * caller-save.c (add_used_regs): Use REG_NREGS instead of
797 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
798 * config/arm/arm.c (output_move_neon): Likewise.
799 (arm_attr_length_move_neon): Likewise.
800 (neon_split_vcombine): Likewise.
801 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
802 (c6x_mark_reg_written): Likewise.
803 (c6x_dwarf_register_span): Likewise.
804 * config/i386/i386.c (ix86_save_reg): Likewise.
805 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
806 (rws_access_reg): Likewise.
807 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
808 * mode-switching.c (create_pre_exit): Likewise.
809 * ree.c (combine_reaching_defs): Likewise.
810 (add_removable_extension): Likewise.
811 * regcprop.c (find_oldest_value_reg): Likewise.
812 (copyprop_hardreg_forward_1): Likewise.
813 * reload.c (reload_inner_reg_of_subreg): Likewise.
814 (push_reload): Likewise.
815 (combine_reloads): Likewise.
816 (find_dummy_reload): Likewise.
817 (reload_adjust_reg_for_mode): Likewise.
818 * reload1.c (find_reload_regs): Likewise.
819 (forget_old_reloads_1): Likewise.
820 (reload_reg_free_for_value_p): Likewise.
821 (reload_adjust_reg_for_temp): Likewise.
822 (emit_reload_insns): Likewise.
823 (delete_output_reload): Likewise.
824 * sel-sched.c (choose_best_reg_1): Likewise.
825 (choose_best_pseudo_reg): Likewise.
827 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
828 Alan Hayward <alan.hayward@arm.com>
829 David Sherwood <david.sherwood@arm.com>
831 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
832 * target.def (slow_unaligned_access): New hook.
833 * targhooks.h (default_slow_unaligned_access): Declare.
834 * targhooks.c (default_slow_unaligned_access): New function.
835 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
836 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
837 * doc/tm.texi: Regenerate.
838 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
839 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
840 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
842 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
843 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
845 (rs6000_slow_unaligned_access): New function.
846 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
847 (expand_block_compare): Likewise.
848 (expand_strn_compare): Likewise.
849 (rs6000_rtx_costs): Likewise.
850 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
851 (riscv_slow_unaligned_access): Likewise.
852 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
853 (riscv_slow_unaligned_access_p): ...this and make static.
854 (riscv_option_override): Update accordingly.
855 (riscv_slow_unaligned_access): New function.
856 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
857 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
858 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
859 (rs6000_slow_unaligned_access): New function.
860 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
861 (rs6000_rtx_costs): Likewise.
862 * config/rs6000/rs6000-string.c (expand_block_compare)
863 (expand_strn_compare): Use targetm.slow_unaligned_access instead
864 of SLOW_UNALIGNED_ACCESS.
865 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
866 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
867 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
868 of SLOW_UNALIGNED_ACCESS.
869 * expmed.c (simple_mem_bitfield_p): Likewise.
870 * expr.c (alignment_for_piecewise_move): Likewise.
871 (emit_group_load_1): Likewise.
872 (emit_group_store): Likewise.
873 (copy_blkmode_from_reg): Likewise.
874 (emit_push_insn): Likewise.
875 (expand_assignment): Likewise.
876 (store_field): Likewise.
877 (expand_expr_real_1): Likewise.
878 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
879 * lra-constraints.c (simplify_operand_subreg): Likewise.
880 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
881 * gimple-ssa-store-merging.c: Likewise in block comment at start
883 * tree-ssa-strlen.c: Include target.h.
884 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
885 of SLOW_UNALIGNED_ACCESS.
886 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
888 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
890 PR rtl-optimization/82185
891 * expmed.c (emit_store_flag_int): Only test tem if it has been
894 2017-09-12 Richard Biener <rguenther@suse.de>
897 * match.pd ((FTYPE) N CMP CST): Fix typo.
899 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
901 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
903 (mips_near_type_p): Add 'short_call' attribute as a synonym
905 * doc/extend.texi (short_call): Document new function attribute.
907 2017-09-12 Jakub Jelinek <jakub@redhat.com>
910 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
911 assertion check that in the condition.
912 (get_atomic_generic_size): Likewise. Before testing if parameter
913 has pointer type, if it has array type, call for C++
914 default_conversion to perform array-to-pointer conversion.
916 2017-09-12 Richard Biener <rguenther@suse.de>
918 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
919 for operations we cannot scalarize.
921 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
923 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
924 vectors heap vectors. Clean up comments.
925 Make visited_bbs a reference.
926 (profitable_jump_thread_path): Make GC
927 vectors heap vectors. Clean up comments.
929 (convert_and_register_jump_thread_path): Make GC vectors heap
931 (check_subpath_and_update_thread_path): Same. Clean up comments.
932 Make visited_bbs a reference.
933 (handle_phi): Abstract common code to to
934 register_jump_thread_path_if_profitable.
935 Rename VAR_BB to DEF_BB.
937 Make GC vectors heap vectors.
938 Make visited_bbs a reference.
939 (handle_assignment): Same.
940 (register_jump_thread_path_if_profitable): New.
941 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
943 Make GC vectors heap vectors. Clean up comments.
944 Make visited_bbs a reference.
945 (find_jump_threads_backwards): Make visited_bbs live in the stack.
946 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
949 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
952 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
953 words of E_DImode object are reachable by xtensa_uimm8x4 access.
955 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
957 Revert r251800 and r251799.
959 2017-09-11 Martin Jambor <mjambor@suse.cz>
962 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
963 arguments in advance.
964 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
965 use it to find predecessor edges.
966 (naive_outof_ssa): Collect vector of predecessors.
968 2017-09-08 Jason Merrill <jason@redhat.com>
970 PR c++/70029 - ICE with ref-qualifier and -flto
971 * langhooks.h (struct lang_hooks_for_types): Add
972 copy_lang_qualifiers.
973 * attribs.c (build_type_attribute_qual_variant): Use it.
974 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
976 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
977 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
979 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
982 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
983 (*mulsi3_sp64): New instruction.
984 (mulsi3): New expander.
986 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
988 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
990 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
992 * sancov.c: Include memmodel.h.
994 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
997 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1000 2017-09-07 Carl Love <cel@us.ibm.com>
1002 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1003 the sldi instruction.
1005 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1007 * sancov.c: Include tm_p.h.
1009 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1012 * output.h (switch_to_other_text_partition): New declaration.
1013 * varasm.c (switch_to_other_text_partition): New function.
1014 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1015 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1016 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1017 to the other text partition before emitting LCL label and switch back
1018 after emitting the word after it.
1020 2017-09-07 Richard Biener <rguenther@suse.de>
1022 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1023 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1024 critical edges here, after loop init.
1025 (pass_data_pre): Remove PROP_no_crit_edges flags.
1026 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1027 for valueization of call args to avoid leaking VN_TOP.
1028 (visit_use): Assert we do not visit default defs.
1029 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1030 Use error_mark_node to more easily detect leaking VN_TOP.
1031 All default-defs are varying, not VN_TOP. Mark them visited.
1032 (run_scc_vn): Make code match comment.
1034 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1036 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1037 OPTION_MASK_FLOAT128_KEYWORD.
1038 (POWERPC_MASKS): Likewise.
1039 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1040 support for the -mfloat128-type option, and make -mfloat128
1041 default on PowerPC Linux systems. Define or undefine
1042 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1043 Define __float128 to be __ieee128 if IEEE 128-bit support is
1044 enabled, or undefine it.
1045 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1046 Delete defining __FLOAT128_TYPE__.
1047 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1048 -mfloat128-type option and make -mfloat128 default on PowerPC
1050 (TARGET_FLOAT128_TYPE): Likewise.
1051 (-mfloat128-type): Likewise.
1052 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1053 Delete the -mfloat128-type option and make -mfloat128 default on
1054 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1055 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1056 128-bit floating point is enabled. Change tests from using
1057 -mfloat128-type to -mfloat128.
1058 (rs6000_mangle_type): Use the correct mangling for the __float128
1059 type even if normal long double is restricted to 64-bits.
1060 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1062 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1063 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1064 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1066 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1067 documentation for -mfloat128.
1069 2017-09-06 Olivier Hainque <hainque@adacore.com>
1071 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1073 2017-09-06 Wish Wu <wishwu007@gmail.com>
1074 Jakub Jelinek <jakub@redhat.com>
1076 * asan.c (initialize_sanitizer_builtins): Add
1077 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1078 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1079 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1080 BT_FN_VOID_UINT64_PTR variables.
1081 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1082 (BT_FN_VOID_UINT16_UINT16): Likewise.
1083 (BT_FN_VOID_UINT32_UINT32): Likewise.
1084 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1085 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1086 (BT_FN_VOID_UINT64_PTR): Likewise.
1087 * common.opt (flag_sanitize_coverage): New variable.
1088 (fsanitize-coverage=trace-pc): Remove.
1089 (fsanitize-coverage=): Add.
1090 * flag-types.h (enum sanitize_coverage_code): New enum.
1091 * fold-const.c (fold_range_test): Disable non-short-circuit
1092 optimization if flag_sanitize_coverage.
1093 (fold_truth_andor): Likewise.
1094 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1095 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1096 (coverage_sanitizer_opts): New array.
1097 (get_closest_sanitizer_option): Add OPTS argument, handle also
1098 OPT_fsanitize_coverage_.
1099 (parse_sanitizer_options): Adjusted to also handle
1100 OPT_fsanitize_coverage_.
1101 (common_handle_option): Add OPT_fsanitize_coverage_.
1102 * sancov.c (instrument_comparison, instrument_switch): New function.
1103 (sancov_pass): Add trace-cmp support.
1104 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1105 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1106 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1107 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1108 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1109 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1110 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1111 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1112 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1113 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1115 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1117 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1118 error. Only quit immediately if parsing is complete.
1119 (BEGIN): Initialize fatal_err and parse_done.
1120 (begin fpu, end fpu): Check number of arguments.
1121 (begin arch, end arch): Likewise.
1122 (begin cpu, end cpu): Likewise.
1123 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1124 (optalias): Likewise.
1126 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1128 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1129 * config/arm/arm-isa.h: Delete. Move definitions to ...
1130 * arm-cpus.in: ... here. Use new feature and fgroup values.
1131 * config/arm/arm.c (arm_option_override): Use lower case for feature
1133 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1134 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1135 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1137 (print_isa_bits_for): New function.
1138 (gen_isa): New function.
1139 (gen_comm_data): Use print_isa_bits_for.
1140 (define feature): New keyword.
1141 (define fgroup): New keyword.
1142 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1143 (arm-isa.h): Add rule to generate file.
1144 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1145 case for feature bit names.
1147 2017-09-06 Richard Biener <rguenther@suse.de>
1149 * tree-ssa-pre.c (NECESSARY): Remove.
1150 (create_expression_by_pieces): Do not touch pass-local flags.
1151 (insert_into_preds_of_block): Likewise.
1152 (do_pre_regular_insertion): Likewise.
1153 (eliminate_insert): Likewise.
1154 (eliminate_dom_walker::before_dom_children): Likewise.
1155 (fini_eliminate): Do not look at inserted_exprs.
1156 (mark_operand_necessary): Remove.
1157 (remove_dead_inserted_code): Replace with simple work-list
1158 algorithm based on inserted_exprs and SSA uses.
1159 (pass_pre::execute): Re-order fini_eliminate and
1160 remove_dead_inserted_code.
1162 2017-09-06 Olivier Hainque <hainque@adacore.com>
1164 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1165 for VxWorks 7. Adjust surrounding comments.
1167 2017-09-06 Richard Biener <rguenther@suse.de>
1169 * gimple-ssa-strength-reduction.c
1170 (find_candidates_dom_walker::before_dom_children): Also allow
1173 2017-09-06 Richard Biener <rguenther@suse.de>
1175 PR tree-optimization/82108
1176 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1177 for gap in the non-permutation SLP case.
1179 2017-09-06 Martin Jambor <mjambor@suse.cz>
1181 PR tree-optimization/82078
1182 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1183 add_access_to_work_queue...
1184 (build_accesses_from_assign): ...here.
1185 (propagate_all_subaccesses): Make sure racc is the group
1186 representative, if there is one.
1188 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1191 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1194 2017-09-06 Richard Biener <rguenther@suse.de>
1196 * gimple-ssa-strength-reduction.c
1197 (find_candidates_dom_walker::before_doom_children): Use a
1198 type and not a mode check.
1200 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1203 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1205 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1207 2017-09-05 Jeff Law <law@redhat.com>
1209 PR tree-optimization/64910
1210 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1211 cases where we have 3 or more operands.
1213 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1216 * omp-low.c (lower_omp_for): Recompute tree invariant if
1217 gimple_omp_for_initial/final is ADDR_EXPR.
1220 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1221 into gimple val before gimplification fo the COND_EXPR.
1223 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1225 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1226 REGION_COPY argument.
1227 (thread_through_all_blocks): Remove unused argument to
1228 duplicate_thread_path.
1230 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1231 Alan Hayward <alan.hayward@arm.com>
1232 David Sherwood <david.sherwood@arm.com>
1234 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1235 Take a scalar_mode rather than a machine_mode.
1236 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1237 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1238 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1239 (aarch64_gen_adjusted_ldpstp): Likewise.
1240 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1242 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1243 Alan Hayward <alan.hayward@arm.com>
1244 David Sherwood <david.sherwood@arm.com>
1246 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1247 Take a scalar_int_mode instead of a machine_mode.
1248 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1249 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1250 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1251 (aarch64_simd_attr_length_rglist): Delete.
1252 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1253 a scalar_int_mode instead of a machine_mode.
1254 (aarch64_add_offset): Likewise.
1255 (aarch64_internal_mov_immediate): Likewise
1256 (aarch64_add_constant_internal): Likewise.
1257 (aarch64_add_constant): Likewise.
1258 (aarch64_movw_imm): Likewise.
1259 (aarch64_rtx_arith_op_extract_p): Likewise.
1260 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1261 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1262 Remove assert that the mode isn't a vector.
1263 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1264 (aarch64_expand_mov_immediate): Update calls after above changes.
1265 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1266 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1267 (aarch64_move_imm): Likewise.
1268 (aarch64_can_const_movi_rtx_p): Likewise.
1269 (aarch64_strip_extend): Likewise.
1270 (aarch64_extr_rtx_p): Likewise.
1271 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1272 a CONST_INT when the mode parameter is VOIDmode.
1273 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1275 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1277 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1278 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1279 (bitwise_type_for_mode): Update accordingly.
1281 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1283 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1284 * stor-layout.c (mode_for_size_tree): Likewise.
1285 (mode_for_array): Update accordingly.
1286 (layout_decl): Likewise.
1287 (compute_record_mode): Likewise. Only set the mode once.
1289 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1291 * target.def (get_mask_mode): Change return type to opt_mode.
1293 * doc/tm.texi: Regenerate.
1294 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1295 * targhooks.c (default_get_mask_mode): Likewise.
1296 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1297 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1298 targetm.get_mask_mode.
1299 * tree.c (build_truth_vector_type): Likewise.
1301 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1303 * machmode.h (mode_for_vector): Return an opt_mode.
1304 * stor-layout.c (mode_for_vector): Likewise.
1305 (mode_for_int_vector): Update accordingly.
1306 (layout_type): Likewise.
1307 * config/i386/i386.c (emit_memmov): Likewise.
1308 (ix86_expand_set_or_movmem): Likewise.
1309 (ix86_expand_vector_init): Likewise.
1310 (ix86_get_mask_mode): Likewise.
1311 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1313 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1314 * expmed.c (extract_bit_field_1): Likewise.
1315 * expr.c (expand_expr_real_2): Likewise.
1316 * optabs-query.c (can_vec_perm_p): Likewise.
1317 (can_vec_mask_load_store_p): Likewise.
1318 * optabs.c (expand_vec_perm): Likewise.
1319 * targhooks.c (default_get_mask_mode): Likewise.
1320 * tree-vect-stmts.c (vectorizable_store): Likewise.
1321 (vectorizable_load): Likewise.
1322 (get_vectype_for_scalar_type_and_size): Likewise.
1324 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1326 * machmode.h (mode_for_int_vector): New function.
1327 * stor-layout.c (mode_for_int_vector): Likewise.
1328 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1329 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1330 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1331 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1332 (s390_expand_vcond): Likewise.
1334 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1336 * machmode.h (opt_machine_mode): New type.
1337 (opt_mode<T>): Allow construction from anything that can be
1339 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1340 (mode_for_size): Return an opt_machine_mode.
1341 * stor-layout.c (mode_for_size): Likewise.
1342 (mode_for_size_tree): Update call accordingly.
1343 (bitwise_mode_for_mode): Likewise.
1344 (make_fract_type): Likewise.
1345 (make_accum_type): Likewise.
1346 * caller-save.c (replace_reg_with_saved_mem): Update call
1348 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1349 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1350 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1351 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1352 * expmed.c (extract_bit_field_1): Likewise.
1353 * reload.c (get_secondary_mem): Likewise.
1354 * varasm.c (assemble_integer): Likewise.
1355 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1358 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1360 * machmode.h (decimal_float_mode_for_size): New function.
1361 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1362 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1363 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1364 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1365 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1366 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1368 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1370 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1371 (get_builtin_sync_mode): Likewise.
1372 (expand_ifn_atomic_compare_exchange): Likewise.
1373 (expand_builtin_atomic_clear): Likewise.
1374 (expand_builtin_atomic_test_and_set): Likewise.
1375 (fold_builtin_atomic_always_lock_free): Likewise.
1376 * calls.c (compute_argument_addresses): Likewise.
1377 (emit_library_call_value_1): Likewise.
1378 (store_one_arg): Likewise.
1379 * combine.c (combine_instructions): Likewise.
1380 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1381 * config/arm/arm.c (arm_function_value): Likewise.
1382 (aapcs_allocate_return_reg): Likewise.
1383 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1384 * config/i386/i386.c (construct_container): Likewise.
1385 (ix86_gimplify_va_arg): Likewise.
1386 (ix86_expand_sse_cmp): Likewise.
1387 (emit_memmov): Likewise.
1388 (emit_memset): Likewise.
1389 (expand_small_movmem_or_setmem): Likewise.
1390 (ix86_expand_pextr): Likewise.
1391 (ix86_expand_pinsr): Likewise.
1392 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1393 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1395 * config/mips/mips.c (mips_function_value_1) Likewise.
1396 (mips_block_move_straight): Likewise.
1397 (mips_expand_ins_as_unaligned_store): Likewise.
1398 * config/powerpcspe/powerpcspe.c
1399 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1400 (rs6000_darwin64_record_arg_flush): Likewise.
1401 * config/rs6000/rs6000.c
1402 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1403 (rs6000_darwin64_record_arg_flush): Likewise.
1404 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1405 (sparc_function_value_1): Likewise.
1406 * config/spu/spu.c (adjust_operand): Likewise.
1407 (spu_emit_branch_or_set): Likewise.
1408 (arith_immediate_p): Likewise.
1409 * emit-rtl.c (gen_lowpart_common): Likewise.
1410 * expr.c (expand_expr_real_1): Likewise.
1411 * function.c (assign_parm_setup_block): Likewise.
1412 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1413 * reload1.c (alter_reg): Likewise.
1414 * stor-layout.c (mode_for_vector): Likewise.
1415 (layout_type): Likewise.
1417 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1419 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1420 (spu_convert_move): Likewise.
1421 * lower-subreg.c (resolve_simple_move): Likewise.
1423 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1426 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1427 define_insn to a define_expand.
1428 (altivec_vsum2sws_direct): New define_insn.
1429 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1431 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1433 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1436 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1441 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1442 realign stack if stack alignment needed is less than incoming
1445 2017-09-05 Marek Polacek <polacek@redhat.com>
1448 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1451 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1453 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1455 2017-09-05 Richard Biener <rguenther@suse.de>
1457 PR tree-optimization/82084
1458 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1460 2017-09-05 Richard Biener <rguenther@suse.de>
1462 PR tree-optimization/82102
1463 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1465 2017-09-05 Martin Liska <mliska@suse.cz>
1467 PR tree-optimization/82032
1468 * tree-cfg.c (generate_range_test): New function.
1469 * tree-cfg.h (generate_range_test): Declared here.
1470 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1471 (cleanup_control_expr_graph): Use it.
1472 * tree-switch-conversion.c (try_switch_expansion): Remove
1474 (emit_case_nodes): Use generate_range_test.
1476 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1479 * config/i386/i386.md (*<btsc><mode>_mask): Add
1480 TARGET_USE_BT to insn constraint.
1481 (*btr<mode>_mask): Ditto.
1483 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1485 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1486 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1488 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1491 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1492 TARGET_NEON and TARGET_IWMMXT.
1493 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1494 TARGET_NEON and TARGET_IWMMXT.
1495 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1497 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1499 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1500 (ix86_rewrite_tls_address): Ditto.
1501 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1502 (ix86_rewrite_tls_address_1): Ditto.
1503 (ix86_rewrite_tls_address): Ditto.
1504 * config/i386/predicates.md (tls_address_pattern): New predicate.
1505 * config/i386/i386.md (TLS address splitter): New splitter.
1507 2017-09-04 Richard Biener <rguenther@suse.de>
1509 PR tree-optimization/82084
1510 * fold-const.h (can_native_encode_string_p): Declare.
1511 * fold-const.c (can_native_encode_string_p): Factor out from ...
1512 (native_encode_string): ... here.
1513 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1514 vectorizing stores from constants we later cannot handle.
1516 2017-09-04 Marek Polacek <polacek@redhat.com>
1519 * doc/invoke.texi: Update -Wtautological-compare documentation.
1521 2017-09-04 Jeff Law <law@redhat.com>
1523 PR tree-optimization/64910
1524 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1525 swap the first and last operand if the last is a constant.
1527 2017-09-04 Marek Polacek <polacek@redhat.com>
1530 * convert.c (do_narrow): When sanitizing signed integer overflows,
1531 bail out for signed types.
1532 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1534 2017-09-04 Richard Biener <rguenther@suse.de>
1536 PR tree-optimization/82060
1537 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1538 Move devirtualization after stmt folding and before EH/AB/noreturn
1539 cleanup to get the stmt refs canonicalized. Use a bool instead
1540 of gimple_modified_p since that doesn't work for NOPs. Schedule
1541 NOPs generated by folding for removal.
1543 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1544 Alan Hayward <alan.hayward@arm.com>
1545 David Sherwood <david.sherwood@arm.com>
1547 * coretypes.h (pad_direction): New enum.
1548 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1549 (FUNCTION_ARG_PADDING): Likewise.
1550 * target.def (function_arg_padding): New hook.
1551 * targhooks.h (default_function_arg_padding): Declare.
1552 * targhooks.c (default_function_arg_padding): New function.
1553 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1554 (TARGET_FUNCTION_ARG_PADDING): ...this.
1555 * doc/tm.texi: Regenerate.
1556 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1557 instead of direction.
1558 (compute_argument_addresses): Likewise.
1559 (load_register_parameters): Likewise.
1560 (emit_library_call_value_1): Likewise.
1561 (store_one_arg): Use targetm.calls.function_arg_padding instead
1562 of FUNCTION_ARG_PADDING.
1563 (must_pass_in_stack_var_size_or_pad): Likewise.
1564 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1565 (emit_group_store): Likewise.
1566 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1567 instead of FUNCTION_ARG_PADDING.
1568 (emit_push_insn): Likewise, and propagate enum change throughout
1570 * function.h (direction): Delete.
1571 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1573 * function.c (assign_parm_find_stack_rtl): Likewise.
1574 (assign_parm_setup_block_p): Likewise.
1575 (assign_parm_setup_block): Likewise.
1576 (gimplify_parameters): Likewise.
1577 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1578 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1580 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1581 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1582 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1583 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1584 (aarch64_function_arg_padding): ...this new function.
1585 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1586 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1587 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1588 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1589 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1590 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1591 (arm_pad_arg_upward): Replace with...
1592 (arm_function_arg_padding): ...this new function.
1593 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1595 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1596 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1597 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1598 (ia64_hpux_function_arg_padding): Replace with...
1599 (ia64_function_arg_padding): ...this new function. Use pad_direction
1600 instead of direction. Check for TARGET_HPUX.
1601 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1602 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1603 (iq2000_function_arg_padding): New function.
1604 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1605 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1606 (mips_function_arg_padding): ...this new function.
1607 (mips_pad_reg_upward): Update accordingly.
1608 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1609 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1610 targetm.calls.function_arg_padding.
1611 (FUNCTION_ARG_PADDING): Delete.
1612 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1613 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1614 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1615 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1616 (nios2_block_reg_padding): Return pad_direction instead of direction.
1617 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1618 instead of direction.
1619 (nios2_function_arg_padding): Likewise. Make static.
1620 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1621 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1622 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1623 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1624 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1625 (pa_function_arg_padding): Make static. Return pad_direction instead
1627 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1628 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1629 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1630 instead of direction. Use targetm.calls.function_arg_padding.
1631 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1632 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1633 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1634 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1635 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1637 (function_arg_padding): Rename to...
1638 (rs6000_function_arg_padding): ...this. Make static. Return
1639 pad_direction instead of direction.
1640 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1641 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1642 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1643 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1644 instead of direction. Use targetm.calls.function_arg_padding.
1645 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1646 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1647 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1648 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1649 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1650 (function_arg_padding): Rename to...
1651 (rs6000_function_arg_padding): ...this. Make static. Return
1652 pad_direction instead of direction.
1653 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1654 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1655 * config/s390/s390.c (s390_function_arg_padding): New function.
1656 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1657 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1658 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1659 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1660 (function_arg_padding): Rename to...
1661 (sparc_function_arg_padding): ...this. Make static. Return
1662 pad_direction instead of direction.
1663 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1664 * config/spu/spu.c (spu_function_arg_padding): New function.
1665 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1666 * system.h (FUNCTION_ARG_PADDING): Poison.
1668 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1669 Alan Hayward <alan.hayward@arm.com>
1670 David Sherwood <david.sherwood@arm.com>
1672 * target.def (modes_tieable_p): New hook.
1673 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1674 (TARGET_MODES_TIEABLE_P): ...this.
1675 * doc/tm.texi.in: Regenerate.
1676 * hooks.h (hook_bool_mode_mode_true): Declare.
1677 * hooks.c (hook_bool_mode_mode_true): New function.
1678 * combine.c (subst): Use targetm.modes_tieable_p instead of
1680 * dse.c (find_shift_sequence): Likewise.
1681 * expmed.c (extract_low_bits): Likewise.
1682 * lower-subreg.c: Include target.h.
1683 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1685 * rtlanal.c (rtx_cost): Likewise.
1686 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1687 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1688 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1689 (TARGET_MODES_TIEABLE_P): Redefine.
1690 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1691 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1692 (TARGET_MODES_TIEABLE_P): Redefine.
1693 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1694 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1695 (arc_modes_tieable_p): New function.
1696 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1697 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1698 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1699 (arm_modes_tieable_p): Make static.
1700 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1701 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1702 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1703 (TARGET_MODES_TIEABLE_P): Redefine.
1704 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1705 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1706 (TARGET_MODES_TIEABLE_P): Redefine.
1707 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1708 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1709 (cr16_modes_tieable_p): New function.
1710 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1711 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1712 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1713 (TRULY_NOOP_TRUNCATION): Update comment.
1714 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1715 (TRULY_NOOP_TRUNCATION): Update comment.
1716 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1717 (frv_modes_tieable_p): New function.
1718 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1719 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1720 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1721 (TARGET_MODES_TIEABLE_P): Redefine.
1722 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1723 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1724 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1725 (TARGET_MODES_TIEABLE_P): Redefine.
1726 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1727 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1728 (ia64_modes_tieable_p): New function.
1729 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1730 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1731 (iq2000_modes_tieable_p): New function.
1732 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1733 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1734 (lm32_modes_tieable_p): New function.
1735 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1736 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1737 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1738 (TARGET_MODES_TIEABLE_P): Redefine.
1739 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1740 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1741 (m32r_modes_tieable_p): New function.
1742 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1743 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1744 (m68k_modes_tieable_p): New function.
1745 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1746 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1747 (mcore_modes_tieable_p): New function.
1748 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1749 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1751 (TARGET_MODES_TIEABLE_P): Redefine.
1752 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1753 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1754 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1755 (TARGET_MODES_TIEABLE_P): Redefine.
1756 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1757 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1758 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1759 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1760 (mn10300_modes_tieable_p): ...this and make static.
1761 (TARGET_MODES_TIEABLE_P): Redefine.
1762 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1763 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1764 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1765 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1766 (msp430_modes_tieable_p): Make static.
1767 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1768 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1769 (TARGET_MODES_TIEABLE_P): Redefine.
1770 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1771 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1772 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1773 (TARGET_MODES_TIEABLE_P): Redefine.
1774 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1775 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1776 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1777 (TARGET_MODES_TIEABLE_P): Redefine.
1778 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1779 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1780 (pdp11_modes_tieable_p): New function.
1781 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1782 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1783 (rs6000_modes_tieable_p): New function.
1784 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1785 * config/powerpcspe/powerpcspe.md: Update comment.
1786 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1787 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1788 (TARGET_MODES_TIEABLE_P): Redefine.
1789 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1790 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1791 (rl78_modes_tieable_p): New function.
1792 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1793 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1794 (rs6000_modes_tieable_p): New function.
1795 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1796 * config/rs6000/rs6000.md: Update comment.
1797 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1798 * config/rx/rx.c (rx_modes_tieable_p): New function.
1799 (TARGET_MODES_TIEABLE_P): Redefine.
1800 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1801 * config/s390/s390.c (s390_modes_tieable_p): New function.
1802 (TARGET_MODES_TIEABLE_P): Redefine.
1803 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1804 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1805 (sh_modes_tieable_p): New function.
1806 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1807 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1808 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1809 (sparc_modes_tieable_p): Make static.
1810 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1811 * config/spu/spu.c (spu_modes_tieable_p): New function.
1812 (TARGET_MODES_TIEABLE_P): Redefine.
1813 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1814 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1815 (TARGET_MODES_TIEABLE_P): Redefine.
1816 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1817 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1818 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1819 * config/v850/v850.c (v850_modes_tieable_p): New function.
1820 (TARGET_MODES_TIEABLE_P): Redefine.
1821 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1822 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1823 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1824 (visium_modes_tieable_p): New function.
1825 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1826 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1827 (xtensa_modes_tieable_p): New function.
1828 * system.h (MODES_TIEABLE_P): Poison.
1830 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1831 Alan Hayward <alan.hayward@arm.com>
1832 David Sherwood <david.sherwood@arm.com>
1834 * target.def (hard_regno_mode_ok): New hook.
1835 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1836 (TARGET_HARD_REGNO_MODE_OK): ...this.
1837 * doc/tm.texi.in: Regenerate.
1838 * hooks.h (hook_bool_uint_mode_true): Declare.
1839 * hooks.c (hook_bool_uint_mode_true): New function.
1840 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1842 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1843 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1844 instead of HARD_REGNO_MODE_OK.
1845 * caller-save.c: Include target.h.
1846 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1848 * combine.c (can_combine_p): Likewise.
1849 (combinable_i3pat): Likewise.
1850 (can_change_dest_mode): Likewise.
1851 * expr.c (init_expr_target): Likewise.
1852 (convert_move): Likewise.
1853 (convert_modes): Likewise.
1854 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1855 (setup_prohibited_mode_move_regs): Likewise.
1856 * ira.h (target_ira): Likewise.
1857 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1858 * lra-constraints.c (process_alt_operands): Likewise.
1859 (split_reg): Likewise.
1860 * recog.c (peep2_find_free_register): Likewise.
1861 * ree.c (combine_reaching_defs): Likewise.
1862 * regcprop.c (maybe_mode_change): Likewise.
1863 * reginfo.c (init_reg_sets_1): Likewise.
1864 (choose_hard_reg_mode): Likewise.
1865 (simplifiable_subregs): Likewise.
1866 * regrename.c (check_new_reg_p): Likewise.
1867 * reload.c (find_valid_class): Likewise.
1868 (find_valid_class_1): Likewise.
1869 (reload_inner_reg_of_subreg): Likewise.
1870 (push_reload): Likewise.
1871 (combine_reloads): Likewise.
1872 (find_dummy_reload): Likewise.
1873 (find_reloads): Likewise.
1874 * reload1.c (find_reg): Likewise.
1875 (set_reload_reg): Likewise.
1876 (allocate_reload_reg): Likewise.
1877 (choose_reload_regs): Likewise.
1878 (reload_adjust_reg_for_temp): Likewise.
1879 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1880 (simplify_subreg_regno): Likewise.
1881 * sel-sched.c (init_regs_for_mode): Likewise.
1882 * varasm.c (make_decl_rtl): Likewise.
1883 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1884 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1886 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1887 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1889 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1890 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1892 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1893 (arc_mode_class): Delete.
1894 (HARD_REGNO_MODE_OK): Delete.
1895 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1896 (arc_hard_regno_mode_ok): Rename old array to...
1897 (arc_hard_regno_mode_ok_modes): ...this.
1898 (arc_conditional_register_usage): Update accordingly.
1899 (arc_mode_class): Make static.
1900 (arc_hard_regno_mode_ok): New function.
1901 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1902 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1903 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1904 (arm_hard_regno_mode_ok): Make static.
1905 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1907 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1908 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1909 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1911 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1912 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1913 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1914 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1915 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1916 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1917 * config/bfin/predicates.md (valid_reg_operand): Use
1918 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1919 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1920 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1921 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1922 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1923 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1924 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1925 (cr16_hard_regno_mode_ok): Make static and return a bool.
1926 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1927 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1928 (cris_hard_regno_mode_ok): New function.
1929 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1930 (epiphany_mode_class): Delete.
1931 (HARD_REGNO_MODE_OK): Delete.
1932 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1933 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1934 (hard_regno_mode_ok): Rename to...
1935 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1936 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1937 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1939 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1940 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1941 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1942 (frv_hard_regno_mode_ok): Make static and return a bool.
1943 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1945 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1946 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1947 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1948 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1951 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1952 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1953 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1956 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1957 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1958 (ia64_hard_regno_mode_ok): New function.
1959 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1960 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1961 (iq2000_hard_regno_mode_ok): New function.
1962 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1963 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1964 (lm32_hard_regno_mode_ok): New function.
1965 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1966 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1967 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1968 instead of HARD_REGNO_MODE_OK.
1969 (m32c_hard_regno_ok): Rename to...
1970 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1971 (m32c_cannot_change_mode_class): Update accordingly.
1972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1973 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1974 (m32r_mode_class): Delete.
1975 (HARD_REGNO_MODE_OK): Delete.
1976 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1977 (m32r_hard_regno_mode_ok): Rename to...
1978 (m32r_hard_regno_modes): ...this.
1979 (m32r_mode_class): Make static.
1980 (m32r_hard_regno_mode_ok): New function.
1981 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1982 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1983 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1984 (m68k_hard_regno_mode_ok): Make static.
1985 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1986 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1987 (mcore_hard_regno_mode_ok): New function.
1988 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1989 (HARD_REGNO_MODE_OK): Delete.
1990 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1992 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1993 (microblaze_hard_regno_mode_ok): New function.
1994 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1995 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1996 (mips_hard_regno_mode_ok): Delete.
1997 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1998 (mips_hard_regno_mode_ok_p): ...this and make static.
1999 (mips_hard_regno_mode_ok_p): Rename to...
2000 (mips_hard_regno_mode_ok_uncached): ...this.
2001 (mips_hard_regno_mode_ok): New function.
2002 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2003 of HARD_REGNO_MODE_OK.
2004 (mips_option_override): Update after above name changes.
2005 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2006 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2007 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2008 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2009 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2010 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2011 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2012 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2013 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2014 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2015 (msp430_hard_regno_mode_ok): Make static and return a bool.
2016 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2017 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2018 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2020 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2021 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2022 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2023 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2024 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2025 (PA_HARD_REGNO_MODE_OK): ...this
2026 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2027 (PA_HARD_REGNO_MODE_OK): ...this.
2028 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2029 (pa_hard_regno_mode_ok): New function.
2030 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2031 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2032 (pdp11_hard_regno_mode_ok): New function.
2033 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2034 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2036 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2038 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2039 (rs6000_hard_regno_mode_ok): Rename to...
2040 (rs6000_hard_regno_mode_ok_uncached): ...this.
2041 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2042 (rs6000_hard_regno_mode_ok): New function.
2043 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2044 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2045 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2046 (riscv_hard_regno_mode_ok): ...this and make static.
2047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2048 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2049 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2050 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2051 (rl78_hard_regno_mode_ok): Make static and return bool.
2052 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2053 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2055 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2056 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2057 (rs6000_hard_regno_mode_ok): Rename to...
2058 (rs6000_hard_regno_mode_ok_uncached): ...this.
2059 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2060 (rs6000_hard_regno_mode_ok): New function.
2061 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2062 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2063 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2064 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2065 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2066 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2067 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2068 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2069 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2070 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2071 (sh_hard_regno_mode_ok): Make static.
2072 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2073 instead of HARD_REGNO_MODE_OK.
2074 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2075 (sparc_mode_class): Delete.
2076 (HARD_REGNO_MODE_OK): Delete.
2077 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2078 (hard_regno_mode_classes): Make static.
2079 (sparc_mode_class): Likewise.
2080 (sparc_hard_regno_mode_ok): New function.
2081 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2082 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2083 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2086 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2087 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2088 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2089 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2091 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2092 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2093 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2094 (visium_hard_regno_mode_ok): New function.
2095 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2096 instead of HARD_REGNO_MODE_OK.
2097 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2098 (HARD_REGNO_MODE_OK): Delete.
2099 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2100 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2101 (xtensa_option_override): Update accordingly.
2102 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2103 (xtensa_hard_regno_mode_ok): New function.
2104 * system.h (HARD_REGNO_MODE_OK): Poison.
2106 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2107 Alan Hayward <alan.hayward@arm.com>
2108 David Sherwood <david.sherwood@arm.com>
2110 * target.def (hard_regno_call_part_clobbered): New hook.
2111 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2112 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2113 * doc/tm.texi: Regenerate.
2114 * hooks.h (hook_bool_uint_mode_false): Declare.
2115 * hooks.c (hook_bool_uint_mode_false): New function.
2116 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2117 * cselib.c (cselib_process_insn): Use
2118 targetm.hard_regno_call_part_clobbered instead of
2119 HARD_REGNO_CALL_PART_CLOBBERED.
2120 * ira-conflicts.c (ira_build_conflicts): Likewise.
2121 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2122 * lra-constraints.c (need_for_call_save_p): Likewise.
2123 * lra-lives.c: Include target.h.
2124 (check_pseudos_live_through_calls): Use
2125 targetm.hard_regno_call_part_clobbered instead of
2126 HARD_REGNO_CALL_PART_CLOBBERED.
2127 * regcprop.c: Include target.h.
2128 (copyprop_hardreg_forward_1): Use
2129 targetm.hard_regno_call_part_clobbered instead of
2130 HARD_REGNO_CALL_PART_CLOBBERED.
2131 * reginfo.c (choose_hard_reg_mode): Likewise.
2132 * regrename.c (check_new_reg_p): Likewise.
2133 * reload.c (find_equiv_reg): Likewise.
2134 * reload1.c (emit_reload_insns): Likewise.
2135 * sched-deps.c (deps_analyze_insn): Likewise.
2136 * sel-sched.c (init_regs_for_mode): Likewise.
2137 (mark_unavailable_hard_regs): Likewise.
2138 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2139 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2140 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2142 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2143 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2144 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2146 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2148 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2149 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2150 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2152 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2153 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2154 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2156 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2157 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2159 * config/powerpcspe/powerpcspe.c
2160 (rs6000_hard_regno_call_part_clobbered): New function.
2161 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2162 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2163 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2165 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2166 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2167 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2169 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2170 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2171 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2173 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2174 Alan Hayward <alan.hayward@arm.com>
2175 David Sherwood <david.sherwood@arm.com>
2177 * rtl.h (subreg_memory_offset): Declare.
2178 * emit-rtl.c (subreg_memory_offset): New function.
2179 * expmed.c (store_bit_field_1): Use it.
2180 * expr.c (undefined_operand_subword_p): Likewise.
2181 * simplify-rtx.c (simplify_subreg): Likewise.
2183 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2185 PR rtl-optimization/57448
2188 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2189 using atomic_load pattern.
2190 (expand_atomic_store): Likewise.
2192 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2195 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2196 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2197 handling. Use replace_call_with_value with NULL instead of
2198 gsi_replace, unlink_stmt_vdef and release_defs.
2200 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2203 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2205 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2208 * rtl.h (emit_library_call_value_1): Declare.
2209 (emit_library_call): Replace declaration with a series of overloads.
2210 Remove the parameter count argument.
2211 (emit_library_call_value): Likewise.
2212 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2213 with an "rtx_mode_t *".
2214 (emit_library_call_value): Delete.
2215 (emit_library_call): Likewise.
2216 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2217 (asan_emit_allocas_unpoison): Likewise.
2218 * builtins.c (expand_builtin_powi): Likewise.
2219 (expand_asan_emit_allocas_unpoison): Likewise.
2220 * cfgexpand.c (expand_main_function): Likewise.
2221 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2222 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2223 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2224 * config/arm/arm.c (arm_trampoline_init): Likewise.
2225 (arm_call_tls_get_addr): Likewise.
2226 (arm_expand_divmod_libfunc): Likewise.
2227 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2228 (smulsi3_highpart): Likewise.
2229 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2230 (c6x_expand_compare): Likewise.
2231 (c6x_expand_movmem): Likewise.
2232 * config/frv/frv.c (frv_trampoline_init): Likewise.
2233 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2234 (ix86_expand_divmod_libfunc): Likewise.
2235 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2236 (ia64_expand_compare): Likewise.
2237 (ia64_profile_hook): Likewise.
2238 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2239 (nonlocal_goto): Likewise.
2240 (restore_stack_nonlocal): Likewise.
2241 * config/m32r/m32r.c (block_move_call): Likewise.
2242 (m32r_trampoline_init): Likewise.
2243 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2244 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2245 (m68k_call_m68k_read_tp): Likewise.
2246 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2247 (microblaze_expand_divide): Likewise.
2248 * config/mips/mips.h (mips_args): Likewise.
2249 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2250 (MIPS_ICACHE_SYNC): Likewise.
2251 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2252 (nios2_trampoline_init): Likewise.
2253 * config/pa/pa.c (hppa_tls_call): Likewise.
2254 (pa_trampoline_init): Likewise.
2255 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2256 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2257 (expand_strn_compare): Likewise.
2258 (rs6000_generate_compare): Likewise.
2259 (rs6000_expand_float128_convert): Likewise.
2260 (output_profile_hook): Likewise.
2261 (rs6000_trampoline_init): Likewise.
2262 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2263 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2264 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2265 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2266 (rs6000_generate_compare): Likewise.
2267 (rs6000_expand_float128_convert): Likewise.
2268 (output_profile_hook): Likewise.
2269 (rs6000_trampoline_init): Likewise.
2270 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2271 * config/sh/sh.c (sh_trampoline_init): Likewise.
2272 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2273 (sparc_emit_float_lib_cmp): Likewise.
2274 (sparc32_initialize_trampoline): Likewise.
2275 (sparc64_initialize_trampoline): Likewise.
2276 (sparc_profile_hook): Likewise.
2277 * config/spu/spu.c (ea_load_store): Likewise.
2278 * config/spu/spu.md (floatunssidf2): Likewise.
2279 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2280 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2281 * config/visium/visium.c (expand_block_move_4): Likewise.
2282 (expand_block_move_2): Likewise.
2283 (expand_block_move_1): Likewise.
2284 (expand_block_set_4): Likewise.
2285 (expand_block_set_2): Likewise.
2286 (expand_block_set_1): Likewise.
2287 (visium_trampoline_init): Likewise.
2288 (visium_profile_hook): Likewise.
2289 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2290 (xtensa_setup_frame_addresses): Likewise.
2291 (xtensa_trampoline_init): Likewise.
2292 * except.c (sjlj_emit_function_enter): Likewise.
2293 (sjlj_emit_function_exit): Likewise.
2294 * explow.c (allocate_dynamic_stack_space): Likewise.
2295 (probe_stack_range): Likewise.
2296 * expr.c (convert_mode_scalar): Likewise.
2297 * optabs.c (expand_binop): Likewise.
2298 (expand_twoval_binop_libfunc): Likewise.
2299 (expand_unop): Likewise.
2300 (prepare_cmp_insn): Likewise.
2301 (prepare_float_lib_cmp): Likewise.
2302 (expand_float): Likewise.
2303 (expand_fix): Likewise.
2304 (expand_fixed_convert): Likewise.
2305 (maybe_emit_sync_lock_test_and_set): Likewise.
2306 (expand_atomic_compare_and_swap): Likewise.
2307 (expand_mem_thread_fence): Likewise.
2308 (expand_atomic_fetch_op): Likewise.
2310 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2312 * doc/generic.texi (OpenACC): Adjust URL.
2313 * doc/invoke.texi (C Dialect Options): Ditto.
2315 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2317 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2318 predicate for operand 1. Add (m,<S>) constraint.
2319 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2320 Prevent memory operand 1 with register operand 2.
2322 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2324 PR rtl-optimization/82024
2325 * combine.c (try_combine): If the combination result is a PARALLEL,
2326 and we only need to retain the SET in there that would be placed
2327 at I2, check that we can place that at I3 instead, before doing so.
2329 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2332 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2334 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2335 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2338 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2339 Jeff Law <law@redhat.com>
2341 * varasm.c (bss_initializer_p): Do not put constants into .bss
2342 (categorize_decl_for_section): Handle bss_initializer_p returning
2343 false when DECL_INITIAL is NULL.
2345 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2348 * config/s390/s390.c (s390_can_inline_p): New function.
2350 2017-09-01 Jeff Law <law@redhat.com>
2352 PR tree-optimization/82052
2353 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2354 Always initialize the returned slot after a hash table miss
2355 when INSERT is true.
2357 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2359 * config/s390/s390.md (mem_signal_fence): Remove.
2360 * doc/md.texi (mem_signal_fence): Remove.
2361 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2363 * target-insns.def (mem_signal_fence): Remove.
2365 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2368 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2371 * asan.c (create_odr_indicator): Strip name encoding from assembler
2372 name before appending it after __odr_asan_.
2374 2017-09-01 Martin Liska <mliska@suse.cz>
2376 PR tree-optimization/82059
2377 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2378 frequency only when an edge is redirected.
2380 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2382 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2383 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2384 (arc_conditional_register_usage): Remove ARC600 lp_count
2386 (arc_file_start): Emit Tag_ARC_CPU_variation.
2387 (arc_can_use_doloop_p): New conditions to use ZOLs.
2388 (hwloop_fail): New function.
2389 (hwloop_optimize): Likewise.
2390 (hwloop_pattern_reg): Likewise.
2391 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2392 (arc_reorg_loops): New function, calls reorg_loops.
2393 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2394 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2396 (arc_loop_hazard): Remove function, functionality moved into
2398 (arc_hazard): Remove arc_loop_hazard call.
2399 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2400 into hwloop_optimize.
2401 (arc_label_align): Remove ZOL handling.
2402 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2403 * config/arc/arc.md (doloop_begin): Remove pattern.
2404 (doloop_begin_i): Likewise.
2405 (doloop_end_i): Likewise.
2406 (doloop_fallback): Likewise.
2407 (doloop_fallback_m): Likewise.
2408 (doloop_end): Reimplement expand.
2409 (arc_lp): New pattern for LP instruction.
2410 (loop_end): New pattern.
2411 (loop_fail): Likewise.
2412 (decrement_and_branch_until_zero): Likewise.
2413 * config/arc/arc.opt (mlpc-width): New option.
2414 * doc/invoke.texi (mlpc-width): Document option.
2416 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2418 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2419 (arc_ccfsm_advance): Fix checking for delay slots.
2420 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2422 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2424 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2426 (movhi_insn): Update store instruction constraint which are saving
2427 6-bit short immediates.
2428 (movsi_insn): Consider also short scaled load operations.
2429 (zero_extendhisi2_i): Use Usd constraint instead of T.
2430 (extendhisi2_i): Add q constraint.
2431 (arc_clzsi2): Add type and length attributes.
2432 (arc_ctzsi2): Likewise.
2433 * config/arc/constraints.md (Usc): Update constraint, the
2434 assembler can parse two relocations for a single instruction.
2436 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2438 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2439 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2441 2017-08-31 Olivier Hainque <hainque@adacore.com>
2443 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2444 match as powerpc-wrs-vxworks*.
2446 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2448 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2449 register constraint for by-element operand.
2450 (aarch64_mls_elt_merge<mode>): Likewise.
2452 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2454 * config/arc/arc.c (arc_can_follow_jump): Check for short
2457 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2459 * config.gcc: Use g.opt for arc.
2460 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2461 functionality moved to ...
2462 (legitimate_scaled_address_p): New function, ...here.
2463 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2464 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2465 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2467 (arc_override_options): Handle G option.
2468 (arc_output_pic_addr_const): Correct function definition.
2469 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2470 (arc_decl_anon_ns_mem_p): Delete.
2471 (arc_in_small_data_p): Overhaul this function to take into
2472 consideration the value given via G option.
2473 (arc_rewrite_small_data_1): Renamed and corrected old
2474 arc_rewrite_small_data function.
2475 (arc_rewrite_small_data): New function.
2476 (small_data_pattern): Don't use pic_offset_table_rtx.
2477 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2478 * config/arc/simdext.md (movmisalignv2hi): Use
2479 prepare_move_operands function.
2481 (movmisalign*): Likewise.
2482 * doc/invoke.texi (ARC options): Document -G option.
2484 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2486 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2488 * config/arc/arc.c (arc_print_operand): Output scalled address for
2489 sdata whenever is possible.
2490 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2491 load/stores are available.
2492 (compact_sda_memory_operand): Check for the alignment required by
2493 code density instructions.
2494 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2496 * config/arc/constraints.md (Usd): Update constraint.
2497 (Us0): New constraint.
2498 (Usc): Update constraint.
2500 2017-08-31 Richard Biener <rguenther@suse.de>
2503 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2506 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2508 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2509 Resize type_signature.
2511 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2512 Alan Hayward <alan.hayward@arm.com>
2513 David Sherwood <david.sherwood@arm.com>
2515 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2516 subregs whose inner modes can be stored in GPRs.
2517 (aarch64_classify_index): Likewise.
2519 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2520 Alan Hayward <alan.hayward@arm.com>
2521 David Sherwood <david.sherwood@arm.com>
2523 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2524 (V_INT_EQUIV): ...this.
2525 (v_cmp_result): Rename to...
2526 (v_int_equiv): ...this.
2527 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2528 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2529 (copysign<mode>3): Likewise.
2530 (aarch64_simd_bsl<mode>_internal): Likewise.
2531 (aarch64_simd_bsl<mode>): Likewise.
2532 (vec_cmp<mode><mode>): Likewise.
2533 (vcond<mode><mode>): Likewise.
2534 (vcond<v_cmp_mixed><mode>): Likewise.
2535 (vcondu<mode><v_cmp_mixed>): Likewise.
2536 (aarch64_cm<optab><mode>): Likewise.
2537 (aarch64_cmtst<mode>): Likewise.
2538 (aarch64_fac<optab><mode>): Likewise.
2539 (vec_perm_const<mode>): Likewise.
2540 (vcond_mask_<mode><v_cmp_result>): Rename to...
2541 (vcond_mask_<mode><v_int_equiv>): ...this.
2542 (vec_cmp<mode><v_cmp_result>): Rename to...
2543 (vec_cmp<mode><v_int_equiv>): ...this.
2545 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2546 Alan Hayward <alan.hayward@arm.com>
2547 David Sherwood <david.sherwood@arm.com>
2549 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2551 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2552 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2553 (UNSPEC_LD4_DREG): New unspecs.
2554 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2555 (aarch64_ld2<mode>_dreg_be): Replace with...
2556 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2558 (aarch64_ld3<mode>_dreg_le)
2559 (aarch64_ld3<mode>_dreg_be): Replace with...
2560 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2562 (aarch64_ld4<mode>_dreg_le)
2563 (aarch64_ld4<mode>_dreg_be): Replace with...
2564 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2567 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2569 PR tree-optimization/81987
2570 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2571 insert an initializer in a location not dominated by the stride
2574 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2576 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2577 on the entire header of the finally block in the fallthru case.
2579 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2581 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2583 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2585 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2586 rs6000_emit_move_from_cr and call renamed function.
2587 (rs6000_emit_prologue): Call renamed functions.
2588 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2589 movesi_from_cr, remove volatile CRs.
2591 2017-08-30 Jon Beniston <jon@beniston.com>
2592 Richard Biener <rguenther@suse.de>
2594 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2595 of VECTOR_MODE_P check.
2596 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2597 element vector types.
2599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2601 * df.h (df_read_modify_subreg_p): Remove in favor of...
2602 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2603 const_rtx instead of an rtx.
2604 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2605 * df-problems.c (df_word_lr_mark_ref): Likewise.
2606 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2607 (mark_pseudo_reg_dead): Likewise.
2608 (mark_ref_dead): Likewise.
2609 * reginfo.c (init_subregs_of_mode): Likewise.
2610 * sched-deps.c (sched_analyze_1): Likewise.
2611 * df-scan.c (df_def_record_1): Likewise.
2612 (df_uses_record): Likewise.
2613 (df_read_modify_subreg_p): Remove in favor of...
2614 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2615 const_rtx instead of an rtx.
2617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2618 Alan Hayward <alan.hayward@arm.com>
2619 David Sherwood <david.sherwood@arm.com>
2621 * rtl.h (partial_subreg_p): New function.
2622 * caller-save.c (save_call_clobbered_regs): Use it.
2623 * calls.c (expand_call): Likewise.
2624 * combine.c (combinable_i3pat): Likewise.
2625 (simplify_set): Likewise.
2626 (make_extraction): Likewise.
2627 (make_compound_operation_int): Likewise.
2628 (gen_lowpart_or_truncate): Likewise.
2629 (force_to_mode): Likewise.
2630 (make_field_assignment): Likewise.
2631 (reg_truncated_to_mode): Likewise.
2632 (record_truncated_value): Likewise.
2633 (move_deaths): Likewise.
2634 * cse.c (record_jump_cond): Likewise.
2635 (cse_insn): Likewise.
2636 * cselib.c (cselib_lookup_1): Likewise.
2637 * expmed.c (extract_bit_field_using_extv): Likewise.
2638 * function.c (assign_parm_setup_reg): Likewise.
2639 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2640 * ira-build.c (create_insn_allocnos): Likewise.
2641 * lra-coalesce.c (merge_pseudos): Likewise.
2642 * lra-constraints.c (match_reload): Likewise.
2643 (simplify_operand_subreg): Likewise.
2644 (curr_insn_transform): Likewise.
2645 * lra-lives.c (process_bb_lives): Likewise.
2646 * lra.c (new_insn_reg): Likewise.
2647 (lra_substitute_pseudo): Likewise.
2648 * regcprop.c (mode_change_ok): Likewise.
2649 (maybe_mode_change): Likewise.
2650 (copyprop_hardreg_forward_1): Likewise.
2651 * reload.c (push_reload): Likewise.
2652 (find_reloads): Likewise.
2653 (find_reloads_subreg_address): Likewise.
2654 * reload1.c (alter_reg): Likewise.
2655 (eliminate_regs_1): Likewise.
2656 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2658 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2660 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2661 back to if statements, including unpack.
2663 2017-08-30 Martin Liska <mliska@suse.cz>
2666 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2668 (func_checker::compare_asm_inputs_outputs): ... this function.
2669 (func_checker::compare_gimple_asm): Use the function to compare
2670 also ASM constrains.
2671 * ipa-icf-gimple.h: Rename the function.
2673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2674 Alan Hayward <alan.hayward@arm.com>
2675 David Sherwood <david.sherwood@arm.com>
2677 * coretypes.h (complex_mode): New type.
2678 * gdbhooks.py (build_pretty_printer): Handle it.
2679 * machmode.h (complex_mode): New class.
2680 (complex_mode::includes_p): New function.
2681 (is_complex_int_mode): Likewise.
2682 (is_complex_float_mode): Likewise.
2683 * genmodes.c (get_mode_class): Handle complex mode classes.
2684 * function.c (expand_function_end): Use is_complex_int_mode.
2686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2687 Alan Hayward <alan.hayward@arm.com>
2688 David Sherwood <david.sherwood@arm.com>
2690 * coretypes.h (scalar_mode_pod): New typedef.
2691 * gdbhooks.py (build_pretty_printer): Handle it.
2692 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2693 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2694 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2695 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2699 Alan Hayward <alan.hayward@arm.com>
2700 David Sherwood <david.sherwood@arm.com>
2702 * machmode.h (mode_for_vector): Take a scalar_mode instead
2704 * stor-layout.c (mode_for_vector): Likewise.
2705 * explow.c (promote_mode): Use as_a <scalar_mode>.
2706 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2709 Alan Hayward <alan.hayward@arm.com>
2710 David Sherwood <david.sherwood@arm.com>
2712 * target.def (preferred_simd_mode): Take a scalar_mode
2713 instead of a machine_mode.
2714 * targhooks.h (default_preferred_simd_mode): Likewise.
2715 * targhooks.c (default_preferred_simd_mode): Likewise.
2716 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2717 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2718 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2719 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2720 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2721 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2722 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2723 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2725 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2726 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2727 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2728 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2729 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2730 * doc/tm.texi: Regenerate.
2731 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2735 Alan Hayward <alan.hayward@arm.com>
2736 David Sherwood <david.sherwood@arm.com>
2738 * target.def (scalar_mode_supported_p): Take a scalar_mode
2739 instead of a machine_mode.
2740 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2741 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2742 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2743 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2744 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2745 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2746 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2747 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2748 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2749 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2750 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2751 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2752 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2753 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2755 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2756 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2757 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2758 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2759 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2761 * doc/tm.texi: Regenerate.
2763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2764 Alan Hayward <alan.hayward@arm.com>
2765 David Sherwood <david.sherwood@arm.com>
2767 * coretypes.h (opt_scalar_mode): New typedef.
2768 * gdbhooks.py (build_pretty_printers): Handle it.
2769 * machmode.h (mode_iterator::get_2xwider): Add overload for
2771 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2773 * expr.c (convert_mode_scalar): Likewise.
2774 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2775 * optabs.c (expand_float): Likewise.
2776 (expand_fix): Likewise.
2777 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2780 Alan Hayward <alan.hayward@arm.com>
2781 David Sherwood <david.sherwood@arm.com>
2783 * optabs.c (expand_float): Explicitly check for scalars before
2784 using a branching expansion.
2785 (expand_fix): Likewise.
2787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2788 Alan Hayward <alan.hayward@arm.com>
2789 David Sherwood <david.sherwood@arm.com>
2791 * expr.c (convert_mode): Split scalar handling out into...
2792 (convert_mode_scalar): ...this new function. Treat the modes
2795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2796 Alan Hayward <alan.hayward@arm.com>
2797 David Sherwood <david.sherwood@arm.com>
2799 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2801 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2804 Alan Hayward <alan.hayward@arm.com>
2805 David Sherwood <david.sherwood@arm.com>
2807 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2808 rather than a machine_mode.
2809 (fixed_from_string): Likewise.
2810 (fixed_convert): Likewise.
2811 (fixed_convert_from_int): Likewise.
2812 (fixed_convert_from_real): Likewise.
2813 (real_convert_from_fixed): Likewise.
2814 * fixed-value.c (fixed_from_double_int): Likewise.
2815 (fixed_from_string): Likewise.
2816 (fixed_convert): Likewise.
2817 (fixed_convert_from_int): Likewise.
2818 (fixed_convert_from_real): Likewise.
2819 (real_convert_from_fixed): Likewise.
2820 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2823 Alan Hayward <alan.hayward@arm.com>
2824 David Sherwood <david.sherwood@arm.com>
2826 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2827 of separate mode class checks. Do not allow vector modes here.
2828 (immed_wide_int_const): Use as_a <scalar_mode>.
2829 * explow.c (trunc_int_for_mode): Likewise.
2830 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2831 (wi::shwi): Likewise.
2832 (wi::min_value): Likewise.
2833 (wi::max_value): Likewise.
2834 * dwarf2out.c (loc_descriptor): Likewise.
2835 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2839 Alan Hayward <alan.hayward@arm.com>
2840 David Sherwood <david.sherwood@arm.com>
2842 * tree.h (SCALAR_TYPE_MODE): New macro.
2843 * expr.c (expand_expr_addr_expr_1): Use it.
2844 (expand_expr_real_2): Likewise.
2845 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2846 (fold_convert_const_fixed_from_int): Likewise.
2847 (fold_convert_const_fixed_from_real): Likewise.
2848 (native_encode_fixed): Likewise
2849 (native_encode_complex): Likewise
2850 (native_encode_vector): Likewise.
2851 (native_interpret_fixed): Likewise.
2852 (native_interpret_real): Likewise.
2853 (native_interpret_complex): Likewise.
2854 (native_interpret_vector): Likewise.
2855 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2856 (simd_clone_adjust_argument_types): Likewise.
2857 (simd_clone_init_simd_arrays): Likewise.
2858 (simd_clone_adjust): Likewise.
2859 * stor-layout.c (layout_type): Likewise.
2860 * tree.c (build_minus_one_cst): Likewise.
2861 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2862 * tree-inline.c (estimate_move_cost): Likewise.
2863 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2864 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2865 (vectorizable_reduction): Likewise.
2866 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2867 (vect_recog_mixed_size_cond_pattern): Likewise.
2868 (check_bool_pattern): Likewise.
2869 (adjust_bool_pattern): Likewise.
2870 (search_type_for_mask_1): Likewise.
2871 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2872 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2873 (vectorizable_load): Likewise.
2874 (vectorizable_store): Likewise.
2875 * ubsan.c (ubsan_encode_value): Likewise.
2876 * varasm.c (output_constant): Likewise.
2878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2879 Alan Hayward <alan.hayward@arm.com>
2880 David Sherwood <david.sherwood@arm.com>
2882 * coretypes.h (scalar_mode): New class.
2883 * machmode.h (scalar_mode): Likewise.
2884 (scalar_mode::includes_p): New function.
2885 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2886 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2887 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2888 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2889 * expmed.c (store_bit_field_1): Likewise.
2890 (extract_bit_field_1): Likewise.
2891 * expr.c (write_complex_part): Likewise.
2892 (read_complex_part): Likewise.
2893 (emit_move_complex_push): Likewise.
2894 (expand_expr_real_2): Likewise.
2895 * function.c (assign_parm_setup_reg): Likewise.
2896 (assign_parms_unsplit_complex): Likewise.
2897 * optabs.c (expand_binop): Likewise.
2898 * rtlanal.c (subreg_get_info): Likewise.
2899 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2900 * varasm.c (output_constant_pool_2): Likewise.
2902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2903 Alan Hayward <alan.hayward@arm.com>
2904 David Sherwood <david.sherwood@arm.com>
2906 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2908 (expmed_mult_highpart_optab): Likewise.
2909 (expmed_mult_highpart): Likewise.
2911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2912 Alan Hayward <alan.hayward@arm.com>
2913 David Sherwood <david.sherwood@arm.com>
2915 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2916 instead of a machine_mode.
2917 (builtin_memset_read_str): Likewise.
2918 * builtins.c (c_readstr): Likewise.
2919 (builtin_memcpy_read_str): Likewise.
2920 (builtin_strncpy_read_str): Likewise.
2921 (builtin_memset_read_str): Likewise.
2922 (builtin_memset_gen_str): Likewise.
2923 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2924 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2925 instead of a machine_mode.
2926 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2928 (make_extraction): Likewise.
2929 (try_widen_shift_mode): Take and return scalar_int_modes instead
2931 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2932 a scalar_int_mode instead of a machine_mode.
2933 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2934 (avr_addr_space_pointer_mode): Likewise.
2935 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2936 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2937 (msp430_unwind_word_mode): Likewise.
2938 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2939 (spu_addr_space_pointer_mode): Likewise.
2940 (spu_addr_space_address_mode): Likewise.
2941 (spu_libgcc_cmp_return_mode): Likewise.
2942 (spu_libgcc_shift_count_mode): Likewise.
2943 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2944 (rl78_addr_space_pointer_mode): Likewise.
2945 (fl78_unwind_word_mode): Likewise.
2946 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2948 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2949 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2950 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2951 (mips_valid_pointer_mode): Likewise.
2952 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2953 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2954 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2956 (ft32_addr_space_address_mode): Likewise.
2957 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2958 scalar_int_mode instead of a machine_mode.
2959 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2961 (m32c_addr_space_address_mode): Likewise.
2962 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2963 (rs6000_eh_return_filter_mode): Likewise.
2964 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2965 (rs6000_eh_return_filter_mode): Likewise.
2966 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2967 (s390_libgcc_shift_count_mode): Likewise.
2968 (s390_unwind_word_mode): Likewise.
2969 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2971 * target.def (mode_rep_extended): Likewise.
2972 (valid_pointer_mode): Likewise.
2973 (addr_space.valid_pointer_mode): Likewise.
2974 (eh_return_filter_mode): Return a scalar_int_mode rather than
2976 (libgcc_cmp_return_mode): Likewise.
2977 (libgcc_shift_count_mode): Likewise.
2978 (unwind_word_mode): Likewise.
2979 (addr_space.pointer_mode): Likewise.
2980 (addr_space.address_mode): Likewise.
2981 * doc/tm.texi: Regenerate.
2982 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2984 (do_jump): Use scalar_int_mode for local variables.
2985 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2986 rather than a machine_mode.
2987 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2988 (scompare_loc_descriptor_wide): Likewise.
2989 (scompare_loc_descriptor_narrow): Likewise.
2990 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2992 * except.c (sjlj_emit_dispatch_table): Likewise.
2993 (expand_builtin_eh_copy_values): Likewise.
2994 * explow.c (convert_memory_address_addr_space_1): Likewise.
2995 Take a scalar_int_mode rather than a machine_mode.
2996 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2997 than a machine_mode.
2998 (memory_address_addr_space): Use scalar_int_mode for local variables.
2999 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3000 rather than a machine_mode.
3001 * expmed.c (mask_rtx): Likewise.
3002 (init_expmed_one_conv): Likewise.
3003 (expand_mult_highpart_adjust): Likewise.
3004 (extract_high_half): Likewise.
3005 (expmed_mult_highpart_optab): Likewise.
3006 (expmed_mult_highpart): Likewise.
3007 (expand_smod_pow2): Likewise.
3008 (expand_sdiv_pow2): Likewise.
3009 (emit_store_flag_int): Likewise.
3010 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3012 (extract_low_bits): Likewise.
3013 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3015 * expr.c (pieces_addr::adjust): Likewise.
3016 (can_store_by_pieces): Likewise.
3017 (store_by_pieces): Likewise.
3018 (clear_by_pieces_1): Likewise.
3019 (expand_expr_addr_expr_1): Likewise.
3020 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3021 (expand_expr_real_1): Likewise.
3022 (try_casesi): Likewise.
3023 * final.c (shorten_branches): Likewise.
3024 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3025 type of "mode" to machine_mode.
3026 * internal-fn.c (expand_arith_overflow_result_store): Take a
3027 scalar_int_mode rather than a machine_mode.
3028 (expand_mul_overflow): Use scalar_int_mode for local variables.
3029 * loop-doloop.c (doloop_modify): Likewise.
3030 (doloop_optimize): Likewise.
3031 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3032 than a machine_mode.
3033 (expand_doubleword_shift_condmove): Likewise.
3034 (expand_doubleword_shift): Likewise.
3035 (expand_doubleword_clz): Likewise.
3036 (expand_doubleword_popcount): Likewise.
3037 (expand_doubleword_parity): Likewise.
3038 (expand_absneg_bit): Use scalar_int_mode for local variables.
3039 (prepare_float_lib_cmp): Likewise.
3040 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3041 rather than a machine_mode.
3042 (convert_memory_address_addr_space): Likewise.
3043 (get_mode_bounds): Likewise.
3044 (get_address_mode): Return a scalar_int_mode rather than a
3046 * rtlanal.c (get_address_mode): Likewise.
3047 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3048 than a machine_mode.
3049 * targhooks.c (default_mode_rep_extended): Likewise.
3050 (default_valid_pointer_mode): Likewise.
3051 (default_addr_space_valid_pointer_mode): Likewise.
3052 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3053 than a machine_mode.
3054 (default_libgcc_cmp_return_mode): Likewise.
3055 (default_libgcc_shift_count_mode): Likewise.
3056 (default_unwind_word_mode): Likewise.
3057 (default_addr_space_pointer_mode): Likewise.
3058 (default_addr_space_address_mode): Likewise.
3059 * targhooks.h (default_eh_return_filter_mode): Likewise.
3060 (default_libgcc_cmp_return_mode): Likewise.
3061 (default_libgcc_shift_count_mode): Likewise.
3062 (default_unwind_word_mode): Likewise.
3063 (default_addr_space_pointer_mode): Likewise.
3064 (default_addr_space_address_mode): Likewise.
3065 (default_mode_rep_extended): Take a scalar_int_mode rather than
3067 (default_valid_pointer_mode): Likewise.
3068 (default_addr_space_valid_pointer_mode): Likewise.
3069 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3071 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3072 rather than a machine_mode.
3073 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3074 for local variables.
3075 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3076 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3077 than a machine_mode.
3079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3080 Alan Hayward <alan.hayward@arm.com>
3081 David Sherwood <david.sherwood@arm.com>
3083 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3084 the mode argument to scalar_int_mode.
3085 (do_jump_by_parts_zero_rtx): Likewise.
3086 (do_jump_by_parts_equality_rtx): Likewise.
3087 (do_jump_by_parts_greater): Take a mode argument.
3088 (do_jump_by_parts_equality): Likewise.
3089 (do_jump_1): Update calls accordingly.
3091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3092 Alan Hayward <alan.hayward@arm.com>
3093 David Sherwood <david.sherwood@arm.com>
3095 * is-a.h (safe_dyn_cast): New function.
3096 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3097 (jump_table_for_label): Likewise.
3098 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3099 instead of an rtx_insn *.
3100 (shorten_branches): Use dyn_cast instead of LABEL_P and
3101 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3102 rtx_jump_table_data::get_data_mode.
3103 (final_scan_insn): Likewise.
3105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3106 Alan Hayward <alan.hayward@arm.com>
3107 David Sherwood <david.sherwood@arm.com>
3109 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3110 trying to combine a full-register integer set with a subreg
3113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3114 Alan Hayward <alan.hayward@arm.com>
3115 David Sherwood <david.sherwood@arm.com>
3117 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3118 that is always either address_mode or pointer_mode.
3120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3121 Alan Hayward <alan.hayward@arm.com>
3122 David Sherwood <david.sherwood@arm.com>
3124 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3125 when the two are known to be equal.
3127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * simplify-rtx.c (simplify_const_unary_operation): Use
3132 is_a <scalar_int_mode> instead of checking for a nonzero
3133 precision. Forcibly convert op_mode to a scalar_int_mode
3134 in that case. More clearly differentiate the operand and
3135 result modes and use the former when deciding what the value
3136 of a count-bits operation should be. Use is_int_mode instead
3137 of checking for a MODE_INT. Remove redundant check for whether
3138 this mode has a zero precision.
3140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3141 Alan Hayward <alan.hayward@arm.com>
3142 David Sherwood <david.sherwood@arm.com>
3144 * optabs.c (widen_leading): Change the type of the mode argument
3145 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3146 (widen_bswap): Likewise.
3147 (expand_parity): Likewise.
3148 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3149 (expand_ffs): Likewise.
3150 (epand_unop): Check for scalar integer modes before calling the
3153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3154 Alan Hayward <alan.hayward@arm.com>
3155 David Sherwood <david.sherwood@arm.com>
3157 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3159 (expand_expr_real_1): Update call accordingly.
3161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3162 Alan Hayward <alan.hayward@arm.com>
3163 David Sherwood <david.sherwood@arm.com>
3165 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3166 value_mode arguments. Use scalar_int_mode internally.
3167 (store_bit_field_1): Rename the new integer mode from imode
3168 to op0_mode and use it instead of GET_MODE (op0). Update calls
3169 to store_split_bit_field, store_bit_field_using_insv and
3170 store_fixed_bit_field.
3171 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3172 Use scalar_int_mode internally. Use a bit count rather than a mode
3173 when calculating the largest bit size for get_best_mode.
3174 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3175 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3176 Remove assertion that OP0 has a scalar integer mode.
3177 (store_split_bit_field): Add op0_mode and value_mode arguments.
3178 Update calls to extract_fixed_bit_field.
3179 (extract_bit_field_using_extv): Add an op0_mode argument.
3180 Use scalar_int_mode internally.
3181 (extract_bit_field_1): Rename the new integer mode from imode to
3182 op0_mode and use it instead of GET_MODE (op0). Update calls to
3183 extract_split_bit_field, extract_bit_field_using_extv and
3184 extract_fixed_bit_field.
3185 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3186 to extract_split_bit_field and extract_fixed_bit_field_1.
3187 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3188 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3190 (extract_split_bit_field): Add an op0_mode argument. Update call
3191 to extract_fixed_bit_field.
3193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3194 Alan Hayward <alan.hayward@arm.com>
3195 David Sherwood <david.sherwood@arm.com>
3197 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3198 * explow.c (hard_function_value): Likewise.
3199 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3200 convert_to_mode call outside the loop.
3201 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3202 for the mode iterator. Require the mode specified by max_pieces
3204 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3206 (copy_blkmode_to_reg): Likewise.
3207 (set_storage_via_setmem): Likewise.
3208 * optabs.c (prepare_cmp_insn): Likewise.
3209 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3210 * stor-layout.c (finish_bitfield_representative): Likewise.
3212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3213 Alan Hayward <alan.hayward@arm.com>
3214 David Sherwood <david.sherwood@arm.com>
3216 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3217 * expr.c (convert_move): Use them.
3218 (convert_modes): Likewise.
3219 (store_expr_with_bounds): Likewise.
3221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3222 Alan Hayward <alan.hayward@arm.com>
3223 David Sherwood <david.sherwood@arm.com>
3225 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3226 parameter for the mode of "x". Remove the "known_x", "known_mode"
3227 and "known_ret" arguments. Change the type of the mode argument
3229 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3230 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3231 (reg_num_sign_bit_copies_for_combine): Likewise.
3232 * rtlanal.c (nonzero_bits1): Likewise.
3233 (num_sign_bit_copies1): Likewise.
3234 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3235 (reg_num_sign_bit_copies_general): Likewise.
3236 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3237 (reg_nonzero_bits_general): Likewise.
3239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3240 Alan Hayward <alan.hayward@arm.com>
3241 David Sherwood <david.sherwood@arm.com>
3243 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3244 than in subroutines. Return 1 for non-integer modes.
3245 (cached_num_sign_bit_copies): Change the type of the mode parameter
3247 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3248 classes. Handle CONST_INT_P first and then check whether X also
3249 has a scalar integer mode. Check the same thing for inner registers
3250 of a SUBREG and for values that are being extended or truncated.
3252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3253 Alan Hayward <alan.hayward@arm.com>
3254 David Sherwood <david.sherwood@arm.com>
3256 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3257 in subroutines. Return the mode mask for non-integer modes.
3258 (cached_nonzero_bits): Change the type of the mode parameter
3260 (nonzero_bits1): Likewise. Remove early exit for other mode
3261 classes. Handle CONST_INT_P first and then check whether X
3262 also has a scalar integer mode.
3264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3265 Alan Hayward <alan.hayward@arm.com>
3266 David Sherwood <david.sherwood@arm.com>
3268 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3269 Return a scalar_int_mode and assert that the size is greater than
3271 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3272 (op_by_pieces_d::op_by_pieces_d): Likewise.
3273 (op_by_pieces_d::run): Likewise.
3274 (can_store_by_pieces): Likewise.
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 (extract_left_shift): Add a mode argument and update
3282 (make_compound_operation_int): Change the type of the mode parameter
3283 to scalar_int_mode and update the call to extract_left_shift.
3285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3286 Alan Hayward <alan.hayward@arm.com>
3287 David Sherwood <david.sherwood@arm.com>
3289 * combine.c (simplify_and_const_int): Change the type of the mode
3290 parameter to scalar_int_mode.
3291 (simplify_and_const_int_1): Likewise. Update recursive call.
3293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3294 Alan Hayward <alan.hayward@arm.com>
3295 David Sherwood <david.sherwood@arm.com>
3297 * combine.c (simplify_compare_const): Check that the mode is a
3298 scalar_int_mode (rather than VOIDmode) before testing its
3300 (simplify_comparison): Move COMPARISON_P handling out of the
3301 loop and restrict the latter part of the loop to scalar_int_modes.
3302 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3303 and when considering SUBREG_REGs. Use is_int_mode instead of
3304 checking GET_MODE_CLASS against MODE_INT.
3306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3307 Alan Hayward <alan.hayward@arm.com>
3308 David Sherwood <david.sherwood@arm.com>
3310 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3311 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3312 shift_unit_mode and for modes involved in scalar shifts.
3314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3315 Alan Hayward <alan.hayward@arm.com>
3316 David Sherwood <david.sherwood@arm.com>
3318 * combine.c (force_int_to_mode): New function, split out from...
3319 (force_to_mode): ...here. Keep xmode up-to-date and use it
3320 instead of GET_MODE (x).
3322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3323 Alan Hayward <alan.hayward@arm.com>
3324 David Sherwood <david.sherwood@arm.com>
3326 * optabs-query.h (extraction_insn::struct_mode): Change type to
3327 opt_scalar_int_mode and update comment.
3328 (extraction_insn::field_mode): Change type to scalar_int_mode.
3329 (extraction_insn::pos_mode): Likewise.
3330 * combine.c (make_extraction): Update accordingly.
3331 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3332 (get_optab_extraction_insn): Likewise.
3333 * recog.c (simplify_while_replacing): Likewise.
3334 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3335 parameter to opt_scalar_int_mode.
3337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3338 Alan Hayward <alan.hayward@arm.com>
3339 David Sherwood <david.sherwood@arm.com>
3341 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3342 to a scalar_int_mode instead of a machine_mode.
3343 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3344 (get_best_mode): Return a boolean and use a pointer argument to store
3345 the selected mode. Replace the limit mode parameter with a bit limit.
3346 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3347 for the values returned by bit_field_mode_iterator::next_mode.
3348 (store_bit_field): Update call to get_best_mode.
3349 (store_fixed_bit_field): Likewise.
3350 (extract_fixed_bit_field): Likewise.
3351 * expr.c (optimize_bitfield_assignment_op): Likewise.
3352 * fold-const.c (optimize_bit_field_compare): Likewise.
3353 (fold_truth_andor_1): Likewise.
3354 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3355 Update for new type of m_mode.
3356 (get_best_mode): As above.
3358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3359 Alan Hayward <alan.hayward@arm.com>
3360 David Sherwood <david.sherwood@arm.com>
3362 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3363 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3364 (store_bit_field): Check is_a <scalar_int_mode> before calling
3365 strict_volatile_bitfield_p.
3366 (extract_bit_field): Likewise.
3368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3369 Alan Hayward <alan.hayward@arm.com>
3370 David Sherwood <david.sherwood@arm.com>
3372 * target.def (cstore_mode): Return a scalar_int_mode.
3373 * doc/tm.texi: Regenerate.
3374 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3375 * targhooks.h (default_cstore_mode): Likewise.
3376 * targhooks.c (default_cstore_mode): Likewise, using a forced
3378 * expmed.c (emit_cstore): Expect the target of the cstore to be
3381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3382 Alan Hayward <alan.hayward@arm.com>
3383 David Sherwood <david.sherwood@arm.com>
3385 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3387 (niter_desc): Likewise mode.
3388 (iv_analyze): Add a mode parameter.
3390 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3391 and change its type to scalar_int_mode.
3392 * loop-iv.c: Update commentary at head of file.
3393 (iv_constant): Pass the mode paraeter before the rtx it describes
3394 and change its type to scalar_int_mode. Remove VOIDmode handling.
3395 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3396 (iv_extend): Likewise.
3397 (shorten_into_mode): Likewise.
3398 (iv_add): Use scalar_int_mode.
3399 (iv_mult): Likewise.
3400 (iv_shift): Likewise.
3401 (canonicalize_iv_subregs): Likewise.
3402 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3403 it describes and change its mode to scalar_int_mode. Also change
3404 the type of the returned inner_mode to scalar_int_mode.
3405 (get_biv_step): Likewise, turning outer_mode from a pointer
3406 into a direct parameter. Update call to get_biv_step_1.
3407 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3408 iv_constant and get_biv_step.
3409 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3410 and change its type to scalar_int_mode. Don't initialise iv->mode
3411 to VOIDmode and remove later checks for its still being VOIDmode.
3412 Update calls to iv_analyze_op and iv_analyze_expr. Check
3413 is_a <scalar_int_mode> when changing the mode under consideration.
3414 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3415 Update call to iv_analyze_expr.
3416 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3417 inner register is not also a scalar_int_mode. Update call to
3419 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3420 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3421 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3422 separate mode class checks. Update calls to iv_analyze. Remove
3423 fix-up of VOIDmodes after iv_analyze_biv.
3424 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3425 don't have a scalar_int_mode. Update call to biv_p.
3427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3428 Alan Hayward <alan.hayward@arm.com>
3429 David Sherwood <david.sherwood@arm.com>
3431 * cfgexpand.c (convert_debug_memory_address): Use
3432 as_a <scalar_int_mode>.
3433 * combine.c (expand_compound_operation): Likewise.
3434 (make_extraction): Likewise.
3435 (change_zero_ext): Likewise.
3436 (simplify_comparison): Likewise.
3437 * cse.c (cse_insn): Likewise.
3438 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3439 (mem_loc_descriptor): Likewise.
3440 (loc_descriptor): Likewise.
3441 * expmed.c (init_expmed_one_mode): Likewise.
3442 (synth_mult): Likewise.
3443 (emit_store_flag_1): Likewise.
3444 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3445 of a comparison with size.
3446 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3447 (reduce_to_bit_field_precision): Likewise.
3448 * function.c (expand_function_end): Likewise.
3449 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3450 * loop-doloop.c (doloop_modify): Likewise.
3451 * optabs.c (expand_binop): Likewise.
3452 (expand_unop): Likewise.
3453 (expand_copysign_absneg): Likewise.
3454 (prepare_cmp_insn): Likewise.
3455 (maybe_legitimize_operand): Likewise.
3456 * recog.c (const_scalar_int_operand): Likewise.
3457 * rtlanal.c (get_address_mode): Likewise.
3458 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3459 (simplify_cond_clz_ctz): Likewise.
3460 * tree-nested.c (get_nl_goto_field): Likewise.
3461 * tree.c (build_vector_type_for_mode): Likewise.
3462 * var-tracking.c (use_narrower_mode): Likewise.
3464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3465 Alan Hayward <alan.hayward@arm.com>
3466 David Sherwood <david.sherwood@arm.com>
3468 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3469 * builtins.c (expand_builtin_signbit): Use it.
3470 * cfgexpand.c (expand_debug_expr): Likewise.
3471 * dojump.c (do_jump): Likewise.
3472 (do_compare_and_jump): Likewise.
3473 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3474 * expmed.c (make_tree): Likewise.
3475 * expr.c (expand_expr_real_2): Likewise.
3476 (expand_expr_real_1): Likewise.
3477 (try_casesi): Likewise.
3478 * fold-const-call.c (fold_const_call_ss): Likewise.
3479 * fold-const.c (unextend): Likewise.
3480 (extract_muldiv_1): Likewise.
3481 (fold_single_bit_test): Likewise.
3482 (native_encode_int): Likewise.
3483 (native_encode_string): Likewise.
3484 (native_interpret_int): Likewise.
3485 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3486 * internal-fn.c (expand_addsub_overflow): Likewise.
3487 (expand_neg_overflow): Likewise.
3488 (expand_mul_overflow): Likewise.
3489 (expand_arith_overflow): Likewise.
3490 * match.pd: Likewise.
3491 * stor-layout.c (layout_type): Likewise.
3492 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3493 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3494 * tree-ssanames.c (get_range_info): Likewise.
3495 * tree-switch-conversion.c (array_value_type) Likewise.
3496 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3497 (vect_recog_divmod_pattern): Likewise.
3498 (vect_recog_mixed_size_cond_pattern): Likewise.
3499 * tree-vrp.c (extract_range_basic): Likewise.
3500 (simplify_float_conversion_using_ranges): Likewise.
3501 * tree.c (int_fits_type_p): Likewise.
3502 * ubsan.c (instrument_bool_enum_load): Likewise.
3503 * varasm.c (mergeable_string_section): Likewise.
3504 (narrowing_initializer_constant_valid_p): Likewise.
3505 (output_constant): Likewise.
3507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3508 Alan Hayward <alan.hayward@arm.com>
3509 David Sherwood <david.sherwood@arm.com>
3511 * machmode.h (NARROWEST_INT_MODE): New macro.
3512 * expr.c (alignment_for_piecewise_move): Use it instead of
3513 GET_CLASS_NARROWEST_MODE (MODE_INT).
3514 (push_block): Likewise.
3515 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3517 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3520 Alan Hayward <alan.hayward@arm.com>
3521 David Sherwood <david.sherwood@arm.com>
3523 * postreload.c (move2add_valid_value_p): Change the type of the
3524 mode parameter to scalar_int_mode.
3525 (move2add_use_add2_insn): Add a mode parameter and use it instead
3527 (move2add_use_add3_insn): Likewise.
3528 (reload_cse_move2add): Update accordingly.
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 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3536 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3537 * optabs.c (expand_unop): Likewise.
3539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3540 Alan Hayward <alan.hayward@arm.com>
3541 David Sherwood <david.sherwood@arm.com>
3543 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3544 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3545 (popcount_loc_descriptor): Likewise.
3546 (bswap_loc_descriptor): Likewise.
3547 (rotate_loc_descriptor): Likewise.
3548 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3549 calling the functions above.
3551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3552 Alan Hayward <alan.hayward@arm.com>
3553 David Sherwood <david.sherwood@arm.com>
3555 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3557 (try_combine): Likewise.
3558 (simplify_if_then_else): Likewise.
3559 * cse.c (cse_insn): Likewise.
3560 * dwarf2out.c (mem_loc_descriptor): Likewise.
3561 * emit-rtl.c (gen_lowpart_common): Likewise.
3562 * simplify-rtx.c (simplify_truncation): Likewise.
3563 (simplify_binary_operation_1): Likewise.
3564 (simplify_const_relational_operation): Likewise.
3565 (simplify_ternary_operation): Likewise.
3566 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3569 Alan Hayward <alan.hayward@arm.com>
3570 David Sherwood <david.sherwood@arm.com>
3572 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3573 * reload.c (push_reload): Likewise.
3574 (find_reloads): Likewise.
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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3581 (make_compound_operation_int): Likewise.
3582 (change_zero_ext): Likewise.
3583 * expr.c (convert_move): Likewise.
3584 (convert_modes): Likewise.
3585 * fwprop.c (forward_propagate_subreg): Likewise.
3586 * loop-iv.c (get_biv_step_1): Likewise.
3587 * optabs.c (widen_operand): Likewise.
3588 * postreload.c (move2add_valid_value_p): Likewise.
3589 * recog.c (simplify_while_replacing): Likewise.
3590 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3591 (simplify_binary_operation_1): Likewise. Remove redundant
3592 mode equality check.
3594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3595 Alan Hayward <alan.hayward@arm.com>
3596 David Sherwood <david.sherwood@arm.com>
3598 * combine.c (combine_simplify_rtx): Add checks for
3599 is_a <scalar_int_mode>.
3600 (simplify_if_then_else): Likewise.
3601 (make_field_assignment): Likewise.
3602 (simplify_comparison): Likewise.
3603 * ifcvt.c (noce_try_bitop): Likewise.
3604 * loop-invariant.c (canonicalize_address_mult): Likewise.
3605 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3608 Alan Hayward <alan.hayward@arm.com>
3609 David Sherwood <david.sherwood@arm.com>
3611 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3612 is_a <scalar_int_mode> instead of != BLKmode.
3614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3615 Alan Hayward <alan.hayward@arm.com>
3616 David Sherwood <david.sherwood@arm.com>
3618 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3619 instead of != VOIDmode.
3620 * combine.c (if_then_else_cond): Likewise.
3621 (change_zero_ext): Likewise.
3622 * dwarf2out.c (mem_loc_descriptor): Likewise.
3623 (loc_descriptor): Likewise.
3624 * rtlanal.c (canonicalize_condition): Likewise.
3625 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3628 Alan Hayward <alan.hayward@arm.com>
3629 David Sherwood <david.sherwood@arm.com>
3631 * simplify-rtx.c (simplify_binary_operation_1): Use
3632 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3635 Alan Hayward <alan.hayward@arm.com>
3636 David Sherwood <david.sherwood@arm.com>
3638 * wide-int.h (int_traits<unsigned char>) New class.
3639 (int_traits<unsigned short>) Likewise.
3640 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3641 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3643 * combine.c (set_nonzero_bits_and_sign_copies): Use
3644 is_a <scalar_int_mode>.
3645 (find_split_point): Likewise.
3646 (combine_simplify_rtx): Likewise.
3647 (simplify_logical): Likewise.
3648 (expand_compound_operation): Likewise.
3649 (expand_field_assignment): Likewise.
3650 (make_compound_operation): Likewise.
3651 (extended_count): Likewise.
3652 (change_zero_ext): Likewise.
3653 (simplify_comparison): Likewise.
3654 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3655 (ucompare_loc_descriptor): Likewise.
3656 (minmax_loc_descriptor): Likewise.
3657 (mem_loc_descriptor): Likewise.
3658 (loc_descriptor): Likewise.
3659 * expmed.c (init_expmed_one_mode): Likewise.
3660 * lra-constraints.c (lra_constraint_offset): Likewise.
3661 * optabs.c (prepare_libcall_arg): Likewise.
3662 * postreload.c (move2add_note_store): Likewise.
3663 * reload.c (operands_match_p): Likewise.
3664 * rtl.h (load_extend_op): Likewise.
3665 * rtlhooks.c (gen_lowpart_general): Likewise.
3666 * simplify-rtx.c (simplify_truncation): Likewise.
3667 (simplify_unary_operation_1): Likewise.
3668 (simplify_binary_operation_1): Likewise.
3669 (simplify_const_binary_operation): Likewise.
3670 (simplify_const_relational_operation): Likewise.
3671 (simplify_subreg): Likewise.
3672 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3673 * var-tracking.c (adjust_mems): Likewise.
3674 (prepare_call_arguments): Likewise.
3676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3677 Alan Hayward <alan.hayward@arm.com>
3678 David Sherwood <david.sherwood@arm.com>
3680 * machmode.h (is_int_mode): New fuction.
3681 * combine.c (find_split_point): Use it.
3682 (combine_simplify_rtx): Likewise.
3683 (simplify_if_then_else): Likewise.
3684 (simplify_set): Likewise.
3685 (simplify_shift_const_1): Likewise.
3686 (simplify_comparison): Likewise.
3687 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3688 * cse.c (notreg_cost): Likewise.
3689 (cse_insn): Likewise.
3690 * cselib.c (cselib_lookup_1): Likewise.
3691 * dojump.c (do_jump_1): Likewise.
3692 (do_compare_rtx_and_jump): Likewise.
3693 * dse.c (get_call_args): Likewise.
3694 * dwarf2out.c (rtl_for_decl_init): Likewise.
3695 (native_encode_initializer): Likewise.
3696 * expmed.c (emit_store_flag_1): Likewise.
3697 (emit_store_flag): Likewise.
3698 * expr.c (convert_modes): Likewise.
3699 (store_field): Likewise.
3700 (expand_expr_real_1): Likewise.
3701 * fold-const.c (fold_read_from_constant_string): Likewise.
3702 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3703 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3704 * optabs.c (expand_binop): Likewise.
3705 (expand_unop): Likewise.
3706 (expand_abs_nojump): Likewise.
3707 (expand_one_cmpl_abs_nojump): Likewise.
3708 * simplify-rtx.c (mode_signbit_p): Likewise.
3709 (val_signbit_p): Likewise.
3710 (val_signbit_known_set_p): Likewise.
3711 (val_signbit_known_clear_p): Likewise.
3712 (simplify_relational_operation_1): Likewise.
3713 * tree.c (vector_type_mode): Likewise.
3715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3716 Alan Hayward <alan.hayward@arm.com>
3717 David Sherwood <david.sherwood@arm.com>
3719 * machmode.h (smallest_mode_for_size): Fix formatting.
3720 (smallest_int_mode_for_size): New function.
3721 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3722 instead of smallest_mode_for_size.
3723 * combine.c (make_extraction): Likewise.
3724 * config/arc/arc.c (arc_expand_movmem): Likewise.
3725 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3726 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3727 * config/s390/s390.c (s390_expand_insv): Likewise.
3728 * config/sparc/sparc.c (assign_int_registers): Likewise.
3729 * config/spu/spu.c (spu_function_value): Likewise.
3730 (spu_function_arg): Likewise.
3731 * coverage.c (get_gcov_type): Likewise.
3732 (get_gcov_unsigned_t): Likewise.
3733 * dse.c (find_shift_sequence): Likewise.
3734 * expmed.c (store_bit_field_1): Likewise.
3735 * expr.c (convert_move): Likewise.
3736 (store_field): Likewise.
3737 * internal-fn.c (expand_arith_overflow): Likewise.
3738 * optabs-query.c (get_best_extraction_insn): Likewise.
3739 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3740 * stor-layout.c (layout_type): Likewise.
3741 (initialize_sizetypes): Likewise.
3742 * targhooks.c (default_get_mask_mode): Likewise.
3743 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3746 Alan Hayward <alan.hayward@arm.com>
3747 David Sherwood <david.sherwood@arm.com>
3749 * machmode.h (opt_mode::else_blk): New function.
3750 (int_mode_for_mode): Declare.
3751 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3752 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3754 * cfgexpand.c (expand_debug_expr): Likewise.
3755 * combine.c (gen_lowpart_or_truncate): Likewise.
3756 (gen_lowpart_for_combine): Likewise.
3757 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3758 * config/avr/avr.c (avr_to_int_mode): Likewise.
3759 (avr_out_plus_1): Likewise.
3760 (avr_out_plus): Likewise.
3761 (avr_out_round): Likewise.
3762 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3763 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3764 (s390_expand_vcond): Likewise.
3765 * config/spu/spu.c (spu_split_immediate): Likewise.
3766 (spu_expand_mov): Likewise.
3767 * dse.c (get_stored_val): Likewise.
3768 * expmed.c (store_bit_field_1): Likewise.
3769 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3771 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3772 (extract_low_bits): Likewise.
3773 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3774 handling rather than repeating the check.
3775 (emit_group_store): Likewise.
3776 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3777 * optabs.c (expand_absneg_bit): Likewise.
3778 (expand_copysign_absneg): Likewise.
3779 (expand_copysign_bit): Likewise.
3780 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3781 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3782 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3783 * var-tracking.c (prepare_call_arguments): Likewise.
3784 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3785 int_mode_for_mode instead of mode_for_size.
3786 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3789 Alan Hayward <alan.hayward@arm.com>
3790 David Sherwood <david.sherwood@arm.com>
3792 * machmode.h (int_mode_for_size): New function.
3793 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3794 instead of mode_for_size.
3795 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3797 * combine.c (expand_field_assignment): Use int_mode_for_size
3798 instead of mode_for_size.
3799 (make_extraction): Likewise.
3800 (simplify_shift_const_1): Likewise.
3801 (simplify_comparison): Likewise.
3802 * dojump.c (do_jump): Likewise.
3803 * dwarf2out.c (mem_loc_descriptor): Likewise.
3804 * emit-rtl.c (init_derived_machine_modes): Likewise.
3805 * expmed.c (flip_storage_order): Likewise.
3806 (convert_extracted_bit_field): Likewise.
3807 * expr.c (copy_blkmode_from_reg): Likewise.
3808 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3809 * internal-fn.c (expand_mul_overflow): Likewise.
3810 * lower-subreg.c (simple_move): Likewise.
3811 * optabs-libfuncs.c (init_optabs): Likewise.
3812 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3813 * tree.c (vector_type_mode): Likewise.
3814 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3815 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3816 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3817 * tree-vect-stmts.c (vectorizable_load): Likewise.
3818 (vectorizable_store): Likewise.
3820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3821 Alan Hayward <alan.hayward@arm.com>
3822 David Sherwood <david.sherwood@arm.com>
3824 * coretypes.h (pod_mode): New type.
3825 (scalar_int_mode_pod): New typedef.
3826 * machmode.h (pod_mode): New class.
3827 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3828 * genmodes.c (emit_mode_int_n): Update accordingly.
3829 * lower-subreg.h (target_lower_subreg): Change type to
3830 scalar_int_mode_pod.
3831 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3832 scalar_int_mode_pod.
3834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3835 Alan Hayward <alan.hayward@arm.com>
3836 David Sherwood <david.sherwood@arm.com>
3838 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3839 machine_mode to scalar_int_mode.
3840 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3841 (rs6000_option_override_internal): Remove cast to int.
3842 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3843 machine_mode to scalar_int_mode.
3844 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3845 (rs6000_option_override_internal): Remove cast to int.
3846 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3847 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3849 * config/s390/s390.c (s390_expand_builtin): Likewise.
3850 * coretypes.h (scalar_int_mode): New type.
3851 (opt_scalar_int_mode): New typedef.
3852 * machmode.h (scalar_int_mode): New class.
3853 (scalar_int_mode::includes_p): New function.
3854 (byte_mode): Change type to scalar_int_mode.
3855 (word_mode): Likewise.
3856 (ptr_mode): Likewise.
3857 * emit-rtl.c (byte_mode): Likewise.
3858 (word_mode): Likewise.
3859 (ptr_mode): Likewise.
3860 (init_derived_machine_modes): Update accordingly.
3861 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3862 and MODE_PARTIAL_INT.
3863 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3864 opt_scalar_int_mode.
3866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3867 Alan Hayward <alan.hayward@arm.com>
3868 David Sherwood <david.sherwood@arm.com>
3870 * target.def (libgcc_floating_mode_supported_p): Take a
3872 * doc/tm.texi: Regenerate.
3873 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3875 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3876 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3880 Alan Hayward <alan.hayward@arm.com>
3881 David Sherwood <david.sherwood@arm.com>
3883 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3884 * doc/tm.texi: Regenerate.
3885 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3886 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3887 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3888 * targhooks.h (default_floatn_mode): Likewise.
3889 * targhooks.c (default_floatn_mode): Likewise.
3890 * tree.c (build_common_tree_nodes): Update accordingly.
3892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3893 Alan Hayward <alan.hayward@arm.com>
3894 David Sherwood <david.sherwood@arm.com>
3896 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3897 (mode_iterator::iterate_p): Likewise.
3898 (mode_iterator::get_wider): Likewise.
3899 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
3906 * machmode.h (float_mode_for_size): New function.
3907 * emit-rtl.c (double_mode): Delete.
3908 (init_emit_once): Use float_mode_for_size.
3909 * stor-layout.c (layout_type): Likewise.
3910 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3913 Alan Hayward <alan.hayward@arm.com>
3914 David Sherwood <david.sherwood@arm.com>
3916 * output.h (assemble_real): Take a scalar_float_mode.
3917 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3918 * config/arm/arm.md (consttable_4): Likewise.
3919 (consttable_8): Likewise.
3920 (consttable_16): Likewise.
3921 * config/mips/mips.md (consttable_float): Likewise.
3922 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3923 * varasm.c (assemble_real): Take a scalar_float_mode.
3924 (output_constant_pool_2): Update accordingly.
3925 (output_constant): Likewise.
3927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3928 Alan Hayward <alan.hayward@arm.com>
3929 David Sherwood <david.sherwood@arm.com>
3931 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3932 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3933 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3934 (native_encode_real): Likewise.
3935 (native_interpret_real): Likewise.
3936 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3937 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3940 Alan Hayward <alan.hayward@arm.com>
3941 David Sherwood <david.sherwood@arm.com>
3943 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3944 <scalar_float_mode>. Simplify.
3945 (gen_extend_conv_libfunc): Likewise.
3947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3948 Alan Hayward <alan.hayward@arm.com>
3949 David Sherwood <david.sherwood@arm.com>
3951 * coretypes.h (scalar_float_mode): New type.
3952 * machmode.h (mode_traits::from_int): Use machine_mode if
3953 USE_ENUM_MODES is defined.
3954 (is_a): New function.
3956 (dyn_cast): Likewise.
3957 (scalar_float_mode): New class.
3958 (scalar_float_mode::includes_p): New function.
3959 (is_float_mode): Likewise.
3960 * gdbhooks.py (MachineModePrinter): New class.
3961 (build_pretty_printer): Use it for scalar_float_mode.
3962 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3963 (format_helper::format_helper): Turn into a template.
3964 * genmodes.c (get_mode_class): New function.
3965 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3966 or machine_mode if none.
3967 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3968 as_a <scalar_float_mode>.
3969 * dwarf2out.c (mem_loc_descriptor): Likewise.
3970 (insert_float): Likewise.
3971 (add_const_value_attribute): Likewise.
3972 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3973 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3974 (expand_unop): Update accordingly.
3975 (expand_abs_nojump): Likewise.
3976 (expand_copysign_absneg): Take a scalar_float_mode.
3977 (expand_copysign_bit): Likewise.
3978 (expand_copysign): Update accordingly.
3980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3981 Alan Hayward <alan.hayward@arm.com>
3982 David Sherwood <david.sherwood@arm.com>
3984 * coretypes.h (opt_mode): New class.
3985 * machmode.h (opt_mode): Likewise.
3986 (opt_mode::else_void): New function.
3987 (opt_mode::require): Likewise.
3988 (opt_mode::exists): Likewise.
3989 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3990 (GET_MODE_2XWIDER_MODE): Likewise.
3991 (mode_iterator::get_wider): Update accordingly.
3992 (mode_iterator::get_2xwider): Likewise.
3993 (mode_iterator::get_known_wider): Likewise, turning into a template.
3994 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3995 forcing a wider mode to exist.
3996 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3997 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3998 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3999 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4000 * lower-subreg.c (init_lower_subreg): Likewise.
4001 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4002 on the final iteration.
4003 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4004 a wider mode exists before asking for a move pattern.
4005 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4006 forcing a wider mode to exist.
4007 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4008 returning false if no such mode exists.
4009 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4010 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4011 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4012 Avoid checking for a MODE_INT if we already know the mode is not a
4014 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4015 forcing a wider mode to exist.
4016 (expmed_mult_highpart_optab): Likewise.
4017 (expmed_mult_highpart): Likewise.
4018 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4020 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4021 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4022 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4023 * internal-fn.c (expand_mul_overflow): Update use of
4024 GET_MODE_2XWIDER_MODE.
4025 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4026 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4027 GET_MODE_WIDER_MODE.
4028 (convert_plusminus_to_widen): Likewise.
4029 * tree-switch-conversion.c (array_value_type): Likewise.
4030 * var-tracking.c (emit_note_insn_var_location): Likewise.
4031 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4032 Return false inside rather than outside the loop if no wider mode
4034 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4035 and GET_MODE_2XWIDER_MODE
4036 (can_compare_p): Use else_void.
4037 * gdbhooks.py (OptMachineModePrinter): New class.
4038 (build_pretty_printer): Use it for opt_mode.
4040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4041 Alan Hayward <alan.hayward@arm.com>
4042 David Sherwood <david.sherwood@arm.com>
4044 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4045 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4048 Alan Hayward <alan.hayward@arm.com>
4049 David Sherwood <david.sherwood@arm.com>
4051 * machmode.h (mode_traits): New structure.
4052 (get_narrowest_mode): New function.
4053 (mode_iterator::start): Likewise.
4054 (mode_iterator::iterate_p): Likewise.
4055 (mode_iterator::get_wider): Likewise.
4056 (mode_iterator::get_known_wider): Likewise.
4057 (mode_iterator::get_2xwider): Likewise.
4058 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4059 (FOR_EACH_MODE): Likewise.
4060 (FOR_EACH_MODE_FROM): Likewise.
4061 (FOR_EACH_MODE_UNTIL): Likewise.
4062 (FOR_EACH_WIDER_MODE): Likewise.
4063 (FOR_EACH_2XWIDER_MODE): Likewise.
4064 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4065 * combine.c (simplify_comparison): Likewise
4066 * config/i386/i386.c (type_natural_mode): Likewise.
4067 * cse.c (cse_insn): Likewise.
4068 * dse.c (find_shift_sequence): Likewise.
4069 * emit-rtl.c (init_derived_machine_modes): Likewise.
4070 (init_emit_once): Likewise.
4071 * explow.c (hard_function_value): Likewise.
4072 * expmed.c (extract_fixed_bit_field_1): Likewise.
4073 (extract_bit_field_1): Likewise.
4074 (expand_divmod): Likewise.
4075 (emit_store_flag_1): Likewise.
4076 * expr.c (init_expr_target): Likewise.
4077 (convert_move): Likewise.
4078 (alignment_for_piecewise_move): Likewise.
4079 (widest_int_mode_for_size): Likewise.
4080 (emit_block_move_via_movmem): Likewise.
4081 (copy_blkmode_to_reg): Likewise.
4082 (set_storage_via_setmem): Likewise.
4083 (compress_float_constant): Likewise.
4084 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4085 * optabs-query.c (get_best_extraction_insn): Likewise.
4086 * optabs.c (expand_binop): Likewise.
4087 (expand_twoval_unop): Likewise.
4088 (expand_twoval_binop): Likewise.
4089 (widen_leading): Likewise.
4090 (widen_bswap): Likewise.
4091 (expand_parity): Likewise.
4092 (expand_unop): Likewise.
4093 (prepare_cmp_insn): Likewise.
4094 (prepare_float_lib_cmp): Likewise.
4095 (expand_float): Likewise.
4096 (expand_fix): Likewise.
4097 (expand_sfix_optab): Likewise.
4098 * postreload.c (move2add_use_add2_insn): Likewise.
4099 * reg-stack.c (reg_to_stack): Likewise.
4100 * reginfo.c (choose_hard_reg_mode): Likewise.
4101 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4102 * stor-layout.c (mode_for_size): Likewise.
4103 (smallest_mode_for_size): Likewise.
4104 (mode_for_vector): Likewise.
4105 (finish_bitfield_representative): Likewise.
4106 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4107 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4108 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4109 * var-tracking.c (prepare_call_arguments): Likewise.
4111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4112 Alan Hayward <alan.hayward@arm.com>
4113 David Sherwood <david.sherwood@arm.com>
4115 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4116 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4117 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4118 * machmode.h (mode_size): Move earlier in file.
4119 (mode_precision): Likewise.
4120 (mode_inner): Likewise.
4121 (mode_nunits): Likewise.
4122 (mode_unit_size): Likewise.
4123 (unit_unit_precision): Likewise.
4124 (mode_wider): Likewise.
4125 (mode_2xwider): Likewise.
4126 (machine_mode): New class.
4127 (mode_to_bytes): New function.
4128 (mode_to_bits): Likewise.
4129 (mode_to_precision): Likewise.
4130 (mode_to_inner): Likewise.
4131 (mode_to_unit_size): Likewise.
4132 (mode_to_unit_precision): Likewise.
4133 (mode_to_nunits): Likewise.
4134 (GET_MODE_SIZE): Use mode_to_bytes.
4135 (GET_MODE_BITSIZE): Use mode_to_bits.
4136 (GET_MODE_PRECISION): Use mode_to_precision.
4137 (GET_MODE_INNER): Use mode_to_inner.
4138 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4139 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4140 (GET_MODE_NUNITS): Use mode_to_nunits.
4141 * system.h (ALWAYS_INLINE): New macro.
4142 * config/powerpcspe/powerpcspe-c.c
4143 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4144 int for arg1_mode and arg2_mode.
4146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4147 Alan Hayward <alan.hayward@arm.com>
4148 David Sherwood <david.sherwood@arm.com>
4150 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4151 Prefix mode names with E_ in case statements.
4152 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4153 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4154 (aarch64_split_simd_move): Likewise.
4155 (aarch64_gen_storewb_pair): Likewise.
4156 (aarch64_gen_loadwb_pair): Likewise.
4157 (aarch64_gen_store_pair): Likewise.
4158 (aarch64_gen_load_pair): Likewise.
4159 (aarch64_get_condition_code_1): Likewise.
4160 (aarch64_constant_pool_reload_icode): Likewise.
4161 (get_rsqrte_type): Likewise.
4162 (get_rsqrts_type): Likewise.
4163 (get_recpe_type): Likewise.
4164 (get_recps_type): Likewise.
4165 (aarch64_gimplify_va_arg_expr): Likewise.
4166 (aarch64_simd_container_mode): Likewise.
4167 (aarch64_emit_load_exclusive): Likewise.
4168 (aarch64_emit_store_exclusive): Likewise.
4169 (aarch64_expand_compare_and_swap): Likewise.
4170 (aarch64_gen_atomic_cas): Likewise.
4171 (aarch64_emit_bic): Likewise.
4172 (aarch64_emit_atomic_swap): Likewise.
4173 (aarch64_emit_atomic_load_op): Likewise.
4174 (aarch64_evpc_trn): Likewise.
4175 (aarch64_evpc_uzp): Likewise.
4176 (aarch64_evpc_zip): Likewise.
4177 (aarch64_evpc_ext): Likewise.
4178 (aarch64_evpc_rev): Likewise.
4179 (aarch64_evpc_dup): Likewise.
4180 (aarch64_gen_ccmp_first): Likewise.
4181 (aarch64_gen_ccmp_next): Likewise.
4182 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4183 (alpha_emit_xfloating_libcall): Likewise.
4184 (emit_insxl): Likewise.
4185 (alpha_arg_type): Likewise.
4186 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4187 (arc_preferred_simd_mode): Likewise.
4188 (arc_secondary_reload): Likewise.
4189 (get_arc_condition_code): Likewise.
4190 (arc_print_operand): Likewise.
4191 (arc_legitimate_constant_p): Likewise.
4192 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4193 * config/arc/arc.md (casesi_load): Likewise.
4194 (casesi_compact_jump): Likewise.
4195 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4196 (cc_use_register): Likewise.
4197 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4198 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4199 (arm_init_iwmmxt_builtins): Likewise.
4200 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4201 (neon_expand_vector_init): Likewise.
4202 (arm_attr_length_move_neon): Likewise.
4203 (maybe_get_arm_condition_code): Likewise.
4204 (arm_emit_vector_const): Likewise.
4205 (arm_preferred_simd_mode): Likewise.
4206 (arm_output_iwmmxt_tinsr): Likewise.
4207 (thumb1_output_casesi): Likewise.
4208 (thumb2_output_casesi): Likewise.
4209 (arm_emit_load_exclusive): Likewise.
4210 (arm_emit_store_exclusive): Likewise.
4211 (arm_expand_compare_and_swap): Likewise.
4212 (arm_evpc_neon_vuzp): Likewise.
4213 (arm_evpc_neon_vzip): Likewise.
4214 (arm_evpc_neon_vrev): Likewise.
4215 (arm_evpc_neon_vtrn): Likewise.
4216 (arm_evpc_neon_vext): Likewise.
4217 (arm_validize_comparison): Likewise.
4218 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4219 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4220 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4221 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4222 (c6x_preferred_simd_mode): Likewise.
4223 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4224 (epiphany_rtx_costs): Likewise.
4225 * config/epiphany/predicates.md (proper_comparison_operator):
4227 * config/frv/frv.c (condexec_memory_operand): Likewise.
4228 (frv_emit_move): Likewise.
4229 (output_move_single): Likewise.
4230 (output_condmove_single): Likewise.
4231 (frv_hard_regno_mode_ok): Likewise.
4232 (frv_matching_accg_mode): Likewise.
4233 * config/h8300/h8300.c (split_adds_subs): Likewise.
4234 (h8300_rtx_costs): Likewise.
4235 (h8300_print_operand): Likewise.
4236 (compute_mov_length): Likewise.
4237 (output_logical_op): Likewise.
4238 (compute_logical_op_length): Likewise.
4239 (compute_logical_op_cc): Likewise.
4240 (h8300_shift_needs_scratch_p): Likewise.
4241 (output_a_shift): Likewise.
4242 (compute_a_shift_length): Likewise.
4243 (compute_a_shift_cc): Likewise.
4244 (expand_a_rotate): Likewise.
4245 (output_a_rotate): Likewise.
4246 * config/i386/i386.c (classify_argument): Likewise.
4247 (function_arg_advance_32): Likewise.
4248 (function_arg_32): Likewise.
4249 (function_arg_64): Likewise.
4250 (function_value_64): Likewise.
4251 (ix86_gimplify_va_arg): Likewise.
4252 (ix86_legitimate_constant_p): Likewise.
4253 (put_condition_code): Likewise.
4254 (split_double_mode): Likewise.
4255 (ix86_avx256_split_vector_move_misalign): Likewise.
4256 (ix86_expand_vector_logical_operator): Likewise.
4257 (ix86_split_idivmod): Likewise.
4258 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4259 (ix86_build_const_vector): Likewise.
4260 (ix86_build_signbit_mask): Likewise.
4261 (ix86_match_ccmode): Likewise.
4262 (ix86_cc_modes_compatible): Likewise.
4263 (ix86_expand_branch): Likewise.
4264 (ix86_expand_sse_cmp): Likewise.
4265 (ix86_expand_sse_movcc): Likewise.
4266 (ix86_expand_int_sse_cmp): Likewise.
4267 (ix86_expand_vec_perm_vpermi2): Likewise.
4268 (ix86_expand_vec_perm): Likewise.
4269 (ix86_expand_sse_unpack): Likewise.
4270 (ix86_expand_int_addcc): Likewise.
4271 (ix86_split_to_parts): Likewise.
4272 (ix86_vectorize_builtin_gather): Likewise.
4273 (ix86_vectorize_builtin_scatter): Likewise.
4274 (avx_vpermilp_parallel): Likewise.
4275 (inline_memory_move_cost): Likewise.
4276 (ix86_tieable_integer_mode_p): Likewise.
4277 (x86_maybe_negate_const_int): Likewise.
4278 (ix86_expand_vector_init_duplicate): Likewise.
4279 (ix86_expand_vector_init_one_nonzero): Likewise.
4280 (ix86_expand_vector_init_one_var): Likewise.
4281 (ix86_expand_vector_init_concat): Likewise.
4282 (ix86_expand_vector_init_interleave): Likewise.
4283 (ix86_expand_vector_init_general): Likewise.
4284 (ix86_expand_vector_set): Likewise.
4285 (ix86_expand_vector_extract): Likewise.
4286 (emit_reduc_half): Likewise.
4287 (ix86_emit_i387_round): Likewise.
4288 (ix86_mangle_type): Likewise.
4289 (ix86_expand_round_sse4): Likewise.
4290 (expand_vec_perm_blend): Likewise.
4291 (canonicalize_vector_int_perm): Likewise.
4292 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4293 (expand_vec_perm_1): Likewise.
4294 (expand_vec_perm_interleave3): Likewise.
4295 (expand_vec_perm_even_odd_pack): Likewise.
4296 (expand_vec_perm_even_odd_1): Likewise.
4297 (expand_vec_perm_broadcast_1): Likewise.
4298 (ix86_vectorize_vec_perm_const_ok): Likewise.
4299 (ix86_expand_vecop_qihi): Likewise.
4300 (ix86_expand_mul_widen_hilo): Likewise.
4301 (ix86_expand_sse2_abs): Likewise.
4302 (ix86_expand_pextr): Likewise.
4303 (ix86_expand_pinsr): Likewise.
4304 (ix86_preferred_simd_mode): Likewise.
4305 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4306 * config/i386/sse.md (*andnot<mode>3): Likewise.
4307 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4308 (*<code><mode>3): Likewise.
4309 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4310 (ia64_expand_atomic_op): Likewise.
4311 (ia64_arg_type): Likewise.
4312 (ia64_mode_to_int): Likewise.
4313 (ia64_scalar_mode_supported_p): Likewise.
4314 (ia64_vector_mode_supported_p): Likewise.
4315 (expand_vec_perm_broadcast): Likewise.
4316 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4317 (iq2000_function_arg_advance): Likewise.
4318 (iq2000_function_arg): Likewise.
4319 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4320 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4321 (m68k_libcall_value): Likewise.
4322 (m68k_function_value): Likewise.
4323 (sched_attr_op_type): Likewise.
4324 * config/mcore/mcore.c (mcore_output_move): Likewise.
4325 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4327 (microblaze_function_arg): Likewise.
4328 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4329 (mips_print_operand): Likewise.
4330 (mips_mode_ok_for_mov_fmt_p): Likewise.
4331 (mips_vector_mode_supported_p): Likewise.
4332 (mips_preferred_simd_mode): Likewise.
4333 (mips_expand_vpc_loongson_even_odd): Likewise.
4334 (mips_expand_vec_unpack): Likewise.
4335 (mips_expand_vi_broadcast): Likewise.
4336 (mips_expand_vector_init): Likewise.
4337 (mips_expand_vec_reduc): Likewise.
4338 (mips_expand_msa_cmp): Likewise.
4339 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4340 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4341 (cc_flags_for_mode): Likewise.
4342 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4343 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4344 (nds32_output_casesi_pc_relative): Likewise.
4345 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4346 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4347 (nvptx_gen_unpack): Likewise.
4348 (nvptx_gen_pack): Likewise.
4349 (nvptx_gen_shuffle): Likewise.
4350 (nvptx_gen_wcast): Likewise.
4351 (nvptx_preferred_simd_mode): Likewise.
4352 * config/pa/pa.c (pa_secondary_reload): Likewise.
4353 * config/pa/predicates.md (base14_operand): Likewise.
4354 * config/powerpcspe/powerpcspe-c.c
4355 (altivec_resolve_overloaded_builtin): Likewise.
4356 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4358 (rs6000_preferred_simd_mode): Likewise.
4359 (output_vec_const_move): Likewise.
4360 (rs6000_expand_vector_extract): Likewise.
4361 (rs6000_split_vec_extract_var): Likewise.
4362 (reg_offset_addressing_ok_p): Likewise.
4363 (rs6000_legitimate_offset_address_p): Likewise.
4364 (rs6000_legitimize_address): Likewise.
4365 (rs6000_emit_set_const): Likewise.
4366 (rs6000_const_vec): Likewise.
4367 (rs6000_emit_move): Likewise.
4368 (spe_build_register_parallel): Likewise.
4369 (rs6000_darwin64_record_arg_recurse): Likewise.
4370 (swap_selector_for_mode): Likewise.
4371 (spe_init_builtins): Likewise.
4372 (paired_init_builtins): Likewise.
4373 (altivec_init_builtins): Likewise.
4374 (do_load_for_compare): Likewise.
4375 (rs6000_generate_compare): Likewise.
4376 (rs6000_expand_float128_convert): Likewise.
4377 (emit_load_locked): Likewise.
4378 (emit_store_conditional): Likewise.
4379 (rs6000_output_function_epilogue): Likewise.
4380 (rs6000_handle_altivec_attribute): Likewise.
4381 (rs6000_function_value): Likewise.
4382 (emit_fusion_gpr_load): Likewise.
4383 (emit_fusion_p9_load): Likewise.
4384 (emit_fusion_p9_store): Likewise.
4385 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4386 (fusion_gpr_mem_load): Likewise.
4387 (fusion_addis_mem_combo_load): Likewise.
4388 (fusion_addis_mem_combo_store): Likewise.
4389 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4390 (fusion_gpr_mem_load): Likewise.
4391 (fusion_addis_mem_combo_load): Likewise.
4392 (fusion_addis_mem_combo_store): Likewise.
4393 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4395 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4396 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4397 (rs6000_preferred_simd_mode): Likewise.
4398 (output_vec_const_move): Likewise.
4399 (rs6000_expand_vector_extract): Likewise.
4400 (rs6000_split_vec_extract_var): Likewise.
4401 (reg_offset_addressing_ok_p): Likewise.
4402 (rs6000_legitimate_offset_address_p): Likewise.
4403 (rs6000_legitimize_address): Likewise.
4404 (rs6000_emit_set_const): Likewise.
4405 (rs6000_const_vec): Likewise.
4406 (rs6000_emit_move): Likewise.
4407 (rs6000_darwin64_record_arg_recurse): Likewise.
4408 (swap_selector_for_mode): Likewise.
4409 (paired_init_builtins): Likewise.
4410 (altivec_init_builtins): Likewise.
4411 (rs6000_expand_float128_convert): Likewise.
4412 (emit_load_locked): Likewise.
4413 (emit_store_conditional): Likewise.
4414 (rs6000_output_function_epilogue): Likewise.
4415 (rs6000_handle_altivec_attribute): Likewise.
4416 (rs6000_function_value): Likewise.
4417 (emit_fusion_gpr_load): Likewise.
4418 (emit_fusion_p9_load): Likewise.
4419 (emit_fusion_p9_store): Likewise.
4420 * config/rx/rx.c (rx_gen_move_template): Likewise.
4421 (flags_from_mode): Likewise.
4422 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4423 (s390_slb_comparison): Likewise.
4424 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4425 (s390_vector_mode_supported_p): Likewise.
4426 (s390_cc_modes_compatible): Likewise.
4427 (s390_match_ccmode_set): Likewise.
4428 (s390_canonicalize_comparison): Likewise.
4429 (s390_emit_compare_and_swap): Likewise.
4430 (s390_branch_condition_mask): Likewise.
4431 (s390_rtx_costs): Likewise.
4432 (s390_secondary_reload): Likewise.
4433 (__SECONDARY_RELOAD_CASE): Likewise.
4434 (s390_expand_cs): Likewise.
4435 (s390_preferred_simd_mode): Likewise.
4436 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4437 * config/sh/sh.c (sh_print_operand): Likewise.
4438 (dump_table): Likewise.
4439 (sh_secondary_reload): Likewise.
4440 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4441 * config/sh/sh.md (casesi_worker_1): Likewise.
4442 (casesi_worker_2): Likewise.
4443 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4444 (fcc_comparison_operator): Likewise.
4445 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4446 (emit_soft_tfmode_cvt): Likewise.
4447 (sparc_preferred_simd_mode): Likewise.
4448 (output_cbranch): Likewise.
4449 (sparc_print_operand): Likewise.
4450 (sparc_expand_vec_perm_bmask): Likewise.
4451 (vector_init_bshuffle): Likewise.
4452 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4453 (spu_vector_mode_supported_p): Likewise.
4454 (spu_expand_insv): Likewise.
4455 (spu_emit_branch_or_set): Likewise.
4456 (spu_handle_vector_attribute): Likewise.
4457 (spu_builtin_splats): Likewise.
4458 (spu_builtin_extract): Likewise.
4459 (spu_builtin_promote): Likewise.
4460 (spu_expand_sign_extend): Likewise.
4461 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4462 (tilegx_simd_int): Likewise.
4463 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4464 (tilepro_simd_int): Likewise.
4465 * config/v850/v850.c (const_double_split): Likewise.
4466 (v850_print_operand): Likewise.
4467 (ep_memory_offset): Likewise.
4468 * config/vax/vax.c (vax_rtx_costs): Likewise.
4469 (vax_output_int_move): Likewise.
4470 (vax_output_int_add): Likewise.
4471 (vax_output_int_subtract): Likewise.
4472 * config/visium/predicates.md (visium_branch_operator): Likewise.
4473 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4474 (visium_print_operand_address): Likewise.
4475 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4476 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4477 (xtensa_expand_conditional_branch): Likewise.
4478 (xtensa_copy_incoming_a7): Likewise.
4479 (xtensa_output_literal): Likewise.
4480 * dfp.c (decimal_real_maxval): Likewise.
4481 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4484 Alan Hayward <alan.hayward@arm.com>
4485 David Sherwood <david.sherwood@arm.com>
4487 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4488 (mode_nunits_inline): Likewise.
4489 (mode_inner_inline): Likewise.
4490 (mode_unit_size_inline): Likewise.
4491 (mode_unit_precision_inline): Likewise.
4492 (emit_insn_modes_h): Likewise. Also emit a #define of the
4494 (emit_mode_wider): Add an E_ prefix to mode names.
4495 (emit_mode_complex): Likewise.
4496 (emit_mode_inner): Likewise.
4497 (emit_mode_adjustments): Likewise.
4498 (emit_mode_int_n): Likewise.
4499 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4500 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4501 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4502 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4503 (CRC32_BUILTIN, ENTRY): Likewise.
4504 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4505 (aarch64_pop_regs): Likewise.
4506 (aarch64_process_components): Likewise.
4507 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4508 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4509 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4510 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4511 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4512 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4514 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4515 (pcmpistr): Likewise.
4516 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4517 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4518 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4520 * config/rl78/rl78.c (mduc_regs): Likewise.
4521 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4522 (htm_expand_builtin): Likewise.
4523 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4524 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4525 * config/xtensa/xtensa.c (print_operand): Likewise.
4526 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4527 (NUM_MODE_VECTOR_INT): Likewise.
4528 * genoutput.c (null_operand): Likewise.
4529 (output_operand_data): Likewise.
4530 * genrecog.c (print_parameter_value): Likewise.
4531 * lra.c (debug_operand_data): Likewise.
4533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4535 * dwarf2out.c (scompare_loc_descriptor_wide)
4536 (scompare_loc_descriptor_narrow): New functions, split out from...
4537 (scompare_loc_descriptor): ...here.
4538 * expmed.c (emit_store_flag_int): New function, split out from...
4539 (emit_store_flag): ...here.
4541 2017-08-30 Richard Biener <rguenther@suse.de>
4543 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4544 (dwarf2out_early_finish): Move setting of AT_pubnames from
4545 early debug output to early finish.
4547 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4549 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4550 and -mdata-region to the assembler.
4552 2017-08-30 Richard Biener <rguenther@suse.de>
4554 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4556 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4557 (dwarf2out_early_global_decl): Only generate a DIE for the
4558 abstract origin if it doesn't already exist or is a declaration DIE.
4559 (resolve_addr): Do not add the linkage name twice when
4560 generating a stub DIE for the DW_TAG_GNU_call_site target.
4562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4564 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4565 Use machine_mode rather than int for arg1_mode.
4567 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4570 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4571 that the second argument of the built-in functions to unpack
4572 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4573 switch statement instead a lot of if statements.
4574 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4575 Allow 64-bit values to be in Altivec registers as well as
4576 traditional floating point registers.
4577 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4579 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4581 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4582 MAX_REGS_PER_ADDRESS == 1.
4584 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4586 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4587 * config/i386/i386.c (ix86_option_override_internal): Simplify
4588 setting of opts->x_flag_entry.
4590 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4591 Jakub Jelinek <jakub@redhat.com>
4592 Richard Biener <rguenther@suse.de>
4594 PR tree-optimization/81503
4595 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4596 folded constant fits in the target type; reorder tests for clarity.
4598 2017-08-29 Martin Liska <mliska@suse.cz>
4600 * passes.def: Include pass_lower_switch.
4601 * stmt.c (dump_case_nodes): Remove and move to
4602 tree-switch-conversion.
4603 (case_values_threshold): Likewise.
4604 (expand_switch_as_decision_tree_p): Likewise.
4605 (emit_case_decision_tree): Likewise.
4606 (expand_case): Likewise.
4607 (balance_case_nodes): Likewise.
4608 (node_has_low_bound): Likewise.
4609 (node_has_high_bound): Likewise.
4610 (node_is_bounded): Likewise.
4611 (emit_case_nodes): Likewise.
4612 (struct simple_case_node): New struct.
4613 (add_case_node): Remove.
4614 (emit_case_dispatch_table): Use vector instead of case_list.
4615 (reset_out_edges_aux): Remove.
4616 (compute_cases_per_edge): Likewise.
4617 (expand_case): Build list of simple_case_node.
4618 (expand_sjlj_dispatch_table): Use it.
4619 * tree-switch-conversion.c (struct case_node): Moved from
4620 stmt.c and adjusted.
4621 (emit_case_nodes): Likewise.
4622 (node_has_low_bound): Likewise.
4623 (node_has_high_bound): Likewise.
4624 (node_is_bounded): Likewise.
4625 (case_values_threshold): Likewise.
4626 (reset_out_edges_aux): Likewise.
4627 (compute_cases_per_edge): Likewise.
4628 (add_case_node): Likewise.
4629 (dump_case_nodes): Likewise.
4630 (balance_case_nodes): Likewise.
4631 (expand_switch_as_decision_tree_p): Likewise.
4632 (emit_jump): Likewise.
4633 (emit_case_decision_tree): Likewise.
4634 (try_switch_expansion): Likewise.
4635 (do_jump_if_equal): Likewise.
4636 (emit_cmp_and_jump_insns): Likewise.
4637 (fix_phi_operands_for_edge): New function.
4638 (record_phi_operand_mapping): Likewise.
4639 (class pass_lower_switch): New pass.
4640 (pass_lower_switch::execute): New function.
4641 (make_pass_lower_switch): Likewise.
4642 (conditional_probability):
4643 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4644 * tree-pass.h: Add make_pass_lower_switch.
4646 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4649 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4652 2017-08-29 Richard Biener <rguenther@suse.de>
4654 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4655 we do not add a DW_AT_inline attribute twice.
4656 (gen_subprogram_die): Remove code setting DW_AT_inline on
4657 DECL_ABSTRACT_P nodes.
4659 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4661 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4662 calls to internal functions.
4663 (gimplify_modify_expr): Likewise.
4664 * tree-call-cdce.c (use_internal_fn): Likewise.
4665 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4666 (convert_to_divmod): Set the nothrow flag.
4667 * tree-if-conv.c (predicate_mem_writes): Likewise.
4668 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4669 (vectorizable_call): Likewise.
4670 (vectorizable_store): Likewise.
4671 (vectorizable_load): Likewise.
4672 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4673 (vect_recog_mask_conversion_pattern): Likewise.
4675 2017-08-29 Martin Liska <mliska@suse.cz>
4678 * gcc.c (driver_handle_option): Add new argument.
4679 * opts-common.c (handle_option): Pass
4680 target_option_override_hook.
4681 * opts-global.c (lang_handle_option): Add new option.
4682 (set_default_handlers): Add new argument.
4683 (decode_options): Likewise.
4684 * opts.c (target_handle_option): Likewise.
4685 (common_handle_option): Call target_option_override_hook.
4686 * opts.h (struct cl_option_handler_func): Add hook for
4687 target option override.
4688 (struct cl_option_handlers): Likewise.
4689 (set_default_handlers): Add new argument.
4690 (decode_options): Likewise.
4691 (common_handle_option): Likewise.
4692 (target_handle_option): Likewise.
4693 * toplev.c (toplev::main): Pass targetm.target_option.override
4696 2017-08-29 Richard Biener <rguenther@suse.de>
4697 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4699 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4700 life to the active subtree.
4702 2017-08-28 Jeff Law <law@redhat.com>
4704 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4705 derive_equivalences.
4706 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4708 (edge_info::derive_equivalences): New private member function
4710 * tree-ssa-dom.c (class edge_info): Changed from a struct
4711 to a class. Add ctor/dtor, methods and data members.
4712 (edge_info::edge_info): Renamed from allocate_edge_info.
4713 Initialize additional members.
4714 (edge_info::~edge_info): New.
4715 (free_dom_edge_info): Delete the edge info.
4716 (record_edge_info): Use new class & associated member functions.
4717 Tighten forms for testing for edge equivalences.
4718 (record_temporary_equivalences): Iterate over the simple
4719 equivalences rather than assuming there's only one per edge.
4720 (cprop_into_successor_phis): Iterate over the simple
4721 equivalences rather than assuming there's only one per edge.
4722 (optimize_stmt): Use operand_equal_p rather than pointer
4723 equality for mini-DSE code.
4725 2017-08-28 Nathan Sidwell <nathan@acm.org>
4727 * gcc.c (execute): Fold SIGPIPE handling into switch
4728 statement. Adjust internal error message.
4730 2017-08-28 Richard Biener <rguenther@suse.de>
4733 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4734 Do nothing for removed DIEs.
4736 2017-08-28 Richard Biener <rguenther@suse.de>
4738 PR tree-optimization/81977
4739 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4742 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4745 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4746 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4747 using targetm.gen_mem_thread_fence.
4749 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4752 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4753 predicate to register_operand. Reorder operands.
4754 (*btr<mode>): Ditto.
4755 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4756 (*btr<mode>_mask): Ditto.
4758 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4760 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4761 * config/rs6000/xmmintrin.h: New file.
4762 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4764 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4767 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4768 parameter and_insn and return it.
4769 (recombine_lvx_pattern): Insert a copy to ensure availability of
4770 the base register of the copied masking operation at the point of
4771 the instruction replacement.
4772 (recombine_stvx_pattern): Likewise.
4774 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4776 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4777 undocumented switches.
4778 (-mpower9-dform-vector): Likewise.
4779 (-mpower9-dform): Likewise.
4780 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4781 comments to delete references to -mpower9-dform* switches.
4782 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4783 Delete reference to -mpower9-dform* switches, test for
4784 -mpower9-vector instead.
4785 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4786 (OTHER_P9_VECTOR_MASKS): Likewise.
4787 (POWERPC_MASKS): Likewise.
4788 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4789 tests against -mpower9-dform* to -mpower9-vector. Delete code
4790 that checked for -mpower9-dform* consistancy with other options.
4791 Add test for -mpower9-misc to enable other power9 switches.
4792 (rs6000_init_hard_regno_mode_ok): Likewise.
4793 (rs6000_option_override_internal): Likewise.
4794 (rs6000_emit_prologue): Likewise.
4795 (rs6000_emit_epilogue): Likewise.
4796 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4797 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4798 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4800 (emit_fusion_p9_store): Likewise.
4801 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4802 resetting these macros if the assembler does not support ISA 3.0
4804 (TARGET_P9_DFORM_VECTOR): Likewise.
4805 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4806 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4808 2017-08-25 Alan Modra <amodra@gmail.com>
4810 PR rtl-optimization/81747
4811 * cse.c (cse_extended_basic_block): Don't attempt to record
4812 equivalences for degenerate conditional branches that branch
4813 to their fall-through.
4815 2017-08-24 Martin Sebor <msebor@redhat.com>
4818 * gimple-fold.c (size_must_be_zero_p): New function.
4819 (gimple_fold_builtin_memory_op): Call it.
4821 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4823 * config/rs6000/mm_malloc.h: New file.
4825 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4827 PR tree-optimization/81913
4828 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4829 analysis when either IVs in condition can wrap.
4831 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4833 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4834 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4836 2017-08-24 Richard Biener <rguenther@suse.de>
4839 * targhooks.c (default_target_can_inline_p): Properly
4840 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4841 is present and always compare.
4842 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4843 infer -mfpmath=sse from TARGET_SSE_P.
4844 (ix86_can_inline_p): Properly use target_option_default_node when
4845 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4847 2017-08-24 Richard Biener <rguenther@suse.de>
4850 * dwarf2out.c (output_die): Handle flag_generate_offload like
4852 (output_comp_unit): Likewise.
4853 (gen_array_type_die): Likewise.
4854 (dwarf2out_early_finish): Likewise.
4855 (note_variable_value_in_expr): Likewise.
4856 (dwarf2out_finish): Likewise. Adjust assert.
4857 * cgraphunit.c (symbol_table::compile): Move setting of
4858 flag_generate_offload earlier ...
4859 (symbol_table::finalize_compilation_unit): ... here, before
4860 early debug finalization.
4862 2017-08-24 Richard Biener <rguenther@suse.de>
4864 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4865 and ipa-fnsummary.h.
4866 (ix86_can_inline_p): When ix86_fpmath flags do not match
4867 check whether the callee uses FP math at all.
4869 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4872 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4873 instead of TYPE_PRECISION.
4875 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4876 Alan Hayward <alan.hayward@arm.com>
4877 David Sherwood <david.sherwood@arm.com>
4879 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4881 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4883 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4885 (ix86_frame::stack_realign_allocate): New field.
4886 (struct machine_frame_state): Modify comments.
4887 (machine_frame_state::sp_realigned_fp_end): New field.
4888 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4890 (sp_valid_at): Add assertion to assure no attempt to access invalid
4891 offset of a realigned stack.
4892 (fp_valid_at): Likewise.
4893 (choose_baseaddr): Modify comments.
4894 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4895 ix86_expand_prologue.
4896 (ix86_expand_prologue): Modify stack realignment and allocation.
4897 (ix86_expand_epilogue): Modify comments.
4898 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4899 avx2_runtime, avx512f, and avx512f_runtime.
4901 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4903 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4904 (mstackrealign): Do not init to -1.
4905 * config/i386/i386.c (ix86_option_override_internal):
4906 Check opts_set, not opts when setting default value of
4907 opts->x_ix86_force_align_arg_pointer.
4909 2017-08-23 Richard Biener <rguenther@suse.de>
4911 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4912 lang_hooks.decl_printable_name.
4913 * print-rtl-function.c (print_rtx_function): Likewise.
4914 * tree-pretty-print.c (dump_function_header): Likewise.
4916 2017-08-23 Richard Biener <rguenther@suse.de>
4919 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4920 -g0 at compile-time.
4922 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4925 * doc/sourcebuild.texi (Other hardware attributes):
4928 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4931 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4932 Fix single-use check.
4934 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4936 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4938 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4940 * doc/install.texi: Modify to add more details on running selected
4943 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4945 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4946 is combined with -mabi=ms.
4947 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4950 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4952 PR tree-optimization/81488
4953 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4954 and cached_basis fields.
4955 (MAX_SPREAD): New constant.
4956 (alloc_cand_and_find_basis): Initialize new fields.
4957 (clear_visited): New function.
4958 (create_phi_basis_1): Rename from create_phi_basis, set visited
4959 and cached_basis fields.
4960 (create_phi_basis): New wrapper function.
4961 (phi_add_costs_1): Rename from phi_add_costs, add spread
4962 parameter, set visited field, short-circuit when limits reached.
4963 (phi_add_costs): New wrapper function.
4964 (record_phi_increments_1): Rename from record_phi_increments, set
4966 (record_phi_increments): New wrapper function.
4967 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4968 (phi_incr_cost): New wrapper function.
4969 (all_phi_incrs_profitable_1): Rename from
4970 all_phi_incrs_profitable, set visited field.
4971 (all_phi_incrs_profitable): New wrapper function.
4973 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4974 Alan Hayward <alan.hayward@arm.com>
4975 David Sherwood <david.sherwood@arm.com>
4977 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4978 that takes the outer and inner modes.
4979 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4980 comparison as the canonical test for a paradoxical subreg.
4981 * combine.c (simplify_set): Use paradoxical_subreg_p.
4982 (make_extraction): Likewise.
4983 (force_to_mode): Likewise.
4984 (rtx_equal_for_field_assignment_p): Likewise.
4985 (gen_lowpart_for_combine): Likewise.
4986 (simplify_comparison): Likewise.
4987 * cse.c (equiv_constant): Likewise.
4988 * expmed.c (store_bit_field_1): Likewise.
4989 * final.c (alter_subreg): Likewise.
4990 * fwprop.c (propagate_rtx): Likewise.
4991 (forward_propagate_subreg): Likewise.
4992 * ira-conflicts.c (ira_build_conflicts): Likewise.
4993 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4994 * lra-constraints.c (curr_insn_transform): Likewise.
4995 (split_reg): Likewise.
4996 * lra-eliminations.c (move_plus_up): Likewise.
4997 (lra_eliminate_regs_1): Likewise.
4998 * recog.c (general_operand): Likewise.
4999 * ree.c (combine_reaching_defs): Likewise.
5000 * reload.c (push_reload): Likewise.
5001 (find_reloads): Likewise.
5002 * reload1.c (elimination_effects): Likewise.
5003 (compute_reload_subreg_offset): Likewise.
5004 (choose_reload_regs): Likewise.
5005 * rtlanal.c (subreg_lsb_1): Likewise.
5006 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5007 (simplify_subreg): Likewise.
5008 * var-tracking.c (track_loc_p): Likewise.
5009 * emit-rtl.c (byte_lowpart_offset): Likewise.
5010 (paradoxical_subreg_p): Delete out-of-line definition.
5012 2017-08-22 Jeff Law <law@redhat.com>
5014 PR tree-optimization/81741
5015 PR tree-optimization/71947
5016 * tree-ssa-dom.c: Include tree-inline.h.
5017 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5018 equivalences if one is more expensive to compute than the other.
5019 * tree-ssa-scopedtables.h (class const_or_copies): Make
5020 record_const_or_copy_raw method private.
5021 (class avail_exprs_stack): New method simplify_binary_operation.
5022 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5023 avail_exprs_stack::simplify_binary_operation as needed.
5024 (avail_exprs_stack::simplify_binary_operation): New function.
5026 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5028 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5029 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5030 (DOT_SYMBOLS): Likewise.
5031 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5032 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5033 (RS6000_ABI_NAME): Likewise.
5034 (TARGET_CMODEL): Likewise.
5035 (TOC_SECTION_ASM_OP): Likewise.
5036 (SET_CMODEL): New macro.
5037 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5039 2017-08-22 Richard Biener <rguenther@suse.de>
5041 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5042 to allow for free-lang-data replacements similar to verify_type_variant.
5044 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5045 Alan Hayward <alan.hayward@arm.com>
5046 David Sherwood <david.sherwood@arm.com>
5048 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5049 VOIDmode for the LABEL_REF.
5051 2017-08-22 Richard Biener <rguenther@suse.de>
5053 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5055 2017-08-22 Richard Biener <rguenther@suse.de>
5057 * common.opt (feliminate-dwarf2-dups): Ignore.
5058 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5059 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5060 same_die_p_wrap, compute_section_prefix,
5061 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5062 (comdat_symbol_id, comdat_symbol_number): Likewise.
5063 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5065 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5066 (output_die): Mark unreachable path unreachable.
5067 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5068 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5069 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5070 (dwarf2out_early_finish): Likewise.
5072 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5074 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5076 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5079 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5080 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5081 (avr_attribute_table) <io, io_low, address>: Initialize
5082 .decl_required with true.
5084 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5086 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5087 undocumented debugging options.
5088 (-mvsx-scalar-double): Likewise.
5089 (-mallow-df-permute): Likewise.
5090 (-mvectorize-builtins): Likewise.
5091 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5092 (rs6000_builtin_vectorized_function): Likewise.
5093 (rs6000_builtin_md_vectorized_function): Likewise.
5094 (rs6000_opt_vars): Likewise.
5096 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5099 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5100 (*btrq_imm): Rename from *btrq.
5101 (*btcq_imm): Rename from *btcq.
5102 (btsc): New code attribute.
5103 (*<btsc><mode>): New insn pattern.
5104 (*btr<mode>): Ditto.
5105 (*<btsc><mode>_mask): New insn_and_split pattern.
5106 (*btr<mode>_mask): Ditto.
5108 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5109 Alan Hayward <alan.hayward@arm.com>
5110 David Sherwood <david.sherwood@arm.com>
5112 * function.c (pad_below): Simplify padding calculation.
5114 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5115 Alan Hayward <alan.hayward@arm.com>
5116 David Sherwood <david.sherwood@arm.com>
5118 * target.def (function_prologue): Remove frame size argument.
5119 (function_epilogue): Likewise.
5120 * doc/tm.texi: Regenerate.
5121 * final.c (final_start_function): Update call to function_prologue.
5122 (final_end_function): Update call to function_epilogue.
5123 (default_function_pro_epilogue): Remove frame size argument.
5124 * output.h (default_function_pro_epilogue): Likewise.
5125 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5126 (arm_output_function_prologue): Likewise.
5127 * config/frv/frv.c (frv_function_prologue): Likewise.
5128 (frv_function_epilogue): Likewise.
5129 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5130 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5131 (ia64_output_function_epilogue): Likewise.
5132 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5133 (m32r_output_function_epilogue): Likewise.
5134 * config/microblaze/microblaze.c (microblaze_function_prologue)
5135 (microblaze_function_epilogue): Likewise.
5136 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5137 (mips_output_function_epilogue): Likewise.
5138 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5139 (mmix_target_asm_function_epilogue): Likewise.
5140 * config/msp430/msp430.c (msp430_start_function): Likewise.
5141 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5142 (nds32_asm_function_epilogue): Likewise.
5143 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5144 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5145 (pa_output_function_epilogue): Likewise.
5146 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5147 (rs6000_output_function_epilogue): Likewise.
5148 * config/rl78/rl78.c (rl78_start_function): Likewise.
5149 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5150 (rs6000_output_function_epilogue): Likewise.
5151 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5152 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5153 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5154 (sparc_asm_function_epilogue): Likewise.
5156 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5158 * tree.h (type_has_mode_precision_p): New function.
5159 * convert.c (convert_to_integer_1): Use it.
5160 * expr.c (expand_expr_real_2): Likewise.
5161 (expand_expr_real_1): Likewise.
5162 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5163 * match.pd: Likewise.
5164 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5165 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5166 * tree-tailcall.c (process_assignment): Likewise.
5167 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5168 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5169 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5170 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5171 (vectorizable_assignment): Likewise.
5172 (vectorizable_shift): Likewise.
5173 (vectorizable_operation): Likewise.
5174 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5176 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5178 * match.pd: Add pow (C, x) simplification.
5180 2017-08-21 Richard Biener <rguenther@suse.de>
5182 PR tree-optimization/81900
5183 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5184 for blocks with abnormal predecessors.
5185 (compute_antic): Do not set visited flag prematurely.
5187 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5190 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5192 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5194 * stor-layout.h (vector_type_mode): Move to...
5195 * tree.h (vector_type_mode): ...here.
5196 * stor-layout.c (vector_type_mode): Move to...
5197 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5199 2017-08-21 Richard Biener <rguenther@suse.de>
5201 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5202 register_external_die hooks.
5203 (debug_false_tree_charstarstar_uhwistar): Declare.
5204 (debug_nothing_tree_charstar_uhwi): Likewise.
5205 * debug.c (do_nothing_debug_hooks): Adjust.
5206 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5207 (debug_nothing_tree_charstar_uhwi): Likewise.
5208 * dbxout.c (dbx_debug_hooks): Adjust.
5209 (xcoff_debug_hooks): Likewise.
5210 * sdbout.c (sdb_debug_hooks): Likewise.
5211 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5212 * dwarf2out.c (macinfo_label_base): New global.
5213 (dwarf2out_register_external_die): New function for the
5214 register_external_die hook.
5215 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5216 (dwarf2_debug_hooks): Use them.
5217 (dwarf2_lineno_debug_hooks): Adjust.
5218 (struct die_struct): Add with_offset flag.
5219 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5220 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5221 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5222 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5223 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5224 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5225 defining section names for the early LTO debug variants.
5226 (reset_indirect_string): New helper.
5227 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5228 (print_dw_val): Add support for offsetted symbol references.
5229 (get_ultimate_context): Split out from is_cxx.
5230 (is_cxx): Use get_ultimate_context.
5231 (is_fortran): Add decl overload.
5232 (compute_comp_unit_symbol): Split out worker from
5233 compute_section_prefix.
5234 (compute_section_prefix): Call compute_comp_unit_symbol and
5235 set comdat_type_p here.
5236 (output_die): Skip DIE symbol output for the LTO added one.
5237 Handle DIE symbol references with offset.
5238 (output_comp_unit): Guard section name mangling properly.
5239 For LTO debug sections emit a symbol at the section beginning
5240 which we use to refer to its DIEs.
5241 (add_abstract_origin_attribute): For DIEs registered via
5242 dwarf2out_register_external_die directly refer to the early
5243 DIE rather than indirectly through the shadow one we created.
5244 Remove obsolete call to dwarf2out_abstract_function for
5245 non-function/block origins.
5246 (gen_array_type_die): When generating early LTO debug do
5247 not emit DW_AT_string_length.
5248 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5249 late when in LTO. As suggested place a gcc_unreachable for
5250 the DECL_ABSTRACT_P case.
5251 (gen_subprogram_die): Avoid another specification DIE
5252 for early built declarations/definitions for the late LTO case.
5253 (gen_variable_die): Add type references for late duplicated VLA dies
5255 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5256 we have the abstract instance already.
5257 (process_scope_var): Adjust decl DIE contexts in LTO which
5258 first puts them in limbo.
5259 (gen_decl_die): Do not generate type DIEs late apart from
5260 types for VLAs or for decls we do not yet have a DIE. Do not
5261 call dwarf2out_abstract_function late.
5262 (dwarf2out_early_global_decl): Make sure to create DIEs
5263 for abstract instances of a decl first.
5264 (dwarf2out_late_global_decl): Adjust comment.
5265 (output_macinfo_op): With multiple macro sections use
5266 macinfo_label_base to distinguish labels.
5267 (output_macinfo): Likewise. Update macinfo_label_base.
5268 Pass in the line info label.
5269 (note_variable_value_in_expr): When generating LTO resolve
5270 all variable values here by generating DIEs as needed.
5271 (init_sections_and_labels): Add early LTO debug flag parameter
5272 and generate different sections and names if set. Add generation
5273 counter for the labels so we can have multiple of them.
5274 (reset_dies): Helper to allow DIEs to be output multiple times.
5275 (dwarf2out_finish): When outputting DIEs to the fat part of an
5276 LTO object first reset DIEs.
5277 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5278 (modified_type_die): Check for decl_ultimate_origin being self
5280 (gen_type_die_with_usage): Likewise.
5281 (gen_typedef_die): Allow decl_ultimate_origin being self.
5282 (set_decl_abstract_flags): Remove.
5283 (set_block_abstract_flags): Likewise.
5284 (dwarf2out_abstract_function): Treat the early generated DIEs
5285 as the abstract copy and only add DW_AT_inline and
5286 DW_AT_artificial here and call set_decl_origin_self.
5287 If the DIE has an abstract origin don't do anything.
5288 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5289 if we have none yet (Go fails to build one, PR78628).
5290 (variably_modified_type_p): Prevent endless recursion for Ada
5291 cyclic pointer types.
5292 * lto-streamer-in.c: Include debug.h.
5293 (dref_queue): New global.
5294 (lto_read_tree_1): Stream in DIE references.
5295 (lto_input_tree): Register DIE references.
5296 (input_function): Stream DECL_DEBUG_ARGS.
5297 * lto-streamer-out.c: Include debug.h.
5298 (lto_write_tree_1): Output DIE references.
5299 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5300 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5301 (output_function): Stream DECL_DEBUG_ARGS.
5302 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5303 Stream DECL_ABSTRACT_ORIGIN.
5304 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5305 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5306 DECL_CONTEXT for file-scope decls.
5307 * lto-streamer.h (struct dref_entry): Declare.
5308 (dref_queue): Likewise.
5309 * cfgexpand.c (pass_expand::execute): Do not call the
5310 outlining_inline_function hook here.
5311 * lto-wrapper.c (debug_obj): New global.
5312 (tool_cleanup): Unlink it if required.
5313 (debug_objcopy): New function.
5314 (run_gcc): Handle early debug sections in the IL files by
5315 extracting them to separate files, partially linkin them and
5316 feeding the result back as result to the linker.
5317 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5318 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5319 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5320 sections into a separate segment.
5321 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5323 (darwin_asm_dwarf_section): Likewise.
5324 (darwin_asm_output_dwarf_offset): Likewise.
5325 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5327 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5328 Alan Hayward <alan.hayward@arm.com>
5329 David Sherwood <david.sherwood@arm.com>
5331 * read-md.h (md_reader::record_potential_iterator_use): Replace
5332 pointer argument with an rtx and an index.
5333 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5334 (apply_mode_iterator): Likewise.
5335 (apply_code_iterator): Likewise.
5336 (apply_int_iterator): Likewise.
5337 (apply_subst_iterator): Likewise.
5338 (record_iterator_use): Likewise.
5339 (record_attribute_use): Likewise.
5340 (md_reader::record_potential_iterator_use): Likewise. Update calls
5341 to record_iterator_use and apply_iterator.
5342 (iterator_use): Replace ptr with x and index.
5343 (attribute_use): Likewise.
5344 (apply_attribute_uses): Update calls to apply_iterator.
5345 (apply_iterators): Likewise. Update initialization of iterator_use.
5346 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5347 and record_potential_iterator_use.
5348 (rtx_reader::read_rtx_operand): Likewise.
5350 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5351 Alan Hayward <alan.hayward@arm.com>
5352 David Sherwood <david.sherwood@arm.com>
5354 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5357 2017-08-21 Richard Biener <rguenther@suse.de>
5360 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5361 at struct end conservatively when comparing common bases.
5363 2017-08-21 Richard Biener <rguenther@suse.de>
5365 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5366 (mem_ref_in_stmt): Remove.
5367 (determine_max_movement): Use ref index to get at the reference.
5368 (invariantness_dom_walker::before_dom_children): Deal with
5369 lim data already initialized.
5370 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5372 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5374 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5375 (TARGET_ISA_ROUND): Ditto.
5376 (TARGET_ROUND): Ditto.
5377 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5378 * config/i386/i386.md: Ditto.
5379 * config/i386/sse.md: Ditto.
5380 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5381 with OPTION_MASK_ISA_SSE4_1.
5383 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5386 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5387 __builtin_ia32_lzcnt_u16.
5389 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5392 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5393 (rs6000_set_current_function): Rewrite function to use it.
5395 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5398 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5399 and TYPE_WARN_IF_NOT_ALIGN.
5400 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5401 (handle_warn_if_not_align): New.
5402 (place_union_field): Call handle_warn_if_not_align.
5403 (place_field): Call handle_warn_if_not_align.
5404 Copy TYPE_WARN_IF_NOT_ALIGN.
5405 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5406 (layout_type): Likewise.
5407 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5409 (tree_decl_common): Add warn_if_not_align.
5410 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5411 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5412 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5413 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5414 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5415 * doc/extend.texi: Document warn_if_not_aligned attribute.
5416 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5418 2017-08-17 Martin Liska <mliska@suse.cz>
5421 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5422 (get_data_dependence): Use it as pointer type.
5423 (distribute_loop): Likewise.
5425 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5427 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5428 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5429 (p8_vmrgow_<mode>_direct): New define_insn.
5430 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5431 handle endianness for vmrgew and vmrgow permute patterns.
5433 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5435 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5436 * config/rs6000/rs6000-cpus.def: Remove comment.
5437 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5438 (POWERPC_MASKS): Likewise.
5439 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5440 use of TARGET_VSX_TIMODE.
5441 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5442 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5443 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5444 (rs6000_option_override_internal): Remove dead code.
5445 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5446 (rs6000_legitimize_reload_address): Likewise.
5447 (rs6000_legitimate_address_p): Likewise.
5448 (rs6000_opt_masks): Delete "vsx-timode".
5449 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5450 from function comment.
5451 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5452 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5453 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5455 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5456 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5457 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5460 (define_peephole2): Likewise.
5462 2017-08-17 Martin Sebor <msebor@redhat.com>
5465 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5466 past the end of an array.
5467 (test_pp_format): Add test cases.
5469 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5471 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5472 missing ECF_NOTHROW flags.
5474 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5477 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5478 operands residing in integer registers.
5479 (*vsx_le_perm_load_<mode>): Likewise.
5480 (*vsx_le_perm_store_<mode>): Likewise.
5481 (define_peephole2): Add peepholes to optimize the above.
5483 2017-08-17 Marek Polacek <polacek@redhat.com>
5486 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5487 to mimic what shorten_compare did. Change the return type to bool.
5488 (fold_cond_expr_with_comparison): Update call to
5489 operand_equal_for_comparison_p.
5490 (fold_ternary_loc): Likewise.
5492 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5494 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5496 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5498 2017-08-17 Richard Biener <rguenther@suse.de>
5500 * tree-ssa-structalias.c (solve_graph): When propagating
5501 to successors update the graphs succ edges and avoid duplicate work.
5503 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5506 * config/i386/i386.c (ix86_option_override_internal): Save target
5507 specific options after ix86_stack_protector_guard_reg was changed.
5509 2017-08-17 Richard Biener <rguenther@suse.de>
5511 PR tree-optimization/81827
5512 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5513 (new_var_info): Initialize it conservatively.
5514 (get_call_vi): Mark register vars.
5515 (new_scalar_tmp_constraint_exp): Likewise.
5516 (handle_rhs_call): Likewise.
5517 (handle_const_call): Likewise.
5518 (create_function_info_for): Likewise.
5519 (solve_constraints): Sort varinfos to separate register from
5520 non-register vars to pack points-to solution bitmaps during
5523 2017-08-17 Marek Polacek <polacek@redhat.com>
5525 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5527 2017-08-17 Richard Biener <rguenther@suse.de>
5529 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5530 to true when overflow is undefined and we saturated the result.
5532 2017-08-17 Alan Modra <amodra@gmail.com>
5535 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5536 Don't use store multiple if only one reg needs saving.
5537 (interesting_frame_related_regno): New function.
5538 (rs6000_frame_related): Don't emit frame info for regs that
5540 (rs6000_emit_epilogue): Likewise.
5542 2017-08-16 Nathan Sidwell <nathan@acm.org>
5544 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5545 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5546 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5547 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5548 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5549 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5550 (verify_type): Adjust for TYPE_BINFO move.
5551 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5552 process TYPE_BINFO directly.
5553 (hash_tree): Likewise.
5554 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5556 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5559 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5561 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5563 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5566 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5567 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5568 Add ix86_binary_operator_ok to insn constraint.
5569 (*iordi_1_bts): Ditto.
5570 (*xordi_1_btc): Ditto.
5571 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5572 Update corresponding peephole2 pattern.
5576 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5578 PR tree-optimization/81832
5579 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5580 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5582 2017-08-16 Marek Polacek <polacek@redhat.com>
5585 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5586 regarding size_zero_node.
5588 2017-08-16 Martin Liska <mliska@suse.cz>
5591 * config.gcc: Respect previously set extra_objs in case
5594 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5596 PR tree-optimization/81835
5597 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5598 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5599 not depend on the phi.
5601 2017-08-16 Alan Modra <amodra@gmail.com>
5603 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5606 2017-08-16 Alan Modra <amodra@gmail.com>
5608 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5609 (save_reg_p): ..into this. Update all callers.
5610 (first_reg_to_save): Simplify.
5612 2017-08-16 Alan Modra <amodra@gmail.com>
5614 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5617 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5621 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5622 constant count if that count is less than 32.
5624 2017-08-15 Nathan Sidwell <nathan@acm.org>
5626 * gcc.c (execute): Emit friendlier message if inferior is killed
5627 by an external cause.
5629 2017-08-15 Richard Biener <rguenther@suse.de>
5631 PR tree-optimization/81790
5632 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5633 CONSTRUCTORs from simplifying and VN.
5635 2017-08-14 Martin Sebor <msebor@redhat.com>
5637 * builtin-attrs.def: Add comments.
5639 2017-08-14 Martin Sebor <msebor@redhat.com>
5642 * doc/extend.texi (attribute nonstring): Document new attribute.
5644 2017-08-14 Martin Sebor <msebor@redhat.com>
5647 * tree-diagnostic.c (default_tree_printer): Handle %G.
5648 * gimple-pretty-print.h (percent_G_format): Declare new function.
5649 * gimple-pretty-print.c (percent_G_format): Define.
5650 * tree-pretty-print.c (percent_K_format): Add argument.
5652 2017-08-14 Martin Sebor <msebor@redhat.com>
5654 PR translation/79998
5655 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5656 Remove a stray space.
5658 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5661 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5662 (*iordi_1_bts): Ditto.
5663 (*xordi_1_btc): Ditto.
5665 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5668 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5669 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5671 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5672 quoted strings, and make more translator-friendly.
5673 (darwin_rs6000_override_options): Likewise.
5674 (rs6000_option_override_internal): Likewise.
5675 (rs6000_return_in_memory): Fix overlong line.
5676 (init_cmulative_args): Use quoted strings, and make more
5677 translator-friendly.
5678 (rs6000_pass_by_reference): Fix overlong line.
5679 (def_builtin): Use quoted strings.
5680 (altivec_expand_predicate_builtin): Use quoted strings, and make
5681 more translator-friendly.
5682 (htm_expand_builtin): Use quoted strings.
5683 (cpu_expand_builtin): Use quoted strings, and make more
5684 translator-friendly.
5685 (altivec_expand_builtin): Likewise.
5686 (paired_expand_predicate_builtin): Likewise.
5687 (rs6000_invalid_builtin): Likewise.
5688 (builtin_function_type): Use quoted strings.
5689 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5690 more translator-friendly.
5691 (rs6000_trampoline_init): Likewise.
5692 (rs6000_handle_altivec_attribute): Likewise.
5693 (rs6000_inner_target_options): Use quoted strings.
5694 (rs6000_disable_incompatible_switches): Likewise.
5695 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5696 strings, and make more translator-friendly.
5697 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5699 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5701 PR tree-optimization/81799
5702 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5703 cond_expr to simple gimple operand.
5705 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5708 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5710 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5714 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5715 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5716 TARGET_GASISR_PROLOGUES.
5717 * config/avr/avr.c (avr_option_override): Same.
5718 (avr_pass_pre_proep::execute): Same.
5720 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5723 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5724 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5726 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5728 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5729 define_insn_and_split. Split to a memory load from 0 in
5730 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5731 using PTR mode iterator.
5732 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5733 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5734 (*add_tp_<mode>): Redefine as define_insn_and_split.
5735 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5736 address space. Merge with *add_tp_x32 using PTR mode iterator.
5737 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5738 Split to an add with a memory load from 0 in
5739 DEFAULT_TLS_SEG_REG address space.
5741 2017-08-12 Andrew Pinski <apinski@cavium.com>
5743 * config/aarch64/aarch64-option-extensions.def (rdma):
5744 Fix feature string to what Linux prints out in /proc/cpuinfo.
5746 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5749 * dwarf2out.c (modified_type_die): For C typedef types that have
5750 an ultimate origin, process the ultimate origin instead of the
5752 (gen_typedef_die): Assert that input DECLs have no ultimate
5754 (gen_type_die_with_usage): For typedef variants that have an
5755 ultimate origin, just call gen_decl_die on the original DECL.
5756 (process_scope_var): Avoid creating DIEs for local typedefs and
5757 concrete static variables.
5759 2017-08-12 Alan Modra <amodra@gmail.com>
5763 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5764 match gnu-user.h startfile.
5765 (ENDFILE_LINUX_SPEC): Similarly.
5767 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5770 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5772 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5774 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5775 * config/aarch64/aarch64.md (mov<mode>): Change.
5776 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5777 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5778 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5780 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5782 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5783 for storage order barriers.
5785 2017-08-11 Martin Liska <mliska@suse.cz>
5788 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5789 variables of void type.
5791 2017-08-11 Martin Liska <mliska@suse.cz>
5793 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5794 TARGET_SUPPORTS_ALIASES.
5795 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5796 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5797 (optimize_weakref): Likewise.
5798 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5799 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5800 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5802 2017-08-11 Martin Liska <mliska@suse.cz>
5805 * config/i386/i386.c (make_resolver_func): Do complete
5806 refactoring of the function.
5808 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5811 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5812 * config/i386/i386.c (ix86_stack_protect_guard): Use
5813 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5814 * doc/invoke.texi (x86 Options): Document
5815 -mstack-protector-guard-symbol= option.
5817 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5819 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5820 * config/i386/i386.c (ix86_split_stack_guard): New function.
5821 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5822 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5823 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5824 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5825 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5826 (split_stack_space_check): Call ix86_split_stack_guard.
5828 2017-08-10 Martin Sebor <msebor@redhat.com>
5830 * print-tree.c (print_node): Print location using the established
5832 Replace spaces with colons.
5833 (debug_raw, debug): Ditto.
5835 2017-08-10 Martin Sebor <msebor@redhat.com>
5838 * pretty-print.c (pp_format): Correct the handling of %s precision.
5840 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5843 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5845 (ix86_finalize_stack_frame_flags): This. Also clear
5846 frame_pointer_needed if -fno-omit-frame-pointer is used without
5848 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5849 with ix86_finalize_stack_frame_flags.
5850 (ix86_expand_epilogue): Likewise.
5851 (ix86_expand_split_stack_prologue): Likewise.
5852 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5854 2017-08-10 Martin Liska <mliska@suse.cz>
5857 * c-attribs.c (handle_target_attribute):
5858 Report warning for an empty string argument of target attribute.
5860 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5863 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5865 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5866 or DECL_NONLOCAL labels.
5867 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5868 or DECL_NONLOCAL labels here.
5870 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5872 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5873 to indicate when early gimple folding has been disabled.
5874 (rs6000_gimple_fold_builtin): Add debug content.
5875 (rs6000_invalid_builtin): Fix whitespace.
5876 (rs6000_expand_builtin): Fix whitespace.
5877 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5879 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5882 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5883 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5885 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5887 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5888 (qdf24xx): Likewise.
5889 * config/aarch64/aarch64.md: Include falkor.md.
5890 * config/aarch64/falkor.md: New.
5892 2017-08-09 Marek Polacek <polacek@redhat.com>
5895 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5896 * diagnostic.c (emit_diagnostic): Add a comment.
5897 (emit_diagnostic_valist): New function.
5899 2017-08-09 Marek Polacek <polacek@redhat.com>
5902 * input.c (make_location): New overload.
5903 * input.h (make_location): Declare.
5905 2017-08-08 Alan Modra <amodra@gmail.com>
5906 H.J. Lu <hongjiu.lu@intel.com>
5909 * gcc.c (NO_PIE_SPEC): Delete.
5910 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5912 (LINK_PIE_SPEC): ..to here.
5913 (LINK_COMMAND_SPEC): Support -no-pie.
5914 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5915 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5917 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5918 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5919 (ENDFILE_CRTEND_SPEC): Similarly.
5921 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5924 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5925 (mstack-protector-guard-offset=): Ditto.
5926 * config/i386/i386.c (ix86_option_override): Handle
5927 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5929 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5930 ix86_stack_protect_guard_offset variables.
5931 (TARGET_STACK_PROTECT_GUARD): Always define.
5932 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5933 and -mstack-protector-guard-offset= options.
5935 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5937 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5938 boundary case for the last candidate.
5940 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5942 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5943 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5945 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5948 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5949 * config/aarch64/aarch64-builtins.c
5950 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5951 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5952 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5954 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5955 Andrew Pinski <pinskia@gmail.com>
5958 * internal-fn.def (XORSIGN): New.
5959 * optabs.def (xorsign_optab): New.
5960 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5961 (convert_expand_mult_copysign): New.
5962 (pass_optimize_widening_mul::execute): Call
5963 convert_expand_mult_copysign.
5965 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5967 PR tree-optimization/81354
5968 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5969 Insert on edges rather than explicitly creating landing pads.
5970 (analyze_candidates_and_replace): Commit edge inserts.
5972 2017-08-08 Richard Biener <rguenther@suse.de>
5975 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5976 (expand_simple_operations): Also look through ADDR_EXPRs with
5977 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5979 2017-08-08 Richard Biener <rguenther@suse.de>
5981 PR tree-optimization/81723
5982 * tree-vect-slp.c (struct bst_traits): New hash traits.
5983 (bst_fail): New global.
5984 (vect_build_slp_tree_2): New worker, split out from ...
5985 (vect_build_slp_tree): ... this now wrapping it with using
5986 bst_fail set to cache SLP tree build fails. Properly handle
5988 (vect_analyze_slp_instance): Allocate and free bst_fail.
5990 2017-08-08 Martin Liska <mliska@suse.cz>
5993 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5994 LABEL_DECLs that can be from a different function.
5996 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5998 PR tree-optimization/81744
5999 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6000 loop's number of iterations.
6002 2017-08-08 Martin Liska <mliska@suse.cz>
6004 * asan.c: Include header files.
6005 * attribs.c (build_decl_attribute_variant): New function moved
6007 (build_type_attribute_qual_variant): Likewise.
6008 (cmp_attrib_identifiers): Likewise.
6009 (simple_cst_list_equal): Likewise.
6010 (omp_declare_simd_clauses_equal): Likewise.
6011 (attribute_value_equal): Likewise.
6012 (comp_type_attributes): Likewise.
6013 (build_type_attribute_variant): Likewise.
6014 (lookup_ident_attribute): Likewise.
6015 (remove_attribute): Likewise.
6016 (merge_attributes): Likewise.
6017 (merge_type_attributes): Likewise.
6018 (merge_decl_attributes): Likewise.
6019 (merge_dllimport_decl_attributes): Likewise.
6020 (handle_dll_attribute): Likewise.
6021 (attribute_list_equal): Likewise.
6022 (attribute_list_contained): Likewise.
6023 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6024 (lookup_attribute_by_prefix): Likewise.
6025 * bb-reorder.c: Include header files.
6026 * builtins.c: Likewise.
6027 * calls.c: Likewise.
6028 * cfgexpand.c: Likewise.
6029 * cgraph.c: Likewise.
6030 * cgraphunit.c: Likewise.
6031 * convert.c: Likewise.
6032 * dwarf2out.c: Likewise.
6033 * final.c: Likewise.
6034 * fold-const.c: Likewise.
6035 * function.c: Likewise.
6036 * gimple-expr.c: Likewise.
6037 * gimple-fold.c: Likewise.
6038 * gimple-pretty-print.c: Likewise.
6039 * gimple.c: Likewise.
6040 * gimplify.c: Likewise.
6041 * hsa-common.c: Likewise.
6042 * hsa-gen.c: Likewise.
6043 * internal-fn.c: Likewise.
6044 * ipa-chkp.c: Likewise.
6045 * ipa-cp.c: Likewise.
6046 * ipa-devirt.c: Likewise.
6047 * ipa-fnsummary.c: Likewise.
6048 * ipa-inline.c: Likewise.
6049 * ipa-visibility.c: Likewise.
6051 * lto-cgraph.c: Likewise.
6052 * omp-expand.c: Likewise.
6053 * omp-general.c: Likewise.
6054 * omp-low.c: Likewise.
6055 * omp-offload.c: Likewise.
6056 * omp-simd-clone.c: Likewise.
6057 * opts-global.c: Likewise.
6058 * passes.c: Likewise.
6059 * predict.c: Likewise.
6060 * sancov.c: Likewise.
6061 * sanopt.c: Likewise.
6062 * symtab.c: Likewise.
6063 * toplev.c: Likewise.
6064 * trans-mem.c: Likewise.
6065 * tree-chkp.c: Likewise.
6066 * tree-eh.c: Likewise.
6067 * tree-into-ssa.c: Likewise.
6068 * tree-object-size.c: Likewise.
6069 * tree-parloops.c: Likewise.
6070 * tree-profile.c: Likewise.
6071 * tree-ssa-ccp.c: Likewise.
6072 * tree-ssa-live.c: Likewise.
6073 * tree-ssa-loop.c: Likewise.
6074 * tree-ssa-sccvn.c: Likewise.
6075 * tree-ssa-structalias.c: Likewise.
6076 * tree-ssa.c: Likewise.
6077 * tree-streamer-in.c: Likewise.
6078 * tree-vectorizer.c: Likewise.
6079 * tree-vrp.c: Likewise.
6081 * ubsan.c: Likewise.
6082 * varasm.c: Likewise.
6083 * varpool.c: Likewise.
6084 * tree.c: Remove functions moved to attribs.[ch].
6086 * config/aarch64/aarch64.c: Add attrs.h header file.
6087 * config/alpha/alpha.c: Likewise.
6088 * config/arc/arc.c: Likewise.
6089 * config/arm/arm.c: Likewise.
6090 * config/avr/avr.c: Likewise.
6091 * config/bfin/bfin.c: Likewise.
6092 * config/c6x/c6x.c: Likewise.
6093 * config/cr16/cr16.c: Likewise.
6094 * config/cris/cris.c: Likewise.
6095 * config/darwin.c: Likewise.
6096 * config/epiphany/epiphany.c: Likewise.
6097 * config/fr30/fr30.c: Likewise.
6098 * config/frv/frv.c: Likewise.
6099 * config/ft32/ft32.c: Likewise.
6100 * config/h8300/h8300.c: Likewise.
6101 * config/i386/winnt.c: Likewise.
6102 * config/ia64/ia64.c: Likewise.
6103 * config/iq2000/iq2000.c: Likewise.
6104 * config/lm32/lm32.c: Likewise.
6105 * config/m32c/m32c.c: Likewise.
6106 * config/m32r/m32r.c: Likewise.
6107 * config/m68k/m68k.c: Likewise.
6108 * config/mcore/mcore.c: Likewise.
6109 * config/microblaze/microblaze.c: Likewise.
6110 * config/mips/mips.c: Likewise.
6111 * config/mmix/mmix.c: Likewise.
6112 * config/mn10300/mn10300.c: Likewise.
6113 * config/moxie/moxie.c: Likewise.
6114 * config/msp430/msp430.c: Likewise.
6115 * config/nds32/nds32-isr.c: Likewise.
6116 * config/nds32/nds32.c: Likewise.
6117 * config/nios2/nios2.c: Likewise.
6118 * config/nvptx/nvptx.c: Likewise.
6119 * config/pa/pa.c: Likewise.
6120 * config/pdp11/pdp11.c: Likewise.
6121 * config/powerpcspe/powerpcspe.c: Likewise.
6122 * config/riscv/riscv.c: Likewise.
6123 * config/rl78/rl78.c: Likewise.
6124 * config/rx/rx.c: Likewise.
6125 * config/s390/s390.c: Likewise.
6126 * config/sh/sh.c: Likewise.
6127 * config/sol2.c: Likewise.
6128 * config/sparc/sparc.c: Likewise.
6129 * config/spu/spu.c: Likewise.
6130 * config/stormy16/stormy16.c: Likewise.
6131 * config/tilegx/tilegx.c: Likewise.
6132 * config/tilepro/tilepro.c: Likewise.
6133 * config/v850/v850.c: Likewise.
6134 * config/vax/vax.c: Likewise.
6135 * config/visium/visium.c: Likewise.
6136 * config/xtensa/xtensa.c: Likewise.
6138 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6141 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6142 constraints since the -mupper-regs-* switches have been
6144 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6145 into a vector from a double word element that was extracted from
6146 another vector, and eliminate extra XXPERMDI instructions.
6147 (vsx_concat_<mode>_2): Likewise.
6148 (vsx_concat_<mode>_3): Likewise.
6149 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6150 concat to allow optimizing inserts from previous extracts.
6152 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6154 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6155 memory reference to a SSP offset in TLS address space.
6156 (ix86_print_operand) <case '@'>: Remove.
6157 (ix86_print_operand_punct_valid_p): Remove '@' code.
6158 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6160 (stack_tls_protect_set_<mode>): Remove.
6161 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6162 (stack_tls_protect_test_<mode>): Remove.
6163 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6165 2017-08-07 Olivier Hainque <hainque@adacore.com>
6168 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6170 2017-08-07 Douglas Rupp <rupp@adacore.com>
6172 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6173 variable was referenced as multidir in command.
6175 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6178 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6181 2017-08-07 Martin Liska <mliska@suse.cz>
6183 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6184 * config/rl78/rl78.c: Add include of attribs.h.
6185 * config/sh/sh.c: Likewise.
6186 * config/v850/v850.c: Likewise.
6188 2017-08-07 Tom de Vries <tom@codesourcery.com>
6191 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6193 2017-08-07 Martin Liska <mliska@suse.cz>
6195 * config/mips/mips.c: Include attribs.h.
6197 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6200 * doc/invoke.texi: Document change in behvaior for -Ofast for
6203 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6205 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6207 (aarch64_pop_regs): Likewise.
6208 (aarch64_gen_load_pair): Likewise.
6209 (aarch64_save_callee_saves): Likewise.
6210 (aarch64_restore_callee_saves): Likewise.
6212 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6214 * config/i386/i386.c: Revert the last change.
6216 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6219 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6221 (ix86_finalize_stack_frame_flags): This. Also clear
6222 frame_pointer_needed if -fno-omit-frame-pointer is used without
6224 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6225 with ix86_finalize_stack_frame_flags.
6226 (ix86_expand_epilogue): Likewise.
6227 (ix86_expand_split_stack_prologue): Likewise.
6229 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6232 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6233 to P_AES for Westmere.
6235 2017-08-07 Jonathan Yong <10walls@gmail.com>
6237 * config/i386/mingw.opt (fset-stack-executable): Removed.
6238 * config/i386/cygming.opt (fset-stack-executable): Moved
6240 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6242 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6244 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6246 2017-08-07 Marek Polacek <polacek@redhat.com>
6249 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6251 2017-08-07 Martin Liska <mliska@suse.cz>
6253 * attribs.h (canonicalize_attr_name): New function.
6254 (cmp_attribs): Move from c-format.c and adjusted.
6255 (is_attribute_p): Moved from tree.h.
6256 * tree-inline.c: Add new includes.
6257 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6258 (private_is_attribute_p): Remove.
6259 (private_lookup_attribute): Likewise.
6260 (private_lookup_attribute_by_prefix): Simplify.
6261 (remove_attribute): Use is_attribute_p.
6262 * tree.h: Remove removed declarations.
6264 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6267 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6268 instead of computing it in the function. Formatting fix.
6269 (expand_case): Don't rely on default_edge being the first edge,
6270 clear it if removing it, pass default_edge to
6271 emit_case_dispatch_table.
6272 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6275 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6277 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6278 insn in the function, emit NOP after the insn.
6280 2017-08-06 Tom de Vries <tom@codesourcery.com>
6282 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6285 2017-08-06 Tom de Vries <tom@codesourcery.com>
6287 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6290 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6292 PR tree-optimization/57371
6293 * match.pd: New pattern.
6295 2017-08-04 Marek Polacek <polacek@redhat.com>
6298 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6299 perform the computation in offset_int.
6301 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6303 PR tree-optimization/81136
6304 * tree-vectorizer.h: Include tree-hash-traits.h.
6305 (vec_base_alignments): New typedef.
6306 (vec_info): Add a base_alignments field.
6307 (vect_record_base_alignments): Declare.
6308 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6310 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6311 (create_data_ref): Add an is_conditional_in_stmt argument.
6312 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6313 the is_conditional_in_stmt field.
6314 (data_ref_loc): Add an is_conditional_in_stmt field.
6315 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6316 (find_data_references_in_stmt): Update call to create_data_ref.
6317 (graphite_find_data_references_in_stmt): Likewise.
6318 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6319 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6320 (vect_record_base_alignment): New function.
6321 (vect_record_base_alignments): Likewise.
6322 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6323 for nested statements even if we fail to compute a misalignment.
6324 Use pooled base alignments for unconditional references.
6325 (vect_find_same_alignment_drs): Compare base addresses instead
6327 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6328 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6330 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6332 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6333 Add an explicit name for the enum. Use auto_vec for slp_instances
6335 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6337 (_bb_vec_info): Add a constructor and destructor.
6338 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6339 (destroy_loop_vec_info): Delete.
6340 (vect_destroy_datarefs): Likewise.
6341 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6342 (vec_info::vec_info): New function.
6343 (vec_info::~vec_info): Likewise.
6344 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6345 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6346 destroy_loop_vec_info.
6347 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6348 (_loop_vec_info::_loop_vec_info): ...this.
6349 (destroy_loop_vec_info): Replace with...
6350 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6351 the stmt_vec_infos. Leave handling of vec_info information to its
6352 destructor. Remove explicit vector releases.
6353 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6354 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6355 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6356 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6357 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6358 (destroy_bb_vec_info): Replace with...
6359 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6360 information to its destructor.
6361 (vect_slp_analyze_bb_1): Use new and delete instead of
6362 new_bb_vec_info and destroy_bb_vec_info.
6363 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6366 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6368 * tree-data-ref.h (subscript): Add access_fn field.
6369 (data_dependence_relation): Add could_be_independent_p.
6370 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6371 (same_access_functions): Move to tree-data-ref.c.
6372 * tree-data-ref.c (ref_contains_union_access_p): New function.
6373 (access_fn_component_p): Likewise.
6374 (access_fn_components_comparable_p): Likewise.
6375 (dr_analyze_indices): Add a reference to access_fn_component_p.
6376 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6378 (constant_access_functions): Likewise.
6379 (add_other_self_distances): Likewise.
6380 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6381 (initialize_data_dependence_relation): Use XCNEW and remove
6382 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6383 of access functions that have the same type. Allow the
6384 subsequence to end with different bases in some circumstances.
6385 Record the chosen access functions in SUB_ACCESS_FN.
6386 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6387 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6388 (subscript_dependence_tester_1): Likewise dra and drb.
6389 (build_classic_dist_vector): Update calls accordingly.
6390 (subscript_dependence_tester): Likewise.
6391 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6392 DDR_COULD_BE_INDEPENDENT_P.
6393 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6394 comp_alias_ddrs instead of may_alias_ddrs.
6395 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6397 (vect_analyze_data_ref_dependence): Use it if
6398 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6399 distance vectors if that fails.
6400 (dependence_distance_ge_vf): New function.
6401 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6402 LOOP_VINFO_MAY_ALIAS_DDRS.
6404 2017-08-04 Richard Biener <rguenther@suse.de>
6407 * fold-const.c (fold_binary_loc): Properly restrict
6408 minus_var0 && minus_var1 case when associating undefined overflow
6411 2017-08-04 Tom de Vries <tom@codesourcery.com>
6413 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6415 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6417 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6418 Don't start diagnostic messages with a capital letter.
6419 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6421 (rs6000_invalid_builtin): Likewise.
6422 (rs6000_trampoline_init): Likewise.
6424 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6427 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6428 after setting changeable df flags.
6430 2017-08-03 Richard Biener <rguenther@suse.de>
6432 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6433 up if the use is in USE - X.
6435 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6437 * toplev.c (dumpfile.h): New include.
6438 (internal_error_reentered): New static function. Use it...
6439 (internal_error_function): ...here to handle reentered internal_error.
6441 2017-08-03 Richard Biener <rguenther@suse.de>
6444 * fold-const.c (split_tree): Add minus_var and minus_con
6445 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6446 here but always use minus_*.
6447 (associate_trees): Assert we never associate with MINUS_EXPR
6448 and NULL first operand. Do not recurse for PLUS_EXPR operands
6449 when associating as MINUS_EXPR either.
6450 (fold_binary_loc): Track minus_var and minus_con.
6452 2017-08-03 Tom de Vries <tom@codesourcery.com>
6455 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6456 ACCEL_COMPILER, apply finish_options on
6457 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6459 2017-08-03 Tom de Vries <tom@codesourcery.com>
6462 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6463 function_entry_patch_area_size > 0.
6465 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6468 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6469 instead of 10??LU, perform unit multiplication in wide_int,
6470 don't change alloc_object_size_limit if the limit is larger
6473 PR tree-optimization/81655
6474 PR tree-optimization/81588
6475 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6476 the case when ranges[i].low and high are 1 for unsigned type with
6480 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6481 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6483 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6485 * tree-vrp.h: Add include guard.
6487 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6490 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6491 (ud2): New insn pattern.
6492 * config/i386/i386.c (ix86_expand_epilogue):
6493 For naked functions, generate ud2 instead of trap insn.
6495 2017-08-02 Marek Polacek <polacek@redhat.com>
6498 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6500 2017-08-02 Tom de Vries <tom@codesourcery.com>
6501 Cesar Philippidis <cesar@codesourcery.com>
6503 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6504 Add missing edge probabilities.
6506 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6508 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6511 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6514 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6515 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6516 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6517 prologue_seq sequences - if any.
6519 2017-08-02 Richard Biener <rguenther@suse.de>
6521 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6522 via vectors if supported, integer extracts via punning if supported
6523 or otherwise vector extracts.
6525 2017-08-02 Richard Biener <rguenther@suse.de>
6527 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6529 (bitmap_insert_into_set): ... this.
6531 2017-08-02 Richard Biener <rguenther@suse.de>
6533 PR tree-optimization/81633
6535 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6537 PR tree-optimization/71752
6538 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6540 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6542 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6543 (machine_function::call_ms2sysv_pad_out): Remove field.
6544 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6545 (ix86_compute_frame_layout): Likewise.
6547 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6550 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6551 attribute with interrupt attribute.
6553 2017-08-01 Andrew Pinski <apinski@cavium.com>
6555 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6556 BIT_INSERT_EXPR's operand 1
6557 to see if the types precision matches.
6559 2017-08-01 Martin Liska <mliska@suse.cz>
6562 * builtins.c (expand_builtin_memcpy_args): Remove.
6563 (expand_builtin_memcpy): Call newly added function
6564 expand_builtin_memory_copy_args.
6565 (expand_builtin_memcpy_with_bounds): Likewise.
6566 (expand_builtin_mempcpy): Remove last argument.
6567 (expand_builtin_mempcpy_with_bounds): Likewise.
6568 (expand_builtin_memory_copy_args): New function created from
6569 expand_builtin_mempcpy_args with small modifications.
6570 (expand_builtin_mempcpy_args): Remove.
6571 (expand_builtin_stpcpy): Remove unused argument.
6572 (expand_builtin): Likewise.
6573 (expand_builtin_with_bounds): Likewise.
6575 2017-08-01 Martin Liska <mliska@suse.cz>
6578 Make mempcpy more optimal (PR middle-end/70140).
6580 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6583 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6584 __builtin_vec_cmpne verify both arguments are compatible vectors
6585 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6586 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6587 move computation of aligned to after checking the argument types.
6591 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6592 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6595 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6596 Doug Rupp <rupp@adacore.com>
6597 Olivier Hainque <hainque@adacore.com>
6599 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6600 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6602 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6603 for TARGET_OS_CPP_BUILTIN.
6604 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6605 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6607 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6608 passing required abi options to the assembler for EABI configurations.
6609 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6610 of .text.hot and .text.unlikely sections for kernel modules when
6611 using ARM style exceptions.
6612 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6613 options. Add EXTRA_CC1_SPEC.
6614 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6616 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6618 (ARM_TARGET2_DWARF_FORMAT): Define.
6619 * config/arm/t-vxworks: Adjust multilib control to removal of the
6620 Diab command line options.
6622 2017-08-01 Martin Liska <mliska@suse.cz>
6624 PR gcov-profile/81561
6625 * gcov.c (unblock): Make unblocking safe as we need to preserve
6626 index correspondence of blocks and block_lists.
6628 2017-08-01 Richard Biener <rguenther@suse.de>
6630 PR tree-optimization/81181
6631 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6632 (compute_antic): ... end of iteration here.
6634 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6636 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6637 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6638 (ftree-slp-vectorize): Likewise.
6639 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6640 can no longer be set independent of flag_tree_loop_vectorize.
6641 * omp-general.c (emp_max_vf): Likewise.
6642 * opts.c (enable_fdo_optimizations): Remove references to
6643 flag_tree_vectorize, these are now implicit.
6644 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6645 and leave it for the options machinery.
6647 2017-08-01 Martin Liska <mliska@suse.cz>
6650 * builtins.c (expand_builtin_memcpy_args): Remove.
6651 (expand_builtin_memcpy): Call newly added function
6652 expand_builtin_memory_copy_args.
6653 (expand_builtin_memcpy_with_bounds): Likewise.
6654 (expand_builtin_mempcpy): Remove last argument.
6655 (expand_builtin_mempcpy_with_bounds): Likewise.
6656 (expand_builtin_memory_copy_args): New function created from
6657 expand_builtin_mempcpy_args with small modifications.
6658 (expand_builtin_mempcpy_args): Remove.
6659 (expand_builtin_stpcpy): Remove unused argument.
6660 (expand_builtin): Likewise.
6661 (expand_builtin_with_bounds): Likewise.
6663 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6666 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6667 print "ds:" only for immediates in generic address space.
6669 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6672 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6673 (ix86_function_ok_for_sibcall): Return false for naked functions.
6675 2017-08-01 Richard Biener <rguenther@suse.de>
6677 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6678 (compute_antic): Seed worklist with exit block predecessors.
6679 * cfganal.c (dfs_find_deadend): For a cycle return the source
6680 of the edge closing it.
6682 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6684 * config/aarch64/aarch64.c
6685 (aarch64_can_const_movi_rtx_p): Move 0 check.
6687 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6689 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6690 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6692 * match.pd: Ditto in address comparison pattern.
6694 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6696 PR tree-optimization/81627
6697 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6698 closed ssa form for store-store chain.
6700 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6702 PR tree-optimization/81620
6703 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6704 for store-store chain.
6706 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6708 PR tree-optimization/81588
6709 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6710 ranges[i].in_p, invert comparison code ccode. For >/>=,
6711 swap rhs1 and rhs2 and comparison code unconditionally,
6712 for </<= don't do that. Don't swap rhs1/rhs2 again if
6713 ranges[i].in_p, instead invert comparison code ccode if
6714 opcode or oe->rank is BIT_IOR_EXPR.
6717 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6718 a direct optab to conversion optab.
6719 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6720 with GET_MODE_INNER as last argument instead of optab_handler.
6721 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6722 vector extraction if possible and optab is available.
6723 * expr.c (store_constructor): Use convert_optab_handler instead
6724 of optab_handler. Use vector initialization from smaller
6725 vectors if possible and optab is available.
6726 * tree-vect-stmts.c (vectorizable_load): Likewise.
6727 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6729 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6730 of vec_init from half-sized vectors with the same element mode.
6731 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6732 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6733 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6734 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6735 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6736 after mode in gen_vec_extract* calls.
6737 (vec_extract<mode>): Renamed to ...
6738 (vec_extract<mode><ssescalarmodelower>): ... this.
6739 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6740 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6741 element mode after mode in gen_vec_init* calls.
6742 (VEC_INIT_HALF_MODE): New mode iterator.
6743 (vec_init<mode>): Renamed to ...
6744 (vec_init<mode><ssescalarmodelower>): ... this.
6745 (vec_init<mode><ssehalfvecmodelower>): New expander.
6746 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6747 (vec_extractv2sfsf): ... this.
6748 (vec_initv2sf): Renamed to ...
6749 (vec_initv2sfsf): ... this.
6750 (vec_extractv2si): Renamed to ...
6751 (vec_extractv2sisi): ... this.
6752 (vec_initv2si): Renamed to ...
6753 (vec_initv2sisi): ... this.
6754 (vec_extractv4hi): Renamed to ...
6755 (vec_extractv4hihi): ... this.
6756 (vec_initv4hi): Renamed to ...
6757 (vec_initv4hihi): ... this.
6758 (vec_extractv8qi): Renamed to ...
6759 (vec_extractv8qiqi): ... this.
6760 (vec_initv8qi): Renamed to ...
6761 (vec_initv8qiqi): ... this.
6762 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6763 (vec_init<mode>): Renamed to ...
6764 (vec_init<mode><VEC_base_l>): ... this.
6765 (vec_extract<mode>): Renamed to ...
6766 (vec_extract<mode><VEC_base_l>): ... this.
6767 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6768 (vec_initv2sfsf): ... this.
6769 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6770 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6771 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6772 element mode after mode in gen_vec_init* calls.
6773 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6774 (vec_init<mode><Vel>): ... this.
6775 (vec_extract<mode>): Renamed to ...
6776 (vec_extract<mode><Vel>): ... this.
6777 * config/aarch64/iterators.md (Vel): New mode attribute.
6778 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6779 Add element mode after mode in gen_vec_extract* calls.
6780 * config/s390/vector.md (non_vec_l): New mode attribute.
6781 (vec_extract<mode>): Renamed to ...
6782 (vec_extract<mode><non_vec_l>): ... this.
6783 (vec_init<mode>): Renamed to ...
6784 (vec_init<mode><non_vec_l>): ... this.
6785 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6786 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6788 * config/arm/iterators.md (V_elem_l): New mode attribute.
6789 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6790 (vec_extract<mode><V_elem_l>): ... this.
6791 (vec_extractv2di): Renamed to ...
6792 (vec_extractv2didi): ... this.
6793 (vec_init<mode>): Renamed to ...
6794 (vec_init<mode><V_elem_l>): ... this.
6795 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6796 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6797 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6798 Add element mode after gen_vec_extract* calls.
6799 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6800 (vec_init<mode><unitmode>): ... this.
6801 (vec_extract<mode>): Renamed to ...
6802 (vec_extract<mode><unitmode>): ... this.
6803 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6804 (vec_init<mode><unitmode>): ... this.
6805 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6806 (vec_initv2sfsf): ... this.
6807 (vec_extractv2sf): Renamed to ...
6808 (vec_extractv2sfsf): ... this.
6809 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6810 Add element mode after gen_vec_extract* calls.
6811 * config/mips/mips.md (unitmode): New mode iterator.
6812 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6813 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6814 * config/spu/spu.md (inner_l): New mode attribute.
6815 (vec_init<mode>): Renamed to ...
6816 (vec_init<mode><inner_l>): ... this.
6817 (vec_extract<mode>): Renamed to ...
6818 (vec_extract<mode><inner_l>): ... this.
6819 * config/sparc/sparc.md (veltmode): New mode iterator.
6820 (vec_init<VMALL:mode>): Renamed to ...
6821 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6822 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6823 (vec_initv2sisi): ... this.
6824 (vec_initv2sf): Renamed to ...
6825 (vec_initv2sfsf): ... this.
6826 (vec_extractv2sf): Renamed to ...
6827 (vec_extractv2sfsf): ... this.
6828 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6829 (vec_init<mode>): Renamed to ...
6830 (vec_init<mode><VEC_base_l>): ... this.
6831 (vec_extract<mode>): Renamed to ...
6832 (vec_extract<mode><VEC_base_l>): ... this.
6833 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6834 (vec_initv2sfsf): ... this.
6835 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6836 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6837 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6838 gen_vec_init* calls.
6840 2017-08-01 Richard Biener <rguenther@suse.de>
6842 PR tree-optimization/81297
6843 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6844 TREE_OVERFLOW from INTEGER_CSTs.
6846 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6848 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6850 2017-07-31 Carl Love <cel@us.ibm.com>
6852 * config/rs6000/rs6000-c: Add support for built-in functions
6853 vector signed char vec_xl_be (signed long long, signed char *);
6854 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6855 vector signed int vec_xl_be (signed long long, signed int *);
6856 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6857 vector signed long long vec_xl_be (signed long long, signed long long *);
6858 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6859 vector signed short vec_xl_be (signed long long, signed short *);
6860 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6861 vector double vec_xl_be (signed long long, double *);
6862 vector float vec_xl_be (signed long long, float *);
6863 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6864 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6865 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6867 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6868 (altivec_expand_builtin): Add switch statement to call
6869 altivec_expand_xl_be for each builtin.
6870 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6871 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6872 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6873 __builtin_vsx_le_be_v16qi.
6874 * doc/extend.texi: Update the built-in documentation file for the
6875 new built-in functions.
6877 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6880 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6882 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6884 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6886 * config.gcc: Add z14.
6887 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6888 CPU model numbers for z13s and z14.
6889 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6891 * config/s390/s390-opts.h (enum processor_type): Rename
6892 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6893 * config/s390/s390.c (processor_table): Add field for CPU name to
6894 be passed to Binutils.
6895 (s390_asm_output_machine_for_arch): Use the new field in
6896 processor_table for Binutils.
6897 (s390_expand_builtin): Replace arch12 with z14.
6898 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6899 (s390_get_sched_attrmask): Likewise.
6900 (s390_get_unit_mask): Likewise.
6901 * config/s390/s390.opt: Add z14 to processor_type enum.
6903 2017-07-31 Martin Jambor <mjambor@suse.cz>
6906 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6907 regardless of optimization level.
6909 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6910 Martin Liska <mliska@suse.cz>
6912 * predict.def: Remove old comment and adjust probability.
6913 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6916 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6919 * config/i386/i386.c (ix86_function_naked): New function.
6920 (ix86_can_use_return_insn_p): Return false for naked functions.
6921 (ix86_expand_prologue): Skip prologue for naked functions.
6922 (ix86_expand_epilogue): Skip epilogue for naked functions
6923 and emit trap instruction.
6924 (ix86_warn_func_return): New function.
6925 (ix86_attribute_table): Add "naked" attribute specification.
6926 (TARGET_WARN_FUNC_RETURN): Define.
6927 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6929 2017-07-31 Martin Liska <mliska@suse.cz>
6931 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6932 (dump_gimple_bb_header): Always dump BB info.
6933 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6935 2017-07-31 Martin Liska <mliska@suse.cz>
6938 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6939 also with current_function_decl non-null equality.
6941 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6944 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6945 change type to the element type, instead add eltype variable and
6946 use it where we are interested in the element type.
6948 PR tree-optimization/81603
6949 * ipa-polymorphic-call.c
6950 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6951 offset arithmetic in offset_int, bail out if the resulting bit offset
6952 doesn't fit into shwi.
6954 2017-07-31 Martin Liska <mliska@suse.cz>
6956 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6957 (gimplify_save_expr): Fix comment.
6959 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6962 * config/i386/i386.c (ix86_function_arg): Update arguments for
6964 (ix86_compute_frame_layout): Set the initial stack offset to
6965 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6966 INCOMING_FRAME_SP_OFFSET.
6967 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6968 stack before exception handler returns.
6969 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6970 the 'ERROR_CODE' for exception handler.
6972 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6974 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6975 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6976 (ASM_OUTPUT_REG_POP): Ditto.
6977 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6978 instead of asm_fprintf to output pure string.
6980 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6982 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6983 to imported_module_or_decl hook.
6984 (debug_nothing_tree_tree_tree_bool): Remove.
6985 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6986 * debug.c (do_nothing_debug_hooks): Use
6987 debug_nothing_tree_tree_tree_bool_bool instead of
6988 debug_nothing_tree_tree_tree_bool.
6989 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6990 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6991 * sdbout.c (sdb_debug_hooks): Likewise.
6992 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6993 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6995 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6996 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6997 attribute, don't add anything.
6999 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7001 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7002 (fold_build2_stat_loc): Likewise.
7003 (fold_build3_stat_loc): Likewise.
7004 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7005 (fold_build1_loc): Remove macro.
7006 (fold_build2_loc): Likewise.
7007 (fold_build3_loc): Likewise.
7009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7011 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7012 (gimple_build_debug_bind_source_stat): Likewise.
7013 * gimple.h (gimple_build_debug_bind): Remove macro.
7014 (gimple_build_debug_bind_source): Likewise.
7016 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7018 * bitmap.c (bitmap_alloc): Adjust.
7019 (bitmap_gc_alloc): Likewise.
7020 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7022 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7024 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7025 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7026 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7027 (bitmap_gc_alloc_stat): Likewise.
7028 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7030 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7032 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7033 * rtl.h (shallow_copy_rtx): Remove macro.
7035 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7037 * emit-rtl.c (gen_raw_REG): Adjust.
7038 * gengenrtl.c (gendef): Likewise.
7039 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7040 * rtl.h (rtx_alloc): Remove macro.
7042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7044 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7045 (build_tree_list_stat): Likewise.
7046 * tree.h (build_tree_list): Remove macro.
7047 (build_tree_list_vec): Likewise.
7049 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7051 * tree.c (make_vector_stat): Remove _stat from name.
7052 (build_vector_stat): Likewise.
7053 * tree.h (make_vector_stat): Remove macro.
7054 (build_vector_stat): Likewise.
7056 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7058 * tree.h (build_var_debug_value): Remove prototype.
7060 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7062 * tree.c (tree_cons_stat): Remove _stat from name.
7063 * tree.h (tree_cons): Remove macro.
7065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7067 * tree.c (build_vl_exp_stat): Remove _stat from name.
7068 * tree.h (build_vl_exp): Remove macro.
7070 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7072 * tree.c (build_decl_stat): Remove _stat from name.
7073 * tree.h (build_decl): Remove macro.
7075 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7077 * gimple.c (gimple_build_with_ops_stat): Adjust.
7078 (gimple_alloc_stat): Remove _stat from name.
7079 * gimple.h (gimple_alloc): Remove macro.
7081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7083 * tree.c (make_tree_vec_stat): Remove _stat from name.
7084 (grow_tree_vec_stat): Likewise.
7085 * tree.h (make_tree_vec_stat): Adjust prototype.
7086 (grow_tree_vec_stat): Likewise.
7087 (make_tree_vec): Remove macro.
7088 (grow_tree_vec): Likewise.
7090 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7092 * fold-const.c (fold_build1_stat_loc): Adjust.
7093 (fold_build2_stat_loc): Likewise.
7094 (fold_build3_stat_loc): Likewise.
7095 * tree.c (build0_stat): Remove _stat from name.
7096 (build1_stat): Likewise.
7097 (build2_stat): Likewise.
7098 (build3_stat): Likewise.
7099 (build4_stat): Likewise.
7100 (build5_stat): Likewise.
7101 * tree.h (build1_loc): Remove macro, and rename _stat function
7103 (build2_loc): Likewise.
7104 (build3_loc): Likewise.
7105 (build4_loc): Likewise.
7106 (build5_loc): Likewise.
7108 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7110 * tree.c (make_int_cst_stat): Remove _stat from name.
7111 * tree.h (make_int_cst_stat): Adjust prototype.
7112 (make_int_cst): Remove macro.
7114 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7116 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7117 * tree.h (make_tree_binfo_stat): Adjust prototype.
7118 (make_tree_binfo): Remove.
7120 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7122 * tree.c (copy_node_stat): Rename to copy_node.
7123 (build_distinct_type_copy): Adjust.
7124 * tree.h (copy_node_stat): Adjust prototype.
7125 (copy_node): Remove macro.
7127 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7129 * tree.c (make_node_stat): rename to make_node.
7130 (build_tree_list_stat): Adjust.
7131 (build0_stat): Likewise.
7132 (build2_stat): Likewise.
7133 (build3_stat): Likewise.
7134 (build4_stat): Likewise.
7135 (build5_stat): Likewise.
7136 (build_decl_stat): Likewise.
7137 * tree.h (make_node_stat): Adjust prototype.
7138 (make_node): remove macro.
7140 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7142 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7143 (PPC_FEATURE2_SCV): Likewise.
7144 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7146 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7148 * config/aarch64/aarch64.c
7149 (aarch64_internal_mov_immediate): Add new special pattern.
7150 * config/aarch64/aarch64.md (*movdi_aarch64):
7151 Add reg/32bit const mov case.
7153 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7154 Richard Sandiford <richard.sandiford@linaro.org>
7156 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7157 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7158 Add integer and movi cases.
7159 (movi-split-hf-df-sf split, fp16): New.
7160 (enabled): Added TARGET_FP_F16INST.
7161 * config/aarch64/iterators.md (GPF_HF): New.
7162 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7164 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7166 * config/aarch64/aarch64.c
7167 (aarch64_simd_container_mode): Add prototype.
7168 (aarch64_expand_mov_immediate): Add HI support.
7169 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7170 (aarch64_can_const_movi_rtx_p): New.
7171 (aarch64_preferred_reload_class):
7172 Remove restrictions of using FP registers for certain SIMD operations.
7173 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7174 (aarch64_valid_floating_const): Add integer move validation.
7175 (aarch64_simd_imm_scalar_p): Remove.
7176 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7177 (aarch64_legitimate_constant_p): Expand list of supported cases.
7178 * config/aarch64/aarch64-protos.h
7179 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7180 (aarch64_reinterpret_float_as_int): New.
7181 (aarch64_simd_imm_scalar_p): Remove.
7182 * config/aarch64/constraints.md (Uvi): New.
7183 (Dd): Split into Ds and new Dd.
7184 * config/aarch64/aarch64.md (*movsi_aarch64):
7186 (*movdi_aarch64): Add SIMD mov case.
7188 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7190 * tree-predcom.c: (struct chain): Handle store-store chain in which
7191 stores for elimination only store loop invariant values.
7192 (execute_pred_commoning_chain): Ditto.
7193 (prepare_initializers_chain_store_elim): Ditto.
7194 (prepare_finalizers): Ditto.
7195 (is_inv_store_elimination_chain): New function.
7196 (initialize_root_vars_store_elim_1): New function.
7198 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7200 * tree-predcom.c: Revise general description of the pass.
7201 (enum chain_type): New enum type for store elimination.
7202 (struct chain): New field supporting store elimination.
7203 (struct component): Ditto.
7204 (dump_chain): Dump store-stores chain.
7205 (release_chain): Release resources.
7206 (split_data_refs_to_components): Compute and create component
7207 contains only stores for elimination.
7208 (get_chain_last_ref_at): New function.
7209 (make_invariant_chain): Initialization.
7210 (make_rooted_chain): Specify chain type in parameter and record it.
7211 (add_looparound_copies): Skip for store-stores chain.
7212 (determine_roots_comp): Compute type of chain and pass it to
7214 (initialize_root_vars_store_elim_2): New function.
7215 (finalize_eliminated_stores): New function.
7216 (remove_stmt): Handle store for elimination.
7217 (execute_pred_commoning_chain): Execute predictive commoning on
7219 (determine_unroll_factor): Skip unroll for store-stores chain.
7220 (prepare_initializers_chain_store_elim): New function.
7221 (prepare_initializers_chain): Hanlde store-store chain.
7222 (prepare_finalizers_chain, prepare_finalizers): New function.
7223 (tree_predictive_commoning_loop): Return integer value indicating
7224 if loop is unrolled or lcssa form is corrupted.
7225 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7227 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7229 * tree-predcom.c (initialize_root): Delete.
7230 (execute_pred_commoning_chain): Initialize root vars and replace
7231 reference of non-combined chain directly, rather than call above
7234 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7236 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7237 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7239 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7241 * tree-predcom.c (struct chain): New field init_seq.
7242 (release_chain): Release init_seq.
7243 (prepare_initializers_chain): Record intialization stmts in above
7245 (insert_init_seqs): New function.
7246 (tree_predictive_commoning_loop): Call insert_init_seqs.
7248 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7250 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7252 2017-07-28 Richard Biener <rguenther@suse.de>
7254 * match.pd: Remove superfluous :c.
7255 * genmatch.c (simplify::id): Add member.
7256 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7258 (current_id): New global.
7259 (dt_node::parent): Move from ...
7260 (dt_operand::parent): ... here. Add for_id member.
7261 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7262 (decision_tree::find_node): Relax order requirement when
7263 merging DT_TRUE nodes to ones inbetween the current simplify
7264 and the one we try to merge with. Add diagnostic whenever
7265 we need to enforce pattern order by not merging.
7266 (decision_tree::insert): Set current_id.
7267 (decision_tree::print_node): Dump parent node and for_id.
7268 (parser::last_id): Add member.
7269 (parser::push_simplify): Assign unique id.
7270 (parser::parser): Initialize last_id.
7272 2017-07-28 Martin Liska <mliska@suse.cz>
7275 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7276 gimple_build_debug_bind.
7278 2017-07-28 Richard Biener <rguenther@suse.de>
7280 PR tree-optimization/81502
7281 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7283 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7285 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7286 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7287 for BIT_FIELD_REF args.
7288 * fold-const.c (make_bit_field_ref): Likewise.
7289 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7291 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7294 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7295 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7296 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7297 Or it into SANITIZER_UNDEFINED.
7298 * ubsan.c: Include gimple-fold.h and varasm.h.
7299 (ubsan_expand_ptr_ifn): New function.
7300 (instrument_pointer_overflow): New function.
7301 (maybe_instrument_pointer_overflow): New function.
7302 (instrument_object_size): Formatting fix.
7303 (pass_ubsan::execute): Call instrument_pointer_overflow
7304 and maybe_instrument_pointer_overflow.
7305 * internal-fn.c (expand_UBSAN_PTR): New function.
7306 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7307 * sanitizer.def (__ubsan_handle_pointer_overflow,
7308 __ubsan_handle_pointer_overflow_abort): New builtins.
7309 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7310 * internal-fn.def (UBSAN_PTR): New internal function.
7311 * opts.c (sanitizer_opts): Add pointer-overflow.
7312 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7313 * fold-const.c (build_range_check): Compute pointer range check in
7314 integral type if pointer arithmetics would be needed. Formatting
7317 2017-07-28 Martin Liska <mliska@suse.cz>
7320 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7321 parameters that are of a variable-length.
7323 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7325 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7327 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7328 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7329 (CRT_CALL_STATIC_FUNCTION): Likewise.
7330 (ASM_DEFAULT_SPEC): New define.
7331 (ASM_SPEC32): Likewise.
7332 (ASM_SPEC64): Likewise.
7333 (ASM_SPEC_COMMON): Likewise.
7334 (ASM_SPEC): Likewise.
7335 (INVALID_64BIT): Likewise.
7336 (LINK_OS_DEFAULT_SPEC): Likewise.
7337 (LINK_OS_SPEC32): Likewise.
7338 (LINK_OS_SPEC64): Likewise.
7339 (POWERPC_LINUX): Likewise.
7340 (PTRDIFF_TYPE): Likewise.
7341 (RESTORE_FP_PREFIX): Likewise.
7342 (RESTORE_FP_SUFFIX): Likewise.
7343 (SAVE_FP_PREFIX): Likewise.
7344 (SAVE_FP_SUFFIX): Likewise.
7345 (SIZE_TYPE): Likewise.
7346 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7347 (TARGET_64BIT): Likewise.
7348 (TARGET_64BIT): Likewise.
7349 (TARGET_AIX): Likewise.
7350 (WCHAR_TYPE_SIZE): Likewise.
7351 (WCHAR_TYPE): Undefine.
7352 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7353 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7354 (CPP_OS_RTEMS_SPEC): Delete.
7355 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7356 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7358 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7360 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7362 PR tree-optimization/81578
7363 * tree-parloops.c (build_new_reduction): Bail out if
7364 reduction_code isn't one of the standard OpenMP reductions.
7365 Move the details printing after that decision.
7367 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7369 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7370 related to reload_in_progress.
7371 (splat_input_operand): Likewise.
7372 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7374 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7376 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7377 (TARGET_INSTANTIATE_DECLS): Likewise.
7378 (legitimate_indexed_address_p): Delete reload_in_progress code.
7379 (rs6000_debug_legitimate_address_p): Likewise.
7380 (rs6000_eliminate_indexed_memrefs): Likewise.
7381 (rs6000_emit_le_vsx_store): Likewise.
7382 (rs6000_emit_move_si_sf_subreg): Likewise.
7383 (rs6000_emit_move): Likewise.
7384 (register_to_reg_type): Likewise.
7385 (rs6000_pre_atomic_barrier): Likewise.
7386 (rs6000_machopic_legitimize_pic_address): Likewise.
7387 (rs6000_allocate_stack_temp): Likewise.
7388 (rs6000_address_for_fpconvert): Likewise.
7389 (rs6000_address_for_altivec): Likewise.
7390 (rs6000_secondary_memory_needed_rtx): Delete function.
7391 (rs6000_check_sdmode): Likewise.
7392 (rs6000_alloc_sdmode_stack_slot): Likewise.
7393 (rs6000_instantiate_decls): Likewise.
7394 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7395 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7396 Delete reload_in_progress.
7397 (*vec_reload_and_plus_<mptrsize>): Likewise.
7398 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7399 (vsx_div_v2di): Likewise.
7400 (vsx_udiv_v2di): Likewise.
7402 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7404 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7405 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7406 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7407 (rs6000_debug_reg_global): Delete print of LRA status.
7408 (rs6000_option_override_internal): Delete dead LRA related code.
7409 (rs6000_lra_p): Delete function.
7410 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7412 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7414 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7415 * config/riscv/rtems.h: New file.
7417 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7418 Sudakshina Das <sudi.das@arm.com>
7420 * config/aarch64/aarch64.md
7421 (define_split for and<mode>3nr_compare): Move
7422 non aarch64_logical_operand to a register.
7423 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7424 register immediate operand to a register.
7425 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7427 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7430 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7432 2017-07-27 Richard Biener <rguenther@suse.de>
7434 PR tree-optimization/81573
7435 PR tree-optimization/81494
7436 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7437 multi defuse cycle case.
7439 2017-07-27 Richard Biener <rguenther@suse.de>
7441 PR tree-optimization/81571
7442 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7445 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7447 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7448 earlier and only if MASK_FPU is set. Adjust formatting.
7450 2017-07-27 Martin Liska <mliska@suse.cz>
7452 * opt-functions.awk: Add validation of value of Init.
7453 * optc-gen.awk: Pass new argument.
7455 2017-07-27 Martin Liska <mliska@suse.cz>
7457 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7458 Fix wrong condition.
7460 2017-07-27 Martin Liska <mliska@suse.cz>
7462 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7463 BBs and edges seen by autoFDO.
7465 2017-07-27 Richard Biener <rguenther@suse.de>
7467 PR tree-optimization/81502
7468 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7469 with incompatible but same sized type.
7470 (execute_update_addresses_taken): Likewise.
7472 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7474 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7475 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7477 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7480 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7481 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7482 Change s_operand to memory_operand.
7484 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7486 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7487 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7488 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7489 Emit instructions rather than returning an expression. Handle TFmode
7490 and KFmode by casting to TImode.
7491 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7492 (rs6000_emit_le_vsx_store): Likewise.
7493 * config/rs6000/vsx.md (VSX_TI): New iterator.
7494 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7495 (*vsx_le_undo_permute_<mode>): Likewise.
7496 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7497 emit the split sequence.
7498 (*vsx_le_perm_store_<mode>): Likewise.
7500 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7502 PR tree-optimization/81555
7503 PR tree-optimization/81556
7504 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7505 if true, force CHANGED for the recursive invocation.
7506 (reassociate_bb): Remember original length of ops array, pass
7507 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7509 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7510 attributes for noipa attribute. For naked attribute use
7511 lookup_attribute first before lookup_attribute_spec.
7512 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7514 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7515 for functions with noipa attribute.
7516 (cgraph_externally_visible_p): Return true for functions with noipa
7518 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7519 for functions with noipa attribute.
7520 * doc/extend.texi: Document noipa function attribute.
7521 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7522 also for functions with noipa attribute.
7523 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7525 2017-07-26 Andrew Pinski <apinski@cavium.com>
7527 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7528 vec_unalign_load_cost and vec_unalign_store_cost.
7530 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7532 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7533 -mvsx-small-integer option.
7534 (ISA_3_0_MASKS_IEEE): Likewise.
7535 (OTHER_VSX_VECTOR_MASKS): Likewise.
7536 (POWERPC_MASKS): Likewise.
7537 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7538 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7539 code, only testing for DImode being allowed in non-VSX floating
7541 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7542 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7544 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7545 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7546 TARGET_P8_VECTOR test.
7547 (rs6000_secondary_reload_simple_move): Likewise.
7548 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7549 since TARGET_P9_VECTOR was already tested.
7550 (rs6000_opt_masks): Remove -mvsx-small-integer.
7551 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7552 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7554 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7555 test for TARGET_VEXTRACTUB was used, and that uses
7557 (p9 extract splitter): Likewise.
7558 (vsx_extract_<mode>_di_p9): Likewise.
7559 (vsx_extract_<mode>_store_p9): Likewise.
7560 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7561 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7562 the elimination of TARGET_VSX_SMALL_INTEGER.
7563 (vsx_extract_<mode>_p8): Likewise.
7564 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7565 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7566 (vsx_set_<mode>_p9): Likewise.
7567 (vsx_set_v4sf_p9): Likewise.
7568 (vsx_set_v4sf_p9_zero): Likewise.
7569 (vsx_insert_extract_v4sf_p9): Likewise.
7570 (vsx_insert_extract_v4sf_p9_2): Likewise.
7571 * config/rs6000/rs6000.md (sign extend splitter): Change
7572 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7573 (floatsi<mode>2_lfiwax_mem): Likewise.
7574 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7575 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7576 since a test for TARGET_P9_VECTOR was used.
7577 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7578 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7579 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7580 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7581 TARGET_P8_VECTOR test.
7582 (fix_trunc<mode>si2_stfiwx): Likewise.
7583 (fix_trunc<mode>si2_internal): Likewise.
7584 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7585 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7587 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7588 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7589 TARGET_P8_VECTOR test.
7590 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7591 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7592 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7594 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7595 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7596 since a test for TARGET_P9_VECTOR was used.
7597 (splitter for loading small constants): Likewise.
7599 2017-07-26 Andrew Pinski <apinski@cavium.com>
7601 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7604 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7607 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7608 (fp_valid_at): Likewise.
7610 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7612 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7613 (qdf24xx_addrcost_table): Likewise.
7614 (cortexa57_tunings): Update to use generic_branch_cost.
7615 (cortexa72_tunings): Likewise.
7616 (cortexa73_tunings): Likewise.
7617 (qdf24xx_tunings): Likewise.
7619 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7621 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7622 (thunderx2t99_branch_cost): Likewise.
7623 (cortexa35_tunings): Update to use generic_branch_cost.
7624 (cortexa53_tunings): Likewise.
7625 (cortexa57_tunings): Likewise.
7626 (cortexa72_tunings): Likewise.
7627 (cortexa73_tunings): Likewise.
7628 (thunderx2t99_tunings): Likewise.
7630 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7632 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7633 (sparc_option_override): Honour MASK_FSMULD.
7634 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7635 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7636 * config/sparc/sparc.opt (mfsmuld): New option.
7637 * doc/invoke.texi (mfsmuld): Document option.
7639 2017-07-26 Marek Polacek <polacek@redhat.com>
7642 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7644 2017-07-26 Richard Biener <rguenther@suse.de>
7646 * gimple-match-head.c (do_valueize): Return OP if valueize
7648 (get_def): New helper to get at the def stmt of a SSA name
7650 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7651 do_valueize to get at the def stmt.
7652 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7654 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7657 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7659 2017-07-26 Martin Liska <mliska@suse.cz>
7662 * function.c (expand_function_start): Make expansion of
7663 nonlocal_goto_save_area after parm_birth_insn.
7665 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7667 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7668 from all CPU target flags enable members.
7670 2017-07-26 Richard Biener <rguenther@suse.de>
7672 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7673 (decision_tree::gen): Make 'type' const.
7674 (write_predicate): Likewise.
7676 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7678 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7679 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7680 (rs6000_option_override_internal): Likewise.
7681 (rs6000_expand_vector_set): Likewise.
7682 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7683 (TARGET_UPPER_REGS_SF): Likewise.
7684 (TARGET_UPPER_REGS_DI): Likewise.
7685 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7686 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7687 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7688 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7689 (Splitters for DI constants in Altivec registers): Likewise.
7690 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7691 (vsx_set_v4sf_p9): Likewise.
7692 (vsx_set_v4sf_p9_zero): Likewise.
7693 (vsx_insert_extract_v4sf_p9): Likewise.
7694 (vsx_insert_extract_v4sf_p9_2): Likewise.
7696 2017-07-25 Carl Love <cel@us.ibm.com>
7698 * doc/extend.texi: Update the built-in documentation file for the
7699 existing built-in functions
7700 vector signed char vec_cnttz (vector signed char);
7701 vector unsigned char vec_cnttz (vector unsigned char);
7702 vector signed short vec_cnttz (vector signed short);
7703 vector unsigned short vec_cnttz (vector unsigned short);
7704 vector signed int vec_cnttz (vector signed int);
7705 vector unsigned int vec_cnttz (vector unsigned int);
7706 vector signed long long vec_cnttz (vector signed long long);
7707 vector unsigned long long vec_cnttz (vector unsigned long long);
7709 2017-07-25 Andrew Pinski <apinski@cavium.com>
7711 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7712 accesses where the use is for the first operand of a BIT_INSERT.
7714 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7717 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7718 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7720 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7722 * config/i386/gstabs.h: Delete.
7723 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7725 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7727 * config/i386/i386.c (ix86_decompose_address): Do not check for
7728 register RTX when looking at index_reg or base_reg.
7729 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7731 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7733 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7734 to update EH info here.
7736 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7738 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7740 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7742 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7744 2017-07-25 Torsten Duwe <duwe@suse.de>
7746 * common.opt: Introduce -fpatchable-function-entry
7747 command line option, and its variables function_entry_patch_area_size
7748 and function_entry_patch_area_start.
7749 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7750 including a two-value parser.
7751 * target.def (print_patchable_function_entry): New target hook.
7752 * targhooks.h (default_print_patchable_function_entry): New function.
7753 * targhooks.c (default_print_patchable_function_entry): Likewise.
7754 * toplev.c (process_options): Switch off IPA-RA if
7755 patchable function entries are being generated.
7756 * varasm.c (assemble_start_function): Look at the
7757 patchable-function-entry command line switch and current
7758 function attributes and maybe generate NOP instructions by
7759 calling the print_patchable_function_entry hook.
7760 * doc/extend.texi: Document patchable_function_entry attribute.
7761 * doc/invoke.texi: Document -fpatchable_function_entry
7762 command line option.
7763 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7765 * doc/tm.texi: Re-generate.
7767 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7770 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7771 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7773 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7775 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7777 2017-07-25 Richard Biener <rguenther@suse.de>
7779 PR tree-optimization/81455
7780 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7781 not walk in cycles when looking for guards.
7783 2017-07-25 Richard Biener <rguenther@suse.de>
7785 PR tree-optimization/81529
7786 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7787 when optimizing backedge uses.
7789 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7791 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7793 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7794 to dl_section_ref. On AIX, append an expression to subtract
7795 the size of the section length to dl_section_ref.
7797 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7799 * configure.ac: If any of the config.* scripts fail, exit 1.
7800 * configure: Regenerate.
7802 2017-07-25 Richard Biener <rguenther@suse.de>
7805 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7806 of immediate uses, be more verbose on errors.
7808 2017-07-25 Richard Biener <rguenther@suse.de>
7810 PR tree-optimization/81510
7811 * tree-vect-loop.c (vect_is_simple_reduction): When the
7812 reduction stmt is not inside the loop bail out.
7814 2017-07-25 Richard Biener <rguenther@suse.de>
7816 PR tree-optimization/81303
7817 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7818 profitability check against LOOP_VINFO_NITERSM1.
7820 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7822 * domwalk.c (cmp_bb_postorder): Simplify.
7823 (sort_bbs_postorder): New function. Use it...
7824 (dom_walker::walk): ...here to optimize common cases.
7826 2017-07-25 Martin Liska <mliska@suse.cz>
7829 * ipa-visibility.c (function_and_variable_visibility): Make the
7830 redirection just on target that supports aliasing.
7831 Fix GNU coding style.
7833 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7836 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7838 * config/arm/rtems.h: Likewise.
7839 * config/bfin/rtems.h: Likewise.
7840 * config/i386/rtemself.h: Likewise.
7841 * config/lm32/rtems.h: Likewise.
7842 * config/m32c/rtems.h: Likewise.
7843 * config/m68k/rtemself.h: Likewise.
7844 * config/microblaze/rtems.h: Likewise.
7845 * config/mips/rtems.h: Likewise.
7846 * config/moxie/rtems.h: Likewise.
7847 * config/nios2/rtems.h: Likewise.
7848 * config/powerpcspe/rtems.h: Likewise.
7849 * config/rs6000/rtems.h: Likewise.
7850 * config/rtems.h: Likewise.
7851 * config/sh/rtems.h: Likewise.
7852 * config/sh/rtemself.h: Likewise.
7853 * config/sparc/rtemself.h: Likewise.
7855 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7858 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7859 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7860 * tree-ssa-structalias.c (alias_get_name): Same.
7862 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7865 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7866 instructions if no du chain is found.
7868 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7870 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7872 2017-07-25 Richard Biener <rguenther@suse.de>
7875 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7878 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7880 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7882 (ISA_2_7_MASKS_SERVER): Likewise.
7883 (ISA_3_0_MASKS_IEEE): Likewise.
7884 (OTHER_P8_VECTOR_MASKS): Likewise.
7885 (OTHER_VSX_VECTOR_MASKS): Likewise.
7886 (POWERPC_MASKS): Likewise.
7887 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7888 duplicate list of options.
7889 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7890 explicit -mupper-regs options.
7891 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7892 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7893 alias for -mupper-regs-df.
7894 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7895 (rs6000_init_hard_regno_mode_ok): Likewise.
7896 (rs6000_option_override_internal): Likewise.
7897 (rs6000_opt_masks): Likewise.
7898 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7899 options in terms of whether -mvsx or -mpower8-vector was used.
7900 (TARGET_UPPER_REGS_DI): Likewise.
7901 (TARGET_UPPER_REGS_SF): Likewise.
7902 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7903 -mupper-regs-* options.
7905 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7907 * passes.c (emergency_dump_function): Print some empty lines and a
7908 header before the RTL dump.
7910 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7912 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7914 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7917 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7918 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7920 2017-07-24 Carl Love <cel@us.ibm.com>
7922 * config/rs6000/rs6000-c.c: Add support for built-in functions
7923 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7924 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7925 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7926 vec_extract_fp_from_shortl): Add defines for the two builtins.
7927 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7928 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7930 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7931 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7932 * doc/extend.texi: Update the built-in documentation file for the
7933 new built-in function.
7935 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7938 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7940 * doc/generic.texi: Likewise.
7941 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7942 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7944 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7946 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7947 (aarch64_mls_elt_merge<mode>): Likewise.
7949 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7951 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7952 having __cxa_atexit.
7954 2017-07-23 Michael Collison <michael.collison@arm.com>
7956 * config/arm/arm.c (arm_option_override): Deprecate
7957 use of -mstructure-size-boundary.
7958 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7959 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7961 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7964 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7965 Reduce cost estimate for direct moves.
7967 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7970 * config/i386/i386.c (ix86_option_override_internal): Disable
7971 BMI, BMI2 and TBM instructions for -m16.
7973 2017-07-21 Carl Love <cel@us.ibm.com>
7975 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7976 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7977 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7978 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7979 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7980 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7981 VMULOSW): New enum "unspec" values.
7982 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7983 altivec_vmulosw): New patterns.
7984 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7985 VMULOSW): Add definitions.
7987 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7989 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7990 (qdf24xx): Likewise.
7991 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7992 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7993 (AARCH64_FL_V8_1): Renumber.
7994 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7995 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7996 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7997 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7998 rdma to feature modifiers list.
8000 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8003 * ipa-visibility (function_and_variable_visibility): Convert
8004 recursive PLT call to direct call if appropriate.
8006 2017-07-21 Andrew Pinski <apinski@cavium.com>
8008 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8009 operand 1 to see if the types precision matches.
8010 * fold-const.c (operand_equal_p): Likewise.
8012 2017-07-21 Richard Biener <rguenther@suse.de>
8014 PR tree-optimization/81303
8015 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8016 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8017 (vect_peeling_hash_get_lowest_cost): Adjust.
8018 (vect_enhance_data_refs_alignment): Likewise. Use
8019 vect_get_peeling_costs_all_drs to compute the penalty for no
8020 peeling to match up costs.
8022 2017-07-21 Richard Biener <rguenther@suse.de>
8024 PR tree-optimization/81500
8025 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8026 we didn't identify a reduction path.
8028 2017-07-21 Tom de Vries <tom@codesourcery.com>
8029 Cesar Philippidis <cesar@codesourcery.com>
8031 PR gcov-profile/81442
8032 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8035 2017-07-21 Tom de Vries <tom@codesourcery.com>
8038 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8040 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8041 nvptx_override_options_after_change.
8043 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8045 * dwarf2out.c (output_file_names): Avoid double testing for
8048 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8050 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8052 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8054 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8056 (try_crossjump_to_edge): Do not punt on partitioned functions.
8058 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8060 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8061 Put all BBs reachable only via paths crossing cold region to cold
8063 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8065 2016-07-21 Richard Biener <rguenther@suse.de>
8067 PR tree-optimization/81303
8068 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8069 into account prologue and epilogue iterations when raising
8070 min_profitable_iters to sth at least covering one vector iteration.
8072 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8074 * config/arm/arm.c (arm_test_cpu_arch_dat):
8077 2017-07-20 Nathan Sidwell <nathan@acm.org>
8079 Remove TYPE_METHODS.
8080 * tree.h (TYPE_METHODS): Delete.
8081 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8082 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8083 (dbxout_type_methods): Scan TYPE_FIELDS.
8084 (dbxout_type): Don't check TYPE_METHODS here.
8085 * function.c (use_register_for_decl): Always ignore register for
8086 class types when not optimizing.
8087 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8088 * tree.c (free_lang_data_in_type): Stitch out member functions and
8089 templates from TYPE_FIELDS.
8090 (build_distinct_type_copy, verify_type_variant,
8091 verify_type): Member fns are on TYPE_FIELDS.
8092 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8093 * tree-pretty-print.c (dump_generic_node): Likewise.
8095 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8098 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8099 V2TImode and V4TImode.
8100 (ix86_expand_vector_extract): Likewise.
8101 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8102 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8103 (ssescalarmode): Handle V4TImode and V2TImode.
8104 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8105 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8106 (VEXTRACTI128_MODE): New mode iterator.
8107 (splitter for *vec_extractv?ti first element): New.
8108 (VEC_INIT_MODE): New mode iterator.
8109 (vec_init<mode>): Consolidate 3 expanders into one using
8110 VEC_INIT_MODE mode iterator.
8112 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8114 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8115 non_spilled_static_chain_regno_p.
8117 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8119 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8121 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8123 * bb-reorder.c (connect_traces): Allow copying of blocks within
8126 2017-07-20 Richard Biener <rguenther@suse.de>
8128 * gimple.h (gimple_phi_result): Add gphi * overload.
8129 (gimple_phi_result_ptr): Likewise.
8130 (gimple_phi_arg): Likewise. Adjust index assert to only
8131 allow actual argument accesses rather than all slots available
8133 (gimple_phi_arg_def): Add gphi * overload.
8134 * tree-phinodes.c (make_phi_node): Initialize only actual
8136 (resize_phi_node): Clear memory not covered by old node,
8137 do not initialize excess argument slots.
8138 (reserve_phi_args_for_new_edge): Initialize new argument slot
8141 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8143 PR tree-optimization/81388
8145 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8147 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8148 by removing computation of may_be_zero.
8150 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8151 Tom de Vries <tom@codesourcery.com>
8154 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8155 when gimple level profile disagrees with what RTL expander did.
8157 2017-07-20 Richard Biener <rguenther@suse.de>
8159 PR tree-optimization/61171
8160 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8161 (vect_analyze_stmt): Add slp instance parameter.
8162 (vectorizable_reduction): Likewise.
8163 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8164 (vect_is_simple_reduction): Deal with chains not detected
8165 as SLP reduction chain, specifically not properly associated
8166 chains containing a mix of plus/minus.
8167 (get_reduction_op): Remove.
8168 (get_initial_defs_for_reduction): Simplify, pass in whether
8169 this is a reduction chain, pass in the SLP node for the PHIs.
8170 (vect_create_epilog_for_reduction): Get the SLP instance as
8172 (vectorizable_reduction): Get the SLP instance as arg.
8173 During analysis remember the SLP node with the PHIs in the
8174 instance. Simplify getting at the vectorized reduction PHIs.
8175 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8176 through SLP instance.
8177 (vect_slp_analyze_operations): Likewise.
8178 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8179 (vect_transform_stmt): Likewise.
8181 2017-07-20 Tom de Vries <tom@codesourcery.com>
8183 PR tree-optimization/81489
8184 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8185 read of phi arg location to before loop that modifies phi.
8187 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8189 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8192 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8195 * except.c (execute): Fix ordering issue.
8197 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8199 PR rtl-optimization/81423
8200 * combine.c (make_compound_operation_int): Don't try to optimize
8201 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8203 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8205 PR rtl-optimization/81423
8206 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8207 with a constant that is -1 in the truncated to mode.
8209 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8211 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8212 (determine_unlikely_bbs): ... here.
8213 * predict.h (propagate_unlikely_bbs_forward): Declare.
8214 * cfgexpand.c (pass_expand::execute): Use it.
8215 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8217 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8218 propagate_unlikely_bbs_forward.
8220 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8223 * except.c (maybe_add_nop_after_section_switch): New function.
8226 2017-07-19 Tom de Vries <tom@codesourcery.com>
8228 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8230 2017-07-19 Tom de Vries <tom@codesourcery.com>
8232 * gimple.h (gimple_phi_arg): Make assert more strict.
8234 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8236 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8237 * config/rs6000/mmintrin.h: New file.
8238 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8240 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8242 PR tree-optimization/81346
8243 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8245 2017-07-19 Tom de Vries <tom@codesourcery.com>
8247 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8249 2017-07-19 Tom de Vries <tom@codesourcery.com>
8251 * config/nvptx/nvptx-modes.def: Add V2DImode.
8252 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8253 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8254 (nvptx_output_mov_insn): Handle lack of mov.b128.
8255 (nvptx_print_operand): Handle 'H' and 'L' codes.
8256 (nvptx_vector_mode_supported): Allow V2DImode.
8257 (nvptx_preferred_simd_mode): New function.
8258 (nvptx_data_alignment): New function.
8259 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8260 nvptx_preferred_simd_mode.
8261 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8263 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8265 2017-07-19 Tom de Vries <tom@codesourcery.com>
8267 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8268 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8269 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8270 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8271 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8272 (mov<VECIM>_insn): New define_insn.
8273 (define_expand "mov<VECIM>): New define_expand.
8275 2017-07-19 Tom de Vries <tom@codesourcery.com>
8277 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8279 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8281 PR tree-optimization/81346
8282 * fold-const.h (fold_div_compare, range_check_type): Declare.
8283 * fold-const.c (range_check_type): New function.
8284 (build_range_check): Use range_check_type.
8285 (fold_div_compare): No longer static, rewritten into
8286 a match.pd helper function.
8287 (fold_comparison): Don't call fold_div_compare here.
8288 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8291 2017-07-19 Nathan Sidwell <nathan@acm.org>
8293 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8294 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8295 * tree.c (find_decls_types_r, verify_type): Use
8296 TYPE_{MIN,MAX}_VALUE_RAW.
8297 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8298 (hash_tree): Likewise.
8299 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8301 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8304 2017-07-18 Tom de Vries <tom@codesourcery.com>
8307 * omp-expand.c (expand_omp_for_static_chunk): Handle
8308 equal-argument loop exit phi.
8310 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8313 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8314 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8315 operand 2 predicate.
8316 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8317 operand 2 predicate.
8318 (ror,rol -> rorx splitters): Use const_int_operand as
8319 operand 2 predicate.
8321 2017-06-18 Richard Biener <rguenther@suse.de>
8323 PR tree-optimization/81410
8324 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8325 the gap in the ! slp_perm SLP case after each group.
8327 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8330 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8333 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8336 * predict.c (set_even_probabilities): Cleanup; do not affect
8337 probabilities that are already known.
8338 (combine_predictions_for_bb): Call even when count is set.
8340 2017-07-18 Nathan Sidwell <nathan@acm.org>
8342 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8345 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8348 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8349 optimization for loop niter analysis.
8351 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8354 * config/avr/avr.c (avr_optimize_casesi): Don't use
8355 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8357 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8359 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8360 body_cost_vec from _vect_peel_extended_info.
8361 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8362 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8365 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8367 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8369 2017-07-18 Richard Biener <rguenther@suse.de>
8371 PR tree-optimization/80620
8372 PR tree-optimization/81403
8373 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8374 info when re-using a VN table entry.
8376 2017-07-18 Richard Biener <rguenther@suse.de>
8378 PR tree-optimization/81418
8379 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8380 vectype_in. Verify that with lane-reducing reduction operations
8381 we have a single def-use cycle.
8383 2017-07-17 Carl Love <cel@us.ibm.com>
8385 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8387 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8388 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8389 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8390 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8391 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8392 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8393 VMULOSW): New enum "unspec" values.
8394 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8395 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8396 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8397 altivec_vmulosw): New patterns.
8398 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8399 VMULOSW): Add definitions.
8400 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8402 * config/alpha/alpha.c: Include predict.h.
8404 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8406 * tree-vrp.c (compare_assert_loc): Fix comparison function
8407 to return predictable results.
8409 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8411 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8414 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8415 * doc/invoke.texi (mexpand-adddi): Update text.
8417 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8419 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8420 that also clobbers the CC register. The old expand code is moved
8422 (*arc_clzsi2): ... here.
8423 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8424 the CC register. The old expand code is moved to ...
8425 (arc_ctzsi2): ... here.
8427 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8429 * config/arc/arc.opt (mindexed-loads): Use initial value
8430 TARGET_INDEXED_LOADS_DEFAULT.
8431 (mauto-modify-reg): Use initial value
8432 TARGET_AUTO_MODIFY_REG_DEFAULT.
8433 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8434 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8435 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8436 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8438 2017-07-17 Martin Liska <mliska@suse.cz>
8441 * opts.c (finish_options): Do not allow -fgnu-tm
8442 w/ -fsanitize={kernel-,}address. Say sorry.
8444 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8447 * tree-loop-distribution.c (classify_partition): Only assert on
8448 numer of iterations.
8449 (merge_dep_scc_partitions): Delete prameter. Update function call.
8450 (distribute_loop): Remove code handling loop with unknown niters.
8451 (pass_loop_distribution::execute): Skip loop with unknown niters.
8453 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8456 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8457 function sort_partitions_by_post_order.
8459 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8461 PR tree-optimization/81374
8462 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8463 the max index of basic blocks, rather than number of basic blocks.
8465 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8467 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8469 (arc_legitimate_pic_operand_p): Likewise.
8470 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8472 (arc_needs_pcl_p): Likewise.
8473 (arc_legitimate_pc_offset_p): Likewise.
8474 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8475 function is also used in constrains.md.
8476 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8477 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8478 PLUS. Only return true/false in known cases, otherwise assert.
8479 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8480 is already called in arc_legitimate_constant_p.
8481 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8483 (LEGITIMATE_PIC_OPERAND_P): Use
8484 arc_raw_symbolic_reference_mentioned_p function.
8485 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8490 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8491 Andrew Burgess <andrew.burgess@embecosm.com>
8493 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8494 (arc_return_address_register): New function.
8495 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8496 (arc_handle_fndecl_attribute): Add naked attribute.
8497 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8498 (TARGET_WARN_FUNC_RETURN): Likewise.
8499 (arc_allocate_stack_slots_for_args): New function.
8500 (arc_warn_func_return): Likewise.
8501 (machine_function): Change type fn_type.
8502 (arc_compute_function_type): Consider new naked function type,
8503 change function return type.
8504 (arc_must_save_register): Adapt to handle new
8505 arc_compute_function_type's return type.
8506 (arc_expand_prologue): Likewise.
8507 (arc_expand_epilogue): Likewise.
8508 (arc_return_address_regs): Delete.
8509 (arc_return_address_register): New function.
8510 (arc_epilogue_uses): Use above function.
8511 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8512 (arc_function_type): Change encoding, add naked type.
8513 (ARC_INTERRUPT_P): Change to handle the new encoding.
8514 (ARC_FAST_INTERRUPT_P): Likewise.
8515 (ARC_NORMAL_P): Define.
8516 (ARC_NAKED_P): Likewise.
8517 (arc_compute_function_type): Delete prototype.
8518 * config/arc/arc.md (in_ret_delay_slot): Use
8519 arc_return_address_register function.
8520 (simple_return): Likewise.
8521 (p_return_i): Likewise.
8523 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8525 PR tree-optimization/81428
8526 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8527 can't be built for those types.
8529 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8531 Remove stuff dead since r239246.
8533 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8534 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8535 (avr_inform_devices): Remove dead stuff.
8537 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8539 * config/arm/arm_neon.h: Fix softp typo.
8541 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8543 PR tree-optimization/81365
8544 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8545 aggregate moves onto bb predecessor edges, make sure there are no
8546 loads that could alias the lhs in between the start of bb and the
8549 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8552 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8553 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8554 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8556 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8558 PR tree-optimization/81396
8559 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8560 (init_symbolic_number): Initialize it to 1.
8561 (perform_symbolic_merge): Add n_ops from both operands into the new
8563 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8564 without base_addr as useless if they need more than one operation.
8565 (bswap_replace): Handle !bswap case for NULL base_addr.
8567 2017-07-17 Tom de Vries <tom@codesourcery.com>
8570 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8573 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8575 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8576 conditional builtin define __FIX_LEON3FT_B2BST.
8578 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8580 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8581 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8584 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8586 PR rtl-optimization/81424
8587 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8588 to remove potential trapping from operands if -fnon-call-exceptions.
8590 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8592 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8593 profile_proability for scalling.
8594 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8596 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8598 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8600 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8602 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8603 fixpoint arithmetics.
8605 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8607 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8608 fixpoint arithmetics.
8610 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8612 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8613 fixpoint arithmetics.
8615 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8617 * profile-count.h (profile_probability::from_reg_br_prob_note,
8618 profile_probability::to_reg_br_prob_note): New functions.
8619 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8620 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8621 * predict.c (probability_reliable_p): Update.
8622 (edge_probability_reliable_p): Update.
8623 (br_prob_note_reliable_p): Update.
8624 (invert_br_probabilities): Update.
8625 (add_reg_br_prob_note): New function.
8626 (combine_predictions_for_insn): Update.
8627 * asan.c (asan_clear_shadow): Update.
8628 * cfgbuild.c (compute_outgoing_frequencies): Update.
8629 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8630 (update_br_prob_note): Update.
8631 (rtl_verify_edges): Update.
8632 (purge_dead_edges): Update.
8633 (fixup_reorder_chain): Update.
8634 * emit-rtl.c (try_split): Update.
8635 * ifcvt.c (cond_exec_process_insns): Update.
8636 (cond_exec_process_if_block): Update.
8637 (dead_or_predicable): Update.
8638 * internal-fn.c (expand_addsub_overflow): Update.
8639 (expand_neg_overflow): Update.
8640 (expand_mul_overflow): Update.
8641 * loop-doloop.c (doloop_modify): Update.
8642 * loop-unroll.c (compare_and_jump_seq): Update.
8643 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8644 * predict.h: Update.
8645 * reorg.c (mostly_true_jump): Update.
8647 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8648 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8649 * config/arc/arc.c: (emit_unlikely_jump): Update.
8650 * config/arm/arm.c: (emit_unlikely_jump): Update.
8651 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8652 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8653 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8654 (ix86_print_operand): Update.
8655 (ix86_split_fp_branch): Update.
8656 (predict_jump): Update.
8657 * config/ia64/ia64.c (ia64_print_operand): Update.
8658 * config/mmix/mmix.c (mmix_print_operand): Update.
8659 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8660 (rs6000_expand_split_stack_prologue): Update.
8661 * config/rs6000/rs6000.c: Update.
8662 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8663 (s390_expand_vec_movstr): Update.
8664 (s390_expand_cs_tdsi): Update.
8665 (s390_expand_split_stack_prologue): Update.
8666 * config/sh/sh.c (sh_print_operand): Update.
8667 (expand_cbranchsi4): Update.
8668 (expand_cbranchdi4): Update.
8669 * config/sparc/sparc.c (output_v9branch): Update.
8670 * config/spu/spu.c (get_branch_target): Update.
8671 (ea_load_store_inline): Update.
8672 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8673 * config/tilepro/tilepro.c: Update.
8675 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8677 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8678 (gimplify_save_expr): Likewise.
8680 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8682 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8684 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8686 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8688 * timevar.def (TV_IPA_FNSUMMARY): Define.
8690 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8692 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8693 to back store errata sensitive sequence from being generated.
8694 (sqrtdf2_fix): Likewise.
8696 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8698 * tree-ssa-threadupdate.c (compute_path_counts,
8699 update_joiner_offpath_counts): Use profile_probability.
8701 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8704 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8706 * config/arm/arm-c.c (arm_cpu_builtins): Define
8707 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8709 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8711 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8712 array entries to represent __ieee128 versions of the
8713 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8714 scalar_extract_sig, and scalar_insert_exp built-in functions.
8715 (altivec_resolve_overloaded_builtin): Add special case handling
8716 for the __builtin_scalar_insert_exp function, as represented by
8717 the P9V_BUILTIN_VEC_VSIEDP constant.
8718 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8719 exponent support for __ieee128 argument.
8720 (VSESQP): Add scalar extract signature support for __ieee128
8722 (VSTDCNQP): Add scalar test negative support for __ieee128
8724 (VSIEQP): Add scalar insert exponent support for __int128 argument
8725 with __ieee128 result.
8726 (VSIEQPF): Add scalar insert exponent support for __ieee128
8727 argument with __ieee128 result.
8728 (VSTDCQP): Add scalar test data class support for __ieee128
8730 (VSTDCNQP): Add overload support for scalar test negative with
8732 (VSTDCQP): Add overload support for scalar test data class
8734 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8736 (UNSPEC_VSX_SIEXPQP): New constant.
8737 (xsxexpqp): New insn for VSX scalar extract exponent quad
8739 (xsxsigqp): New insn for VSX scalar extract significand quad
8741 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8742 precision with floating point argument.
8743 (xststdcqp): New expand for VSX scalar test data class quad
8745 (xststdcnegqp): New expand for VSX scalar test negative quad
8747 (xststdcqp): New insn to match expansions for VSX scalar test data
8748 class quad precision and VSX scalar test negative quad precision.
8749 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8750 special case operand checking to enforce that second operand of
8751 VSX scalar test data class with quad precision argument is a 7-bit
8753 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8754 prototypes and descriptions of __ieee128 versions of
8755 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8756 scalar_test_data_class, and scalar_test_neg built-in functions.
8758 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8760 PR tree-optimization/81162
8761 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8762 replace a negate with an add.
8764 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8766 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8768 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8770 * config/arm/arm-c.c (arm_cpu_builtins): Define
8771 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8773 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8775 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8776 (armv8-r): Set ARM Cortex-R52 as default CPU.
8777 * config/arm/arm-tables.opt: Regenerate.
8778 * config/arm/arm-tune.md: Regenerate.
8779 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8781 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8782 extension for -mcpu=cortex-r52.
8784 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8786 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8787 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8788 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8789 (fp-armv8): Define it as FP_ARMv8 only.
8790 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8791 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8793 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8794 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8795 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8796 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8797 than TARGET_FPU_ARMV8.
8798 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8799 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8800 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8802 * config/arm/neon.md (neon_vrint): Likewise.
8803 (neon_vcvt): Likewise.
8804 (neon_<fmaxmin_op><mode>): Likewise.
8805 (<fmaxmin><mode>3): Likewise.
8806 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8807 * config/arm/predicates.md (arm_cond_move_operator): Check against
8808 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8810 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8812 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8815 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8817 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8818 loop in comment with memset.
8820 2017-07-14 Martin Liska <mliska@suse.cz>
8822 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8823 * dwarf2out.c (is_java): Remove the function.
8824 (output_pubname): Remove usage of the function.
8825 (lower_bound_default): Remove usage of DW_LANG_Java.
8826 (gen_compile_unit_die): Likewise.
8827 * gcc.c: Remove compiler defaults for .java and .zip files.
8828 * gimple-expr.c (remove_suffix): Change as there's no longer
8829 extension than 4-letter one.
8830 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8831 (gimplify_save_expr): Likewise.
8832 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8833 as it's possible even for other languages than Java.
8834 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8835 * lto-opts.c (lto_write_options): Remove reference to Java.
8836 * opts.c (strip_off_ending): Update file extension handling.
8837 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8838 * tree-eh.c (lower_resx): Likewise.
8839 * tree.c (free_lang_data_in_type): Remove dead code.
8840 (find_decls_types_r): Likewise.
8841 (build_common_builtin_nodes): Remove Java from a comment.
8842 (verify_type): Remove dead code.
8843 * varasm.c (assemble_external): Remove Java from a comment.
8845 2017-07-14 Martin Liska <mliska@suse.cz>
8847 * opts.c (finish_options): Add quotes.
8848 (common_handle_option): Likewise.
8850 2017-07-14 Martin Liska <mliska@suse.cz>
8852 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8853 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8855 * dwarf2out.c (lower_bound_default): Do not handle
8857 (gen_compile_unit_die): Likewise.
8858 * gcc.c: Remove default extension binding for GNU Pascal.
8859 * stmt.c: Remove Pascal language from a comment.
8860 * xcoffout.c: Likewise.
8862 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8865 * diagnostic-show-locus.c (fixit_cmp): New function.
8866 (layout::layout): Sort m_fixit_hints.
8867 (column_range::column_range): Assert that the values are valid.
8868 (struct char_span): New struct.
8869 (correction::overwrite): New method.
8870 (struct source_line): New struct.
8871 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8872 calls in terms of classes source_line and char_span, and
8873 correction::overwrite.
8874 (selftest::test_overlapped_fixit_printing_2): New function.
8875 (selftest::diagnostic_show_locus_c_tests): Call it.
8877 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8879 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8880 early if there is no lhs.
8882 2017-07-13 Martin Liska <mliska@suse.cz>
8884 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8885 (gen_reference_type_die): Likewise.
8886 * stor-layout.c: Remove Pascal-related comment.
8888 2017-07-13 Martin Liska <mliska@suse.cz>
8890 * opts.c (finish_options): Add quotes to error messages.
8891 (parse_sanitizer_options): Likewise.
8893 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8895 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8897 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8899 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8901 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8903 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8905 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8907 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8910 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8911 provides the hardware capability bits, define the macro
8912 __BUILTIN_CPU_SUPPORTS__.
8913 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8914 if GLIBC does not provide the hardware capability bits. Add a
8915 gcc_unreachable call if the built-in cpu function is neither
8916 __builtin_cpu_is nor __builtin_cpu_supports.
8917 (rs6000_get_function_versions_dispatcher): Change the warning
8918 that an old GLIBC is used which does not export the capability
8919 bits to be an error.
8920 * doc/extend.texi (target_clones attribute): Document the
8921 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8922 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8923 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8924 the macros defined by GCC if the newer GLIBC is available.
8926 2017-07-12 Jeff Law <law@redhat.com>
8928 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8929 remaining includes slightly.
8930 * config/riscv/riscv-builtins.c: Include profile-count.h.
8932 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8935 * config/avr/avr.c (avr_set_current_function): In diagnostic
8936 messages: Quote keywords and (parts of) identifiers.
8937 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8940 2017-07-12 Carl Love <cel@us.ibm.com>
8942 * config/rs6000/rs6000-c.c: Add support for built-in functions
8943 vector bool char vec_revb (vector bool char);
8944 vector bool short vec_revb (vector short char);
8945 vector bool int vec_revb (vector bool int);
8946 vector bool long long vec_revb (vector bool long long);
8947 * doc/extend.texi: Update the built-in documentation file for the
8948 new built-in functions.
8950 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8952 * config/s390/s390.md: Remove movcc splitter.
8954 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8956 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8957 load/store on condition.
8959 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8962 * config/avr/avr.c (avr_encode_section_info)
8963 [progmem && !TREE_READONLY]: Error if progmem object needs
8966 2017-07-11 Michael Collison <michael.collison@arm.com>
8968 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8971 2017-07-11 Carl Love <cel@us.ibm.com>
8973 * config/rs6000/rs6000-c.c: Add support for builtins
8974 vector unsigned int vec_parity_lsbb (vector signed int);
8975 vector unsigned int vec_parity_lsbb (vector unsigned int);
8976 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8977 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8978 vector unsigned long long vec_parity_lsbb (vector signed long long);
8979 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8980 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8981 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8982 * doc/extend.texi: Update the built-in documentation file for the
8983 new built-in functions.
8985 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8987 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8988 (layout::m_primary_loc): New field.
8989 (layout::layout): Initialize new field. Move location filtering
8990 logic from here to...
8991 (layout::maybe_add_location_range): ...this new method. Add
8992 support for filtering to just the lines already specified by other
8994 (layout::will_show_line_p): New method.
8995 (gcc_rich_location::add_location_if_nearby): New method.
8996 (selftest::test_add_location_if_nearby): New test function.
8997 (selftest::diagnostic_show_locus_c_tests): Call it.
8998 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9001 2017-07-11 Tom de Vries <tom@codesourcery.com>
9003 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9004 (bb_first_real_insn): New function.
9005 (nvptx_single): Add extra initialization of broadcasted condition
9008 2017-07-11 Nathan Sidwell <nathan@acm.org>
9010 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9012 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9014 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9015 Binutils doc as TEXI will mess them up.
9016 * doc/invoke.texi (AVR Options): Same here.
9018 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9020 * config/sparc/sparc.opt (mfix-ut700): New option.
9021 (mfix-gr712rc): Likewise.
9022 (sparc_fix_b2bst): New variable.
9023 * doc/invoke.texi (SPARC options): Document them.
9024 (ARM options): Fix warnings.
9025 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9026 instructions to prevent sequences that can trigger the store-store
9027 errata for certain LEON3FT processors.
9028 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9029 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9030 * config/sparc/sparc.md (fix_b2bst): New attribute.
9031 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9033 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9036 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9038 (*rsqrtsf2_sse): Ditto.
9040 (div<mode>3): Macroize insn from divdf3 and divsf3
9041 using MODEF mode iterator.
9043 2017-07-10 Martin Sebor <msebor@redhat.com>
9045 PR tree-optimization/80397
9046 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9047 instead of testing for equality to INTEGER_TYPE.
9049 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9051 * config.gcc: Remove uclibc from arc target spec.
9053 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9055 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9057 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9060 * lto-wrapper.c (remove_option): New function.
9061 (merge_and_complain): Merge PIC/PIE options more realistically.
9063 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9065 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9069 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9070 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9071 * config.in: Regenerate.
9072 * configure: Regenerate.
9073 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9074 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9075 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9076 (TARGET_GASISR_PROLOGUES): ...target mask.
9077 * common/config/avr/avr-common.c
9078 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9079 Set -mgas-isr-prologues.
9080 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9081 INSERT_PASS_BEFORE for it.
9082 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9083 * config/avr/avr.c (avr_option_override)
9084 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9085 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9086 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9087 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9088 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9090 (make_avr_pass_pre_proep): New function.
9091 (emit_push_sfr) <treg>: Add argument to function and use it
9093 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9094 and set machine->gasisr.yes.
9095 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9096 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9097 __gcc_isr.n_pushed to .L__stack_usage.
9098 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9099 (avr_asm_final_postscan_insn): ...this new static function.
9100 * config/avr/avr.h (machine_function)
9101 <is_no_gccisr, use_L__stack_usage>: New fields.
9102 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9103 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9104 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9105 (gasisr, *gasisr): New expander and insn.
9106 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9107 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9108 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9110 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9112 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9115 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9117 Move jump-tables out of .text again.
9120 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9121 (ASM_OUTPUT_ADDR_VEC): New function.
9122 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9123 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9124 INSN_ADDRESSes as asm comment.
9125 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9126 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9127 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9128 * config/avr/avr.md (*tablejump): Adjust comment.
9129 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9130 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9132 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9133 (avr_output_addr_vec): New proto.
9134 (avr_log_t) <insn_addresses>: New field.
9136 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9139 * config/i386/i386.c (ix86_function_arg_advance): Set
9140 outgoing_args_on_stack to true if there are outgoing arguments
9142 (ix86_function_arg): Likewise.
9143 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9144 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9145 * config/i386/i386.h (machine_function): Add
9146 outgoing_args_on_stack.
9148 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9150 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9152 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9154 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9156 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9157 (REAL_H): Remove $(MACHMODE_H).
9158 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9160 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9161 $(MACHMODE_H) and double-int.h.
9162 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9164 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9167 2017-07-07 Andrew Pinski <apinski@cavium.com>
9169 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9170 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9172 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9174 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9175 Add warning if GCC was not configured to link against a GLIBC that
9176 exports the hardware capability bits.
9177 (make_resolver_func): Make resolver function private and not a
9178 COMDAT function. Create the name with clone_function_name instead
9179 of make_unique_name.
9182 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9183 correct operand in doing the split.
9185 2017-07-07 Carl Love <cel@us.ibm.com>
9187 * config/rs6000/rs6000-c: Add support for built-in function
9188 vector unsigned short vec_pack_to_short_fp32 (vector float,
9190 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9191 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9192 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9193 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9194 (convert_4f32_8i16): Add define_expand.
9195 * doc/extend.texi: Update the built-in documentation file for the
9196 new built-in function.
9198 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9200 * config/sparc/m8.md: New file.
9201 * config/sparc/sparc.md: Include m8.md.
9203 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9205 * config/sparc/sparc.opt: New option -mvis4b.
9206 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9207 (sparc_option_override): Handle VIS4B.
9208 (enum sparc_builtins): Define
9209 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9210 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9211 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9212 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9213 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9214 (check_constant_argument): New function.
9215 (sparc_vis_init_builtins): Define builtins
9216 __builtin_vis_dictunpack{8,16,32},
9217 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9218 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9219 __builtin_vis_fpcmpde{8,16,32}shl and
9220 __builtin_vis_fpcmpur{8,16,32}shl.
9221 (sparc_expand_builtin): Check that the constant operands to
9222 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9223 constant and in range.
9224 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9226 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9227 (SPARC_IMM5_P): Likewise.
9228 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9229 (enabled): Handle vis4b.
9230 (UNSPEC_DICTUNPACK): New unspec.
9231 (UNSPEC_FPCMPSHL): Likewise.
9232 (UNSPEC_FPUCMPSHL): Likewise.
9233 (UNSPEC_FPCMPDESHL): Likewise.
9234 (UNSPEC_FPCMPURSHL): Likewise.
9235 (cpu_feature): New CPU feature `vis4b'.
9236 (dictunpack{8,16,32}): New insns.
9237 (FPCSMODE): New mode iterator.
9238 (fpcscond): New code iterator.
9239 (fpcsucond): Likewise.
9240 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9241 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9242 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9243 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9244 * config/sparc/constraints.md: Define constraints `q' for unsigned
9245 2-bit integer constants and `t' for unsigned 5-bit integer
9247 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9249 (imm5_operand_dictunpack16): Likewise.
9250 (imm5_operand_dictunpack32): Likewise.
9251 (imm2_operand): Likewise.
9252 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9253 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9254 ditunpack* and fpcmp*shl builtins.
9256 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9258 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9259 * config.in: Add HAVE_AS_SPARC6 define.
9260 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9262 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9264 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9265 (CPP_CPU_SPEC): Handle m8.
9266 (ASM_CPU_SPEC): Likewise.
9267 * config/sparc/sparc-opts.h (enum processor_type): Add
9269 * config/sparc/sparc.c (m8_costs): New struct.
9270 (sparc_option_override): Handle TARGET_CPU_m8.
9271 (sparc32_initialize_trampoline): Likewise.
9272 (sparc64_initialize_trampoline): Likewise.
9273 (sparc_issue_rate): Likewise.
9274 (sparc_register_move_cost): Likewise.
9275 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9276 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9277 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9278 (CPP_CPU_SPEC): Handle M8.
9279 (ASM_CPU_SPEC): Likewise.
9280 (AS_M8_FLAG): Define.
9281 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9282 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9283 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9285 * configure: Regenerate.
9286 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9289 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9291 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9293 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9294 ("*movdi_insn_sp32"): Do not set v3pipe.
9295 ("*movsi_insn"): Likewise.
9296 ("*movdi_insn_sp64"): Likewise.
9297 ("*movsf_insn"): Likewise.
9298 ("*movdf_insn_sp32"): Likewise.
9299 ("*movdf_insn_sp64"): Likewise.
9300 ("*zero_extendsidi2_insn_sp64"): Likewise.
9301 ("*sign_extendsidi2_insn"): Likewise.
9302 ("*mov<VM32:mode>_insn"): Likewise.
9303 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9304 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9305 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9306 ("<vlop:code><VL:mode>3"): Likewise.
9307 ("*not_<vlop:code><VL:mode>3"): Likewise.
9308 ("*nand<VL:mode>_vis"): Likewise.
9309 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9310 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9311 ("one_cmpl<VL:mode>2"): Likewise.
9312 ("faligndata<VM64:mode>_vis"): Likewise.
9313 ("alignaddrsi_vis"): Likewise.
9314 ("alignaddrdi_vis"): Likweise.
9315 ("alignaddrlsi_vis"): Likewise.
9316 ("alignaddrldi_vis"): Likewise.
9317 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9318 ("bmaskdi_vis"): Likewise.
9319 ("bmasksi_vis"): Likewise.
9320 ("bshuffle<VM64:mode>_vis"): Likewise.
9321 ("cmask8<P:mode>_vis"): Likewise.
9322 ("cmask16<P:mode>_vis"): Likewise.
9323 ("cmask32<P:mode>_vis"): Likewise.
9324 ("pdistn<P:mode>_vis"): Likewise.
9325 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9327 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9329 * config/sparc/sparc.md ("subtype"): New insn attribute.
9330 ("*wrgsr_sp64"): Set insn subtype.
9331 ("*rdgsr_sp64"): Likewise.
9332 ("alignaddrsi_vis"): Likewise.
9333 ("alignaddrdi_vis"): Likewise.
9334 ("alignaddrlsi_vis"): Likewise.
9335 ("alignaddrldi_vis"): Likewise.
9336 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9337 ("fexpand_vis"): Likewise.
9338 ("fpmerge_vis"): Likewise.
9339 ("faligndata<VM64:mode>_vis"): Likewise.
9340 ("bshuffle<VM64:mode>_vis"): Likewise.
9341 ("cmask8<P:mode>_vis"): Likewise.
9342 ("cmask16<P:mode>_vis"): Likewise.
9343 ("cmask32<P:mode>_vis"): Likewise.
9344 ("fchksm16_vis"): Likewise.
9345 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9346 ("fmean16_vis"): Likewise.
9347 ("fp<plusminus_insn>64_vis"): Likewise.
9348 ("<plusminus_insn>v8qi3"): Likewise.
9349 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9350 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9351 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9352 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9353 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9354 ("*movqi_insn"): Likewise.
9355 ("*movhi_insn"): Likewise.
9356 ("*movsi_insn"): Likewise.
9357 ("movsi_pic_gotdata_op"): Likewise.
9358 ("*movdi_insn_sp32"): Likewise.
9359 ("*movdi_insn_sp64"): Likewise.
9360 ("movdi_pic_gotdata_op"): Likewise.
9361 ("*movsf_insn"): Likewise.
9362 ("*movdf_insn_sp32"): Likewise.
9363 ("*movdf_insn_sp64"): Likewise.
9364 ("*zero_extendhisi2_insn"): Likewise.
9365 ("*zero_extendqihi2_insn"): Likewise.
9366 ("*zero_extendqisi2_insn"): Likewise.
9367 ("*zero_extendqidi2_insn"): Likewise.
9368 ("*zero_extendhidi2_insn"): Likewise.
9369 ("*zero_extendsidi2_insn_sp64"): Likewise.
9370 ("ldfsr"): Likewise.
9371 ("prefetch_64"): Likewise.
9372 ("prefetch_32"): Likewise.
9373 ("tie_ld32"): Likewise.
9374 ("tie_ld64"): Likewise.
9375 ("*tldo_ldub_sp32"): Likewise.
9376 ("*tldo_ldub1_sp32"): Likewise.
9377 ("*tldo_ldub2_sp32"): Likewise.
9378 ("*tldo_ldub_sp64"): Likewise.
9379 ("*tldo_ldub1_sp64"): Likewise.
9380 ("*tldo_ldub2_sp64"): Likewise.
9381 ("*tldo_ldub3_sp64"): Likewise.
9382 ("*tldo_lduh_sp32"): Likewise.
9383 ("*tldo_lduh1_sp32"): Likewise.
9384 ("*tldo_lduh_sp64"): Likewise.
9385 ("*tldo_lduh1_sp64"): Likewise.
9386 ("*tldo_lduh2_sp64"): Likewise.
9387 ("*tldo_lduw_sp32"): Likewise.
9388 ("*tldo_lduw_sp64"): Likewise.
9389 ("*tldo_lduw1_sp64"): Likewise.
9390 ("*tldo_ldx_sp64"): Likewise.
9391 ("*mov<VM32:mode>_insn"): Likewise.
9392 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9393 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9395 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9397 * config/sparc/sparc.md ("type"): New insn type viscmp.
9398 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9400 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9401 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9402 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9403 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9405 ("n7_vis_logical_11cycle"): Likewise.
9406 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9407 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9408 * config/sparc/niagara.md ("niag_vis"): Likewise.
9409 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9410 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9412 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9414 * config/sparc/sparc.md: New instruction type `bmask'.
9415 (bmaskdi_vis): Use the `bmask' type.
9416 (bmasksi_vis): Likewise.
9417 * config/sparc/ultra3.md (us3_array): Likewise.
9418 * config/sparc/niagara7.md (n7_array): Likewise.
9419 * config/sparc/niagara4.md (n4_array): Likewise.
9420 * config/sparc/niagara2.md (niag2_vis): Likewise.
9421 (niag3_vis): Likewise.
9422 * config/sparc/niagara.md (niag_vis): Likewise.
9424 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9426 * ipa-comdats.c: Remove optimize check from gate.
9427 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9428 for functions not optimized.
9429 (ipa_fn_summary_read): Skip optimize check.
9430 (ipa_fn_summary_write): Likewise.
9431 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9433 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9435 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9436 (check_callers): Give up on caller which is not optimized.
9437 (inline_small_functions): Likewise.
9438 (ipa_inline): Do not give up when not optimizing.
9439 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9440 away unoptimizes cdtors.
9441 (whole_program_function_and_variable_visibility): Do
9442 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9443 * ipa.c (process_references): Do not check optimize.
9444 (symbol_table::remove_unreachable_nodes): Update optimize check.
9445 (set_writeonly_bit): Update optimize check.
9446 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9447 (pass_ipa_single_use::gate): Remove.
9449 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9451 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9452 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9453 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9454 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9455 permute_load, permute_store, adjust_extract, adjust_splat,
9456 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9457 replace_swap_with_copy, dump_swap_insn_table,
9458 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9459 recombine_lvx_pattern, recombine_stvx_pattern,
9460 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9461 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9462 to file rs6000-p8swap.c.
9463 * config/rs6000/rs6000-p8swap.c: New file.
9464 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9465 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9466 and rs6000*-*-* targets.
9468 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9470 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9472 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9474 * lto-wrapper.c (merge_and_complain): Do not merge
9475 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9476 fsigned_zeros, ftrapping_math, fwrapv.
9477 (append_compiler_options): Do not track these options.
9478 (append_linker_options): Likewie
9480 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9482 * cgraphunit.c (cgraph_node::finalize_function): When
9483 !flag_toplevel_reorde set no_reorder flag.
9484 (varpool_node::finalize_decl): Likewise.
9485 (symbol_table::compile): Drop no toplevel reorder path.
9487 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9489 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9490 edges; zero probability is not better than uninitialized.
9492 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9494 * asan.h (asan_sanitize_allocas_p): Declare.
9495 * asan.c (asan_sanitize_allocas_p): New function.
9496 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9497 (handle_builtin_alloca): Likewise.
9498 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9499 if !asan_sanitize_allocas_p.
9500 * params.def (asan-instrument-allocas): Add new option.
9501 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9502 * opts.c (common_handle_option): Disable allocas sanitization for
9505 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9507 * asan.c: Include gimple-fold.h.
9508 (get_last_alloca_addr): New function.
9509 (handle_builtin_stackrestore): Likewise.
9510 (handle_builtin_alloca): Likewise.
9511 (asan_emit_allocas_unpoison): Likewise.
9512 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9513 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9514 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9515 (instrument_builtin_call): Pass gimple iterator to
9516 get_mem_refs_of_builtin_call.
9517 (last_alloca_addr): New global.
9518 * asan.h (asan_emit_allocas_unpoison): Declare.
9519 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9520 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9521 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9522 if function calls alloca.
9523 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9524 * gimple-fold.h (replace_call_with_value): Declare.
9525 * internal-fn.c: Include asan.h.
9526 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9527 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9529 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9531 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9532 (C_SELFTEST_FLAGS): New.
9533 (CPP_SELFTEST_FLAGS): New.
9534 (SELFTEST_DEPS): New, from deps of s-selftest.
9535 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9536 (CPP_SELFTEST_DEPS): New.
9537 (selftest): Add dependency on s-selftest-c++.
9538 (s-selftest): Rename to...
9539 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9540 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9541 than SELFTEST_FLAGS.
9542 (selftest-gdb): Rename to...
9543 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9545 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9546 (selftest-valgrind): Rename to...
9547 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9549 (selftest-valgrind): Reintroduce as an alias for
9550 selftest-c-valgrind.
9551 (s-selftest-c++): New.
9552 (selftest-c++-gdb): New.
9553 (selftest-c++-valgrind): New.
9555 2017-07-06 Olivier Hainque <hainque@adacore.com>
9557 * gcc.c (process_command): When deciding if undefined variables
9558 should be ignored when processing specs, accept "gcc -v" as well.
9560 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9562 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9563 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9565 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9567 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9568 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9569 * config/arm/arm-tables.opt: Regenerate.
9570 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9572 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9574 2017-07-06 Carl Love <cel@us.ibm.com>
9576 * ChangeLog: Clean up from mid air collision
9578 2017-07-06 Carl Love <cel@us.ibm.com>
9580 * config/rs6000/rs6000-c.c: Add support for built-in functions
9581 vector signed int vec_subc (vector signed int, vector signed int);
9582 vector signed __int128 vec_subc (vector signed __int128,
9583 vector signed __int128);
9584 vector unsigned __int128 vec_subc (vector unsigned __int128,
9585 vector unsigned __int128);
9586 vector signed int vec_sube (vector signed int, vector signed int,
9588 vector unsigned int vec_sube (vector unsigned int,
9589 vector unsigned int,
9590 vector unsigned int);
9591 vector signed __int128 vec_sube (vector signed __int128,
9592 vector signed __int128,
9593 vector signed__int128);
9594 vector unsigned __int128 vec_sube (vector unsigned __int128,
9595 vector unsigned __int128,
9596 vector unsigned __int128);
9597 vector signed int vec_subec (vector signed int, vector signed int,
9599 vector unsigned int vec_subec (vector unsigned int,
9600 vector unsigned int,
9601 vector unsigned int);
9602 vector signed __int128 vec_subec (vector signed __int128,
9603 vector signed __int128,
9604 vector signed__int128);
9605 vector unsigned __int128 vec_subec (vector unsigned __int128,
9606 vector unsigned __int128,
9607 vector unsigned __int128);
9608 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9609 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9610 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9611 BU_ALTIVEC_OVERLOAD_X definitions.
9612 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9613 * doc/extend.texi: Update the built-in documentation file for the new
9616 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9619 * diagnostic-show-locus.c (layout::layout): Use start and finish
9620 spelling location for the start and finish of each range.
9621 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9622 Add unused aspect param.
9623 * input.c (expand_location_1): Add "aspect" param, and use it
9624 to access the correct part of the location.
9625 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9627 (expand_location_to_spelling_point): Likewise.
9628 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9629 param, and pass it to expand_location_1.
9631 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9633 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9634 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9635 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9636 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9637 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9638 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9639 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9640 _mm_maskz_getmant_ss): New intrinsics.
9641 (__builtin_ia32_getexpss128_mask): Changed to ...
9642 __builtin_ia32_getexpss128_round ... this.
9643 (__builtin_ia32_getexpsd128_mask): Changed to ...
9644 __builtin_ia32_getexpsd128_round ... this.
9645 * config/i386/i386-builtin-types.def
9646 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9647 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9648 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9649 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9650 __builtin_ia32_getmantss_mask_round): New builtins.
9651 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9652 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9653 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9654 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9655 * config/i386/sse.md
9656 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9657 avx512f_sgetexp<mode><mask_scalar_name>
9658 <round_saeonly_scalar_name> ... this.
9659 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9660 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9661 vgetexp<ssescalarmodesuffix>
9662 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9663 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9664 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9665 avx512f_vgetmant<mode><mask_scalar_name>
9666 <round_saeonly_scalar_name> ... this.
9667 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9668 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9669 vgetmant<ssescalarmodesuffix>
9670 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9671 %0<mask_scalar_operand4>, %1, %2
9672 <round_saeonly_scalar_mask_op4>, %3} ... this.
9673 * config/i386/subst.md (mask_scalar_operand4,
9674 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9675 round_saeonly_scalar_nimm_predicate): New subst attributes.
9677 2017-07-06 Julia Koval <julia.koval@intel.com>
9679 * config/i386/i386.c (ix86_erase_embedded_rounding):
9680 Remove code for old rounding pattern.
9682 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9684 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9686 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9688 * doc/sourcebuild.texi (Test Directives, Variants of
9689 dg-require-support): Add documentation for dg-require-stack-check.
9691 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9693 * config/i386/subst.md (mask_scalar, round_scalar,
9694 round_saeonly_scalar): New meta-templates.
9695 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9696 round_scalar_mask_operand3, round_scalar_mask_op3,
9697 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9698 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9699 round_saeonly_scalar_constraint,
9700 round_saeonly_scalar_prefix): New subst attribute.
9701 * config/i386/sse.md
9702 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9703 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9704 <round_scalar_name> ... this.
9705 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9706 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9707 <round_scalar_name> ... this.
9708 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9709 <sse>_vm<code><mode>3<mask_scalar_name>
9710 <round_saeonly_scalar_name> ... this.
9711 (v<plusminus_mnemonic><ssescalarmodesuffix>
9712 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9713 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9714 v<plusminus_mnemonic><ssescalarmodesuffix>
9715 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9716 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9717 (v<multdiv_mnemonic><ssescalarmodesuffix>
9718 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9719 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9720 v<multdiv_mnemonic><ssescalarmodesuffix>
9721 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9722 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9723 (v<maxmin_float><ssescalarmodesuffix>
9724 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9725 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9726 v<maxmin_float><ssescalarmodesuffix>
9727 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9728 %0<mask_scalar_operand3>, %1, %<iptr>2
9729 <round_saeonly_scalar_mask_op3>} ... this.
9731 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9733 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9734 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9736 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9737 Alan Hayward <alan.hayward@arm.com>
9738 David Sherwood <david.sherwood@arm.com>
9740 * combine.c (simplify_if_then_else): Remove "enum" before
9742 * compare-elim.c (can_eliminate_compare): Likewise.
9743 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9745 (aarch64_lookup_simd_builtin_type): Likewise.
9746 (aarch64_simd_builtin_type): Likewise.
9747 (aarch64_init_simd_builtin_types): Likewise.
9748 (aarch64_simd_expand_args): Likewise.
9749 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9751 (aarch64_reverse_mask): Likewise.
9752 (aarch64_simd_emit_reg_reg_move): Likewise.
9753 (aarch64_gen_adjusted_ldpstp): Likewise.
9754 (aarch64_ccmp_mode_to_code): Likewise.
9755 (aarch64_operands_ok_for_ldpstp): Likewise.
9756 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9757 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9759 (aarch64_min_divisions_for_recip_mul): Likewise.
9760 (aarch64_reassociation_width): Likewise.
9761 (aarch64_get_condition_code_1): Likewise.
9762 (aarch64_simd_emit_reg_reg_move): Likewise.
9763 (aarch64_simd_attr_length_rglist): Likewise.
9764 (aarch64_reverse_mask): Likewise.
9765 (aarch64_operands_ok_for_ldpstp): Likewise.
9766 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9767 (aarch64_gen_adjusted_ldpstp): Likewise.
9768 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9770 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9771 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9772 (arm_lookup_simd_builtin_type): Likewise.
9773 (arm_simd_builtin_type): Likewise.
9774 (arm_init_simd_builtin_types): Likewise.
9775 (arm_expand_builtin_args): Likewise.
9776 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9777 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9778 (ft32_setup_incoming_varargs): Likewise.
9779 (ft32_function_arg): Likewise.
9780 (ft32_function_arg_advance): Likewise.
9781 (ft32_pass_by_reference): Likewise.
9782 (ft32_arg_partial_bytes): Likewise.
9783 (ft32_valid_pointer_mode): Likewise.
9784 (ft32_addr_space_pointer_mode): Likewise.
9785 (ft32_addr_space_legitimate_address_p): Likewise.
9786 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9788 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9789 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9790 (iamcu_alignment): Likewise.
9791 (canonicalize_vector_int_perm): Likewise.
9792 (ix86_noce_conversion_profitable_p): Likewise.
9793 (ix86_mpx_bound_mode): Likewise.
9794 (ix86_operands_ok_for_move_multiple): Likewise.
9795 * config/microblaze/microblaze-protos.h
9796 (microblaze_expand_conditional_branch_reg): Likewise.
9797 * config/microblaze/microblaze.c
9798 (microblaze_expand_conditional_branch_reg): Likewise.
9799 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9801 (rs6000_reassociation_width): Likewise.
9802 (rs6000_invalid_binary_op): Likewise.
9803 (fusion_p9_p): Likewise.
9804 (emit_fusion_p9_load): Likewise.
9805 (emit_fusion_p9_store): Likewise.
9806 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9808 (riscv_hard_regno_mode_ok_p): Likewise.
9809 (riscv_address_insns): Likewise.
9810 (riscv_split_symbol): Likewise.
9811 (riscv_legitimize_move): Likewise.
9812 (riscv_function_value): Likewise.
9813 (riscv_hard_regno_nregs): Likewise.
9814 (riscv_expand_builtin): Likewise.
9815 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9816 (riscv_build_integer): Likewise.
9817 (riscv_split_integer): Likewise.
9818 (riscv_legitimate_constant_p): Likewise.
9819 (riscv_cannot_force_const_mem): Likewise.
9820 (riscv_regno_mode_ok_for_base_p): Likewise.
9821 (riscv_valid_base_register_p): Likewise.
9822 (riscv_valid_offset_p): Likewise.
9823 (riscv_valid_lo_sum_p): Likewise.
9824 (riscv_classify_address): Likewise.
9825 (riscv_legitimate_address_p): Likewise.
9826 (riscv_address_insns): Likewise.
9827 (riscv_load_store_insns): Likewise.
9828 (riscv_force_binary): Likewise.
9829 (riscv_split_symbol): Likewise.
9830 (riscv_force_address): Likewise.
9831 (riscv_legitimize_address): Likewise.
9832 (riscv_move_integer): Likewise.
9833 (riscv_legitimize_const_move): Likewise.
9834 (riscv_legitimize_move): Likewise.
9835 (riscv_address_cost): Likewise.
9836 (riscv_subword): Likewise.
9837 (riscv_output_move): Likewise.
9838 (riscv_canonicalize_int_order_test): Likewise.
9839 (riscv_emit_int_order_test): Likewise.
9840 (riscv_function_arg_boundary): Likewise.
9841 (riscv_pass_mode_in_fpr_p): Likewise.
9842 (riscv_pass_fpr_single): Likewise.
9843 (riscv_pass_fpr_pair): Likewise.
9844 (riscv_get_arg_info): Likewise.
9845 (riscv_function_arg): Likewise.
9846 (riscv_function_arg_advance): Likewise.
9847 (riscv_arg_partial_bytes): Likewise.
9848 (riscv_function_value): Likewise.
9849 (riscv_pass_by_reference): Likewise.
9850 (riscv_setup_incoming_varargs): Likewise.
9851 (riscv_print_operand): Likewise.
9852 (riscv_elf_select_rtx_section): Likewise.
9853 (riscv_save_restore_reg): Likewise.
9854 (riscv_for_each_saved_reg): Likewise.
9855 (riscv_register_move_cost): Likewise.
9856 (riscv_hard_regno_mode_ok_p): Likewise.
9857 (riscv_hard_regno_nregs): Likewise.
9858 (riscv_class_max_nregs): Likewise.
9859 (riscv_memory_move_cost): Likewise.
9860 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9861 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9862 (rl78_addr_space_address_mode): Likewise.
9863 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9865 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9866 (rs6000_reassociation_width): Likewise.
9867 (rs6000_invalid_binary_op): Likewise.
9868 (fusion_p9_p): Likewise.
9869 (emit_fusion_p9_load): Likewise.
9870 (emit_fusion_p9_store): Likewise.
9871 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9872 (ok_for_simple_move_operands): Likewise.
9873 (ok_for_simple_move_strict_operands): Likewise.
9874 (ok_for_simple_arith_logic_operands): Likewise.
9875 (visium_legitimize_reload_address): Likewise.
9876 (visium_select_cc_mode): Likewise.
9877 (output_cbranch): Likewise.
9878 (visium_split_double_move): Likewise.
9879 (visium_expand_copysign): Likewise.
9880 (visium_expand_int_cstore): Likewise.
9881 (visium_expand_fp_cstore): Likewise.
9882 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9883 (visium_function_arg): Likewise.
9884 (visium_function_arg_advance): Likewise.
9885 (visium_libcall_value): Likewise.
9886 (visium_setup_incoming_varargs): Likewise.
9887 (visium_legitimate_constant_p): Likewise.
9888 (visium_legitimate_address_p): Likewise.
9889 (visium_legitimize_address): Likewise.
9890 (visium_secondary_reload): Likewise.
9891 (visium_register_move_cost): Likewise.
9892 (visium_memory_move_cost): Likewise.
9893 (prepare_move_operands): Likewise.
9894 (ok_for_simple_move_operands): Likewise.
9895 (ok_for_simple_move_strict_operands): Likewise.
9896 (ok_for_simple_arith_logic_operands): Likewise.
9897 (visium_function_value_1): Likewise.
9898 (rtx_ok_for_offset_p): Likewise.
9899 (visium_legitimize_reload_address): Likewise.
9900 (visium_split_double_move): Likewise.
9901 (visium_expand_copysign): Likewise.
9902 (visium_expand_int_cstore): Likewise.
9903 (visium_expand_fp_cstore): Likewise.
9904 (visium_split_cstore): Likewise.
9905 (visium_select_cc_mode): Likewise.
9906 (visium_split_cbranch): Likewise.
9907 (output_cbranch): Likewise.
9908 (visium_print_operand_address): Likewise.
9909 * expmed.c (flip_storage_order): Likewise.
9910 * expmed.h (emit_cstore): Likewise.
9911 (flip_storage_order): Likewise.
9912 * genrecog.c (validate_pattern): Likewise.
9913 * hsa-gen.c (gen_hsa_addr): Likewise.
9914 * internal-fn.c (expand_arith_overflow): Likewise.
9915 * ira-color.c (allocno_copy_cost_saving): Likewise.
9916 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9917 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9918 (process_invariant_for_inheritance): Likewise.
9919 * lra-eliminations.c (move_plus_up): Likewise.
9920 * omp-low.c (lower_oacc_reductions): Likewise.
9921 * simplify-rtx.c (simplify_subreg): Likewise.
9922 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9923 (TARGET_CHKP_BOUND_MODE): Likewise..
9924 * targhooks.c (default_chkp_bound_mode): Likewise.
9925 (default_setup_incoming_vararg_bounds): Likewise.
9926 * targhooks.h (default_chkp_bound_mode): Likewise.
9927 (default_setup_incoming_vararg_bounds): Likewise.
9928 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9929 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9930 (have_whole_vector_shift): Likewise.
9931 * tree-vect-stmts.c (vectorizable_load): Likewise.
9932 * doc/tm.texi: Regenerate.
9934 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9936 Graceful degrade if Binutils PR21472 is not available.
9939 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9940 .rodata in flash test fails.
9941 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9942 * confgure: Regenerate.
9943 * config.in: Regenerate.
9944 * config/avr/avr.c (avr_asm_named_section)
9945 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9946 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9947 (avr_asm_init_sections): Same.
9949 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9951 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9952 (fma<VH:mode>4_intrinsic): Likewise.
9953 (*fmsub<VCVTF:mode>4): Likewise.
9954 (*fmsub<VH:mode>4_intrinsic): Likewise.
9956 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9959 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9960 Don't depend on "optimize > 0".
9961 (out_movhi_r_mr, out_movqi_mr_r): Same.
9962 (out_movhi_mr_r, out_movqi_r_mr): Same.
9963 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9964 io_address_operand on "optimize > 0".
9966 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9968 * tree-loop-distribution.c: Add general explanantion on the pass.
9969 (generate_loops_for_partition): Mark distributed loop.
9970 (pg_add_dependence_edges): New parameter. Handle alias data
9971 dependence specially and record it in the parameter if asked.
9972 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9973 (init_partition_graph_vertices, add_partition_graph_edge): New.
9974 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9975 (free_partition_graph_vdata, build_partition_graph): New.
9976 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9977 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9978 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9979 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9980 (version_for_distribution_p, finalize_partitions): New.
9981 (distribute_loop): Handle alias data dependence specially. Factor
9982 out loop fusion code as functions and call these functions.
9984 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9986 * tree-loop-distribution.c (classify_partition): New parameter and
9987 better handle reduction statement.
9988 (rdg_build_partitions): Revise comment.
9989 (distribute_loop): Compute statements in all partitions and pass it
9990 to classify_partition.
9992 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9994 * tree-loop-distribution.c (enum partition_type): New.
9995 (struct partition): New field type.
9996 (partition_merge_into): Add parameter. Update partition type.
9997 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9998 (build_rdg_partition_for_vertex): Compute partition type.
9999 (rdg_build_partitions): Dump partition type.
10000 (distribute_loop): Update calls to partition_merge_into.
10002 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10004 * tree-loop-distribution.c (struct ddr_hasher): New.
10005 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10007 (classify_partition): Call get_data_dependence.
10008 (pg_add_dependence_edges): Ditto.
10009 (distribute_loop): Release data dependence hash table.
10011 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10013 * tree-loop-distribution.c (ref_base_address): Delete.
10014 (similar_memory_accesses): Rename ...
10015 (share_memory_accesses): ... to this. Check if partitions access
10016 the same memory reference.
10017 (distribute_loop): Call share_memory_accesses.
10019 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10021 * tree-loop-distribution.c (struct partition): New field recording
10022 its data reference.
10023 (partition_alloc, partition_free): Init and release data refs.
10024 (partition_merge_into): Merge data refs.
10025 (build_rdg_partition_for_vertex): Collect data refs for partition.
10026 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10028 (distribute_loop): Remve data refs from vertice data of partition
10031 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10033 * tree-loop-distribution.c (params.h): Include header file.
10034 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10035 (datarefs_vec): New global var.
10036 (create_rdg_vertices): Use datarefs_vec directly.
10037 (free_rdg): Don't free data references.
10038 (build_rdg): Update use. Don't free data references.
10039 (distribute_loop): Compute global variable for data references.
10040 Bail out if there are too many data references.
10042 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10044 * tree-loop-distribution.c (loop_nest): New global var.
10045 (build_rdg): Use loop directly, rather than loop nest.
10046 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10048 (distribute_loop): Compute global variable loop nest. Update use.
10050 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10052 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10053 (partition_merge_into): New parameter. Dump reason for fusion.
10054 (distribute_loop): Update use of partition_merge_into.
10056 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10058 * tree-loop-distribution.c (bb_top_order_index): New.
10059 (bb_top_order_index_size, bb_top_order_cmp): New.
10060 (stmts_from_loop): Use topological order.
10061 (pass_loop_distribution::execute): Compute and release topological
10062 order for basic blocks.
10064 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10066 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10071 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10072 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10073 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10074 * internal-fn.def (LOOP_DIST_ALIAS): New.
10075 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10076 (fold_loop_internal_call): ... this.
10077 (vect_loop_dist_alias_call): New function.
10078 (set_uid_loop_bbs): Call fold_loop_internal_call.
10079 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10082 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10085 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10086 Require dead FLAGS_REG at the beginning of a peephole.
10088 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10091 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10092 arguments in the call to __builtin_ia32_sbb_u32.
10093 (_subborrow_u64): Swap _X and _Y arguments in the call to
10094 __builtin_ia32_sbb_u64.
10096 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10099 * tree-vrp.c (compare_assert_loc): Turn into a function template
10100 with stable template parameter. Only test if a->e is NULL,
10101 !a->e == !b->e has been verified already. Use e == NULL or
10102 e != NULL instead of e or ! e tests. If stable is true, don't use
10103 iterative_hash_expr, on the other side allow a or b or both NULL
10104 and sort the NULLs last.
10105 (process_assert_insertions): Sort using compare_assert_loc<false>
10106 instead of compare_assert_loc, later sort using
10107 compare_assert_loc<true> before calling process_assert_insertions_for
10108 in a loop. Use break instead of continue once seen NULL pointer.
10110 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10112 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10113 Cortex-R7 and Cortex-R8 processors.
10115 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10117 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10118 uninitialized while src is not.
10120 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10122 * common/config/arm/arm-common.c: Adjust include path for
10124 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10125 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10126 (arm-cpu-data.h): Likewise.
10127 (arm-cpu-cdata.h): Likewise.
10128 * config/arm/arm-cpu.h: Delete.
10129 * config/arm/arm-cpu-cdata.h: Delete.
10130 * config/arm/arm-cpu-data.h: Delete.
10132 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10134 * config/arm/arm-cpus.in (cortex-a55): New.
10135 (cortex-a75): Likewise.
10136 (cortex-a75.cortex-a55): Likewise.
10137 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10139 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10140 * config/arm/arm-cpu-cdata.h: Regenerate.
10141 * config/arm/arm-cpu-data.h: Regenerate.
10142 * config/arm/arm-cpu.h: Regenerate.
10143 * config/arm/arm-tables.opt: Regenerate.
10144 * config/arm/arm-tune.md: Regenerate.
10146 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10148 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10150 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10152 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10155 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10157 PR tree-optimization/81292
10158 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10159 full_string_p, also call adjust_related_strinfos if the adjustment
10160 is simple, otherwise invalidate related strinfos.
10162 2017-07-04 Martin Liska <mliska@suse.cz>
10165 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10166 newly created variable as DECL_IGNORED_P.
10168 2017-07-04 Martin Liska <mliska@suse.cz>
10171 * ipa-inline.c (inline_small_functions):
10172 Use xstrdup_for_dump.
10174 2017-07-04 Tom de Vries <tom@codesourcery.com>
10176 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10178 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10181 * config/darwin.c (darwin_function_switched_text_sections):
10184 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10186 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10188 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10190 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10192 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10194 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10195 min_profitable_iters, and th as inclusive lower bounds.
10196 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10197 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10198 for min_profitable_iters and min_profitable_estimate.
10199 (vect_transform_loop): Treat th as an inclusive lower bound.
10200 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10202 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10205 * config/darwin.c (darwin_function_switched_text_sections):
10206 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10207 in two pieces, and suppress the use of buf.
10209 2017-07-03 Nathan Sidwell <nathan@acm.org>
10211 * hash-table.h (hash_table_mod1): Fix indentation.
10213 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10215 PR middle-end/81290
10216 * predict.c (force_edge_cold): Be more careful about propagation
10218 * profile-count.h (profile_probability::guessed,
10219 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10221 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10223 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10225 * doc/invoke.texi (rcpc architecture extension): Document it.
10227 2017-07-03 Richard Biener <rguenther@suse.de>
10229 PR tree-optimization/60510
10230 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10231 the scalar reduction PHI and use it.
10232 (vectorizable_reduction): Properly guard the single_defuse_cycle
10233 path for non-SLP reduction chains where we cannot use it.
10234 Rework reduc_def/index and vector type deduction. Rework
10235 vector operand gathering during reduction op code-gen.
10236 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10237 chains dissolve the chain and leave it to non-SLP reduction
10240 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10242 * tree-data-ref.h (dr_alignment): Declare.
10243 * tree-data-ref.c (dr_alignment): New function.
10244 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10245 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10247 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10249 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10251 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10252 and base_misalignment fields.
10253 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10254 * tree-data-ref.c: Include builtins.h.
10255 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10256 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10257 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10258 * tree-vect-data-refs.c: Include tree-cfg.h.
10259 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10260 fields instead of calculating an alignment here.
10261 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10262 innermost_loop_behavior fields.
10264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10266 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10268 (DR_STEP_ALIGNMENT): New macro.
10269 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10270 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10271 (create_data_ref): Print it.
10272 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10273 to tell whether the step preserves vector (mis)alignment.
10274 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10275 Move the check for an integer step and generalise to all INTEGER_CST.
10276 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10277 Print the outer step alignment.
10279 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10281 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10282 with offset_alignment.
10283 (DR_ALIGNED_TO): Delete.
10284 (DR_OFFSET_ALIGNMENT): New macro.
10285 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10286 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10287 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10288 (create_data_ref): Likewise.
10289 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10290 (vect_analyze_data_refs): Likewise.
10291 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10292 creating dummy innermost behavior.
10294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10296 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10297 with a "innermost_loop_behavior *" and refeence tree.
10298 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10299 (create_data_ref): Update call accordingly.
10300 * tree-predcom.c (find_looparound_phi): Likewise.
10302 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10304 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10305 fields with dr_wrt_vec_loop.
10306 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10307 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10308 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10309 (vect_dr_behavior): New function.
10310 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10311 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10312 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10313 track whether the step preserves the misalignment.
10314 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10315 Use vect_dr_behavior.
10316 (vect_setup_realignment): Update call accordingly.
10317 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10318 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10319 call to vect_create_addr_base_for_vector_ref.
10320 (vect_create_cond_for_align_checks): Likewise.
10321 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10322 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10323 (vect_recog_mask_conversion_pattern): Likewise.
10324 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10325 (new_stmt_vec_info): Remove redundant zeroing.
10327 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10329 * common/config/arm/arm-common.c (arm_be8_option): New function.
10330 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10331 (ISA_ARMv6): Add isa_bit_be8.
10332 * config/arm/arm.h (arm_be8_option): Add prototype.
10333 (BE8_SPEC_FUNCTION): New define.
10334 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10335 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10336 (mlittle-endian): Similarly.
10337 (mbe8, mbe32): New options.
10338 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10339 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10341 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10343 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10345 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10347 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10348 (cleanup_tree_cfg_bb): Use it.
10349 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10351 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10353 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10356 * loop-doloop.c (add_test): Update profile.
10358 2017-07-03 Martin Liska <mliska@suse.cz>
10361 * sanopt.c (rewrite_usage_of_param): New function.
10362 (sanitize_rewrite_addressable_params): Likewise.
10363 (pass_sanopt::execute): Call rewrite_usage_of_param.
10365 2017-07-03 Richard Biener <rguenther@suse.de>
10367 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10368 back to using VIEW_CONVERT_EXPR.
10370 2017-07-03 Martin Liska <mliska@suse.cz>
10373 * doc/extend.texi: Document when a resolver function is
10374 generated for target_clones.
10376 2017-07-03 Martin Liska <mliska@suse.cz>
10378 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10379 and shadow memory of auto variables which are subject of
10380 use-after-scope sanitization.
10381 (asan_expand_mark_ifn): Add do set only when is_poison.
10383 2016-07-03 Richard Biener <rguenther@suse.de>
10385 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10387 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10388 (vectorizable_reduction): Perform reduction PHI creation when
10389 visiting a reduction PHI and adjust and simplify code generation
10390 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10391 (vect_transform_loop): Visit reduction PHIs.
10392 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10393 defs into the SLP tree.
10394 (vect_build_slp_tree): Reduction defs terminate the recursion.
10395 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10397 (vect_get_vec_defs_for_stmt_copy): Export.
10398 (vect_get_vec_defs): Likewise.
10399 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10401 (vect_get_vec_defs_for_stmt_copy): Declare.
10402 (vect_get_vec_defs): Likewise.
10404 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10406 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10407 parameter with a "loop" parameter and use it instead of the
10408 loop containing DR_STMT. Don't check simple_iv when doing
10409 BB analysis. Describe the two analysis modes in the comment.
10411 2017-07-03 Tom de Vries <tom@codesourcery.com>
10413 PR tree-optimization/69468
10414 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10415 (find_same_succ_bb): Handle ignore_edge_flags.
10417 2017-07-03 Tom de Vries <tom@codesourcery.com>
10419 PR tree-optimization/81192
10420 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10422 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10424 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10426 2017-07-03 Tom de Vries <tom@codesourcery.com>
10428 PR tree-optimization/81192
10429 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10430 BB_SAME_SUCC (bb) == NULL.
10432 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10434 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10437 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10439 * dumpfile.c: Include profile-count.h
10440 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10442 (insert_cond_bb): Update profile.
10443 * tree-cfg.h (insert_cond_bb): Update prototype.
10444 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10445 * tree-dump.c: Do not include tree-cfg.
10447 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10449 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10451 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10453 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10456 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10458 * tree-complex.c (expand_complex_div_wide): update profile.
10460 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10461 Alan Hayward <alan.hayward@arm.com>
10462 David Sherwood <david.sherwood@arm.com>
10464 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10465 (CORETYPES_H): New define.
10466 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10467 (insn-modes-inline.h, s-modes-inline-h): New rules.
10468 (generated_files): Add insn-modes-inline.h.
10469 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10470 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10471 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10472 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10473 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10474 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10475 (build/gencodes.o, build/genconditions.o): Likewise.
10476 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10477 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10478 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10479 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10480 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10481 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10482 * coretypes.h: Include everything up to real.h for generators.
10483 Include insn-modes.h first. Include wide-int-print.h after
10484 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10485 * machmode.h: Don't include insn-modes.h here.
10486 * function-tests.c: Remove includes of signop.h, machmode.h,
10487 double-int.h and wide-int.h.
10489 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10491 * optc-save-gen.awk: Likewise.
10492 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10493 * godump.c: Remove include of wide-int-print.h.
10494 * pretty-print.h: Likewise.
10495 * wide-int-print.cc: Likewise.
10496 * wide-int.cc: Likewise.
10497 * hash-map-tests.c: Remove include of signop.h.
10498 * hash-set-tests.c: Likewise.
10499 * rtl-tests.c: Likewise.
10500 * mkconfig.sh: Remove include of machmode.h.
10501 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10503 (emit_insn_modes_inline_h): ...this new function. Emit the code
10504 into an insn-modes-inline.h header file, adding appropriate
10505 include guards and end comments.
10506 (emit_insn_modes_c_header): Remove include of machmode.h.
10507 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10509 (main): Handle -i flag and call emit_insn_modes_inline_h when
10512 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10514 * tree-ssa-strlen.c (strinfo): Rename the length field to
10515 nonzero_chars. Add a full_string_p field.
10516 (compare_nonzero_chars, zero_length_string_p): New functions.
10517 (get_addr_stridx): Add an offset_out parameter.
10518 Use compare_nonzero_chars.
10519 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10520 (new_strinfo): Update after above changes to strinfo.
10521 (set_endptr_and_length): Set full_string_p.
10522 (get_string_length): Update after above changes to strinfo.
10523 (unshare_strinfo): Update call to new_strinfo.
10524 (maybe_invalidate): Likewise.
10525 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10526 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10527 as a uhwi instead of an shwi. Update after above changes to
10528 strinfo and new_strinfo.
10529 (zero_length_string): Assert that chainsi contains full strings.
10530 Use zero_length_string_p. Update call to new_strinfo.
10531 (adjust_related_strinfos): Update after above changes to strinfo.
10532 Copy full_string_p from origsi.
10533 (adjust_last_stmt): Use zero_length_string_p.
10534 (handle_builtin_strlen): Update after above changes to strinfo and
10535 new_strinfo. Install the lhs as the string length if the previous
10536 entry didn't describe a full string.
10537 (handle_builtin_strchr): Update after above changes to strinfo
10539 (handle_builtin_strcpy): Likewise.
10540 (handle_builtin_strcat): Likewise.
10541 (handle_builtin_malloc): Likewise.
10542 (handle_pointer_plus): Likewise.
10543 (handle_builtin_memcpy): Likewise. Track nonzero characters
10544 that aren't necessarily followed by a nul terminator.
10545 (handle_char_store): Likewise.
10547 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10549 PR tree-optimization/80769
10550 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10551 for malloc and calloc. Document the new invariant that all related
10552 strinfos have delayed lengths or none do.
10553 (verify_related_strinfos): Move earlier in file.
10554 (set_endptr_and_length): New function, split out from...
10555 (get_string_length): ...here. Also set the lengths of related
10557 (zero_length_string): Assert that chainsi has known (rather than
10559 (adjust_related_strinfos): Likewise.
10561 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10563 PR tree-optimization/81136
10564 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10565 assert that two references with the same misalignment have the same
10566 compile-time misalignment if those compile-time misalignments
10569 2017-07-01 Andi Kleen <ak@linux.intel.com>
10571 * print-tree.c (print_node): Print all attributes.
10573 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10575 * cfg.c (scale_bbs_frequencies): New function.
10576 * cfg.h (scale_bbs_frequencies): Declare it.
10577 * cfgloopanal.c (single_likely_exit): Cleanup.
10578 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10580 (scale_loop_profile): Likewise.
10581 (loop_version): Likewise.
10582 (create_empty_loop_on_edge): Update.
10583 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10584 scale_loop_frequencies, scale_loop_profile, loopify,
10585 loop_version): Update prototypes.
10586 * modulo-sched.c (sms_schedule): Update.
10587 * predict.c (unlikely_executed_edge_p): Also check probability.
10588 (probably_never_executed_edge_p): Fix typo.
10589 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10590 * tree-parloops.c (gen_parallel_loop): Update.
10591 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10592 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10593 * tree-ssa-loop-split.c (split_loop): Update.
10594 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10595 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10596 (vect_loop_versioning): Update.
10597 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10599 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10601 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10603 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10605 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10606 to keep profile consistent.
10608 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10610 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10611 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10612 * profile-count.h (max_safe_multiplier): Make unsigned.
10613 (profile_count::guessed_zero): New.
10615 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10617 * bb-reorder.c (fix_up_crossing_landing_pad,
10618 fix_crossing_conditional_branches): Use make_single_succ_edge
10619 to keep profile consistent.
10621 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10623 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10626 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10629 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10630 the right scopes, make sure cond_jump isn't preserved between multiple
10631 iterations. Search for fallthru edge whenever there are 3+ edges and
10632 use find_fallthru_edge for it.
10634 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10636 Patch by Alexander Monakov <amonakov@ispras.ru>
10637 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10638 probabilities consistently.
10640 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10642 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10643 emit_cmp_and_jump_insns.
10645 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10648 * tree-inline.c (expand_call_inline): Combine profile statuses.
10650 2017-06-30 Andrew Pinski <apinski@cavium.com>
10652 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10653 fold_stmt returned true.
10655 2017-06-30 Nathan Sidwell <nathan@acm.org>
10657 * ggc.h (empty_string): Delete.
10658 * cfgexpand.c (expand_asm_stmt): Use plain "".
10659 * optabs.c (expand_asm_memory_barrier): Likewise.
10660 * stringpool.c (empty_string): Delete.
10661 (digit_vector, digit_string): Delete.
10662 (ggc_alloc_string): Use plain "", don't optimize single digit
10663 strings. Use ggc_alloc_atomic.
10665 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10667 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10668 comparison set and one other set, use the cost of the non-comparison
10671 2017-06-30 Nathan Sidwell <nathan@acm.org>
10673 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10676 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10678 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10679 loops. Remove now unneeded calls to gimple_switch_set_label() that
10680 just set removed labels to NULL_TREE.
10682 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10684 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10685 (set_range_info): ...here. Only call set_range_info_raw if domain
10687 (set_nonzero_bits): Call set_range_info_raw.
10688 * tree-ssanames.h (set_range_info_raw): New.
10690 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10693 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10694 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10695 of nonimmediate_operand and <store_mask_constraint> instead of m
10696 for the input operand. For V8FI iterator, always split if input
10697 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10698 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10699 <store_mask_predicate> instead of register_operand and
10700 <store_mask_constraint> instead of v for the input operand. Make
10701 sure both operands aren't MEMs for if not <mask_applied>.
10703 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10705 * lto-wrapper.c (copy_file) Close both file descriptors before
10708 2017-06-30 Martin Liska <mliska@suse.cz>
10711 * multiple_target.c (create_dispatcher_calls): Make ifunc
10712 also for function that don't have calls or are not referenced.
10714 2017-06-30 Richard Biener <rguenther@suse.de>
10716 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10717 analyze the first scalar stmt. Move vector type computation
10718 for the BB case here from ...
10719 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10720 live operation processing in the SLP case properly.
10722 2017-06-30 Richard Biener <rguenther@suse.de>
10724 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10726 2017-06-30 Martin Liska <mliska@suse.cz>
10729 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10730 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10732 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10734 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10737 2017-06-30 Richard Biener <rguenther@suse.de>
10739 PR tree-optimization/81249
10740 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10741 condition reduction result to original scalar type.
10743 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10745 * profile-count.h (enum profile_quality): Fix typos and whitespace
10748 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10750 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10751 type for branch probabilities.
10753 2017-06-29 Julian Brown <julian@codesourcery.com>
10754 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10756 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10757 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10758 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10759 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10761 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10763 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10764 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10765 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10766 CC usage from generic code to here.
10767 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10768 CC usage into the target macros.
10770 2017-06-29 Maya Rashish <coypu@sdf.org>
10772 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10775 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10777 * arm/arm-builtins.c: Include profile-count.h
10778 * except.c (sjlj_emit_function_enter): Use
10779 profile_probability::unlikely.
10781 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10783 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10784 and tocrel_offset be pointer args rather than implicitly using
10786 (legitimate_constant_pool_address_p, rs6000_emit_move,
10787 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10788 tocrel_offset and use in toc_relative_expr_p call.
10789 (print_operand, print_operand_address): Use static tocrel_base_oac
10790 and tocrel_offset_oac.
10791 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10794 2017-06-29 Maya Rashish <coypu@sdf.org>
10796 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10798 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10800 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10801 objects, take into account only the alignment of 'op0' and 'mode1' if
10804 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10806 * ccmp.c (ccmp_tree_comparison_p): New function.
10807 (ccmp_candidate_p): Update to use above function.
10808 (get_compare_parts): New function.
10809 (expand_ccmp_next): Update to use new functions.
10810 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10812 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10813 take mode as argument.
10814 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10815 * expr.c (expand_expr_real_1): Pass mode as argument.
10817 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10819 * combine.c (combine_instructions): Print insns to dump_file, together
10822 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10824 * asan.c (asan_emit_stack_protection): Update.
10825 (create_cond_insert_point): Update.
10826 * auto-profile.c (afdo_propagate_circuit): Update.
10827 * basic-block.h (struct edge_def): Turn probability to
10828 profile_probability.
10829 (EDGE_FREQUENCY): Update.
10830 * bb-reorder.c (find_traces_1_round): Update.
10831 (better_edge_p): Update.
10832 (sanitize_hot_paths): Update.
10833 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10834 (make_single_succ_edge): Update.
10835 (check_bb_profile): Update.
10836 (dump_edge_info): Update.
10837 (update_bb_profile_for_threading): Update.
10838 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10839 probabilitycount to 0.
10840 * cfgbuild.c (compute_outgoing_frequencies): Update.
10841 * cfgcleanup.c (try_forward_edges): Update.
10842 (outgoing_edges_match): Update.
10843 (try_crossjump_to_edge): Update.
10844 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10845 (expand_gimple_tailcall): Update.
10846 (construct_init_block): Use make_single_succ_edge.
10847 (construct_exit_block): Use make_single_succ_edge.
10848 * cfghooks.c (verify_flow_info): Update.
10849 (redirect_edge_succ_nodup): Update.
10850 (split_edge): Update.
10851 (account_profile_record): Update.
10852 * cfgloopanal.c (single_likely_exit): Update.
10853 * cfgloopmanip.c (scale_loop_profile): Update.
10854 (set_zero_probability): Remove.
10855 (duplicate_loop_to_header_edge): Update.
10856 * cfgloopmanip.h (loop_version): Update prototype.
10857 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10858 (force_nonfallthru_and_redirect): Update.
10859 (update_br_prob_note): Update.
10860 (rtl_verify_edges): Update.
10861 (purge_dead_edges): Update.
10862 (rtl_lv_add_condition_to_bb): Update.
10863 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10864 * cgraphunit.c (init_lowered_empty_function): Update.
10865 (cgraph_node::expand_thunk): Update.
10866 * cilk-common.c: Include profile-count.h
10867 * dojump.c (inv): Remove.
10868 (jumpifnot): Update.
10869 (jumpifnot_1): Update.
10870 (do_jump_1): Update.
10872 (do_jump_by_parts_greater_rtx): Update.
10873 (do_compare_rtx_and_jump): Update.
10874 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10875 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10876 * dwarf2cfi.c: Include profile-count.h
10877 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10878 (sjlj_emit_dispatch_table): Likewise.
10879 * explow.c: Include profile-count.h
10880 * expmed.c (emit_store_flag_force): Update.
10881 (do_cmp_and_jump): Update.
10882 * expr.c (compare_by_pieces_d::generate): Update.
10883 (compare_by_pieces_d::finish_mode): Update.
10884 (emit_block_move_via_loop): Update.
10885 (store_expr_with_bounds): Update.
10886 (store_constructor): Update.
10887 (expand_expr_real_2): Update.
10888 (expand_expr_real_1): Update.
10889 * expr.h (try_casesi, try_tablejump): Update prototypes.
10890 * gimple-pretty-print.c (dump_probability): Update.
10891 (dump_profile): New.
10892 (dump_gimple_label): Update.
10893 (dump_gimple_bb_header): Update.
10894 * graph.c (draw_cfg_node_succ_edges): Update.
10895 * hsa-gen.c (convert_switch_statements): Update.
10896 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10897 (find_if_case_1): Update.
10898 (find_if_case_2): Update.
10899 * internal-fn.c (expand_arith_overflow_result_store): Update.
10900 (expand_addsub_overflow): Update.
10901 (expand_neg_overflow): Update.
10902 (expand_mul_overflow): Update.
10903 (expand_vector_ubsan_overflow): Update.
10904 * ipa-cp.c (good_cloning_opportunity_p): Update.
10905 * ipa-split.c (split_function): Use make_single_succ_edge.
10906 * ipa-utils.c (ipa_merge_profiles): Update.
10907 * loop-doloop.c (add_test): Update.
10908 (doloop_modify): Update.
10909 * loop-unroll.c (compare_and_jump_seq): Update.
10910 (unroll_loop_runtime_iterations): Update.
10911 * lra-constraints.c (lra_inheritance): Update.
10912 * lto-streamer-in.c (input_cfg): Update.
10913 * lto-streamer-out.c (output_cfg): Update.
10914 * mcf.c (adjust_cfg_counts): Update.
10915 * modulo-sched.c (sms_schedule): Update.
10916 * omp-expand.c (expand_omp_for_init_counts): Update.
10917 (extract_omp_for_update_vars): Update.
10918 (expand_omp_ordered_sink): Update.
10919 (expand_omp_for_ordered_loops): Update.
10920 (expand_omp_for_generic): Update.
10921 (expand_omp_for_static_nochunk): Update.
10922 (expand_omp_for_static_chunk): Update.
10923 (expand_cilk_for): Update.
10924 (expand_omp_simd): Update.
10925 (expand_omp_taskloop_for_outer): Update.
10926 (expand_omp_taskloop_for_inner): Update.
10927 * omp-simd-clone.c (simd_clone_adjust): Update.
10928 * optabs.c (expand_doubleword_shift): Update.
10929 (expand_abs): Update.
10930 (emit_cmp_and_jump_insn_1): Update.
10931 (expand_compare_and_swap_loop): Update.
10932 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10933 * predict.c (predictable_edge_p): Update.
10934 (edge_probability_reliable_p): Update.
10935 (set_even_probabilities): Update.
10936 (combine_predictions_for_insn): Update.
10937 (combine_predictions_for_bb): Update.
10938 (propagate_freq): Update.
10939 (estimate_bb_frequencies): Update.
10940 (force_edge_cold): Update.
10941 * profile-count.c (profile_count::dump): Add missing space into dump.
10942 (profile_count::debug): Add newline.
10943 (profile_count::differs_from_p): Explicitly convert to unsigned.
10944 (profile_count::stream_in): Update.
10945 (profile_probability::dump): New member function.
10946 (profile_probability::debug): New member function.
10947 (profile_probability::differs_from_p): New member function.
10948 (profile_probability::differs_lot_from_p): New member function.
10949 (profile_probability::stream_in): New member function.
10950 (profile_probability::stream_out): New member function.
10951 * profile-count.h (profile_count_quality): Rename to ...
10952 (profile_quality): ... this one.
10953 (profile_probability): New.
10954 (profile_count): Update.
10955 * profile.c (compute_branch_probabilities): Update.
10956 * recog.c (peep2_attempt): Update.
10957 * sched-ebb.c (schedule_ebbs): Update.
10958 * sched-rgn.c (find_single_block_region): Update.
10959 (compute_dom_prob_ps): Update.
10960 (schedule_region): Update.
10961 * sel-sched-ir.c (compute_succs_info): Update.
10962 * stmt.c (struct case_node): Update.
10963 (do_jump_if_equal): Update.
10964 (get_outgoing_edge_probs): Update.
10965 (conditional_probability): Update.
10966 (emit_case_dispatch_table): Update.
10967 (expand_case): Update.
10968 (expand_sjlj_dispatch_table): Update.
10969 (emit_case_nodes): Update.
10970 * targhooks.c: Update.
10971 * tracer.c (better_p): Update.
10972 (find_best_successor): Update.
10973 * trans-mem.c (expand_transaction): Update.
10974 * tree-call-cdce.c: Update.
10975 * tree-cfg.c (gimple_split_edge): Upate.
10976 (move_sese_region_to_fn): Upate.
10977 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10978 * tree-eh.c (lower_resx): Upate.
10979 (cleanup_empty_eh_move_lp): Upate.
10980 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10981 * tree-inline.c (copy_edges_for_bb): Update.
10982 (copy_cfg_body): Update.
10983 * tree-parloops.c (gen_parallel_loop): Update.
10984 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10985 (gimple_gen_time_profiler): Update.
10986 * tree-ssa-dce.c (remove_dead_stmt): Update.
10987 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10988 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10989 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10990 (unloop_loops): Update.
10991 (try_peel_loop): Update.
10992 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10993 * tree-ssa-loop-split.c (connect_loops): Update.
10994 (split_loop): Update.
10995 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10996 (hoist_guard): Update.
10997 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10998 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10999 (value_replacement): Update.
11000 * tree-ssa-reassoc.c (branch_fixup): Update.
11001 * tree-ssa-tail-merge.c (replace_block_by): Update.
11002 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11003 (create_edge_and_update_destination_phis): Update.
11004 (compute_path_counts): Update.
11005 (recompute_probabilities): Update.
11006 (update_joiner_offpath_counts): Update.
11007 (freqs_to_counts_path): Update.
11008 (duplicate_thread_path): Update.
11009 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11010 (struct switch_conv_info): Update.
11011 (gen_inbound_check): Update.
11012 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11013 (vect_do_peeling): Update.
11014 (vect_loop_versioning): Update.
11015 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11016 (optimize_mask_stores): Update.
11017 * ubsan.c (ubsan_expand_null_ifn): Update.
11018 * value-prof.c (gimple_divmod_fixed_value): Update.
11019 (gimple_divmod_fixed_value_transform): Update.
11020 (gimple_mod_pow2): Update.
11021 (gimple_mod_pow2_value_transform): Update.
11022 (gimple_mod_subtract): Update.
11023 (gimple_mod_subtract_transform): Update.
11024 (gimple_ic): Update.
11025 (gimple_stringop_fixed_value): Update.
11026 (gimple_stringops_transform): Update.
11027 * value-prof.h: Update.
11029 2017-06-29 Carl Love <cel@us.ibm.com>
11031 * config/rs6000/rs6000-c.c: Add support for built-in functions
11032 vector signed int vec_signed (vector float);
11033 vector signed long long vec_signed (vector double);
11034 vector signed int vec_signed2 (vector double, vector double);
11035 vector signed int vec_signede (vector double);
11036 vector signed int vec_signedo (vector double);
11037 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11038 instruction generator.
11039 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11040 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11041 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11043 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11044 vunsignede_v2df): Add define_expands.
11045 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11046 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11047 VEC_UNSIGNEDO): Add definitions.
11048 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11049 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11050 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11051 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11052 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11053 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11054 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11055 vec_unsignede, vec_unsignedo): Add builtin defines.
11056 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11058 * doc/extend.texi: Update the built-in documentation file for the
11059 new built-in functions.
11061 2017-06-29 Richard Biener <rguenther@suse.de>
11063 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11064 reduction chains to LOOP_VINFO_REDUCTIONS.
11065 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11066 SLP reductions after processing reduction chains.
11068 2017-06-29 Nathan Sidwell <nathan@acm.org>
11070 * builtins.c (fold_builtin_FUNCTION): Use
11071 lang_hooks.decl_printable_name.
11073 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11075 PR middle-end/81194
11076 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11077 with only one label.
11078 * stmt.c (expand_case): Assert NCASES is greater than one.
11080 2017-06-29 Richard Biener <rguenther@suse.de>
11082 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11084 (group_case_labels): Likewise.
11085 (find_taken_edge): Push sanity checking on val to workers...
11086 (find_taken_edge_cond_expr): ... here
11087 (find_taken_edge_switch_expr): ... and here, handle cases
11088 with just a default label.
11089 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11090 (group_case_labels): Likewise.
11091 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11092 group_case_labels does anything cleanup the CFG again.
11094 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11096 PR tree-optimization/81196
11097 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11098 exit condition comparing two IVs.
11100 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11102 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11103 profile to the dummy entry at the end of the list of architectures.
11104 * config/arm/arm-cpu-cdata.h: Regenerated.
11106 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11107 Michael Collison <michael.collison@arm.com>
11110 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11112 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11113 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11114 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11115 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11116 of shift when the shift amount is masked with constant equal to
11117 the size of the mode.
11118 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11121 2017-06-29 Martin Liska <mliska@suse.cz>
11123 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11125 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11128 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11131 (edge_growth_cache_entry::edge_growth_cache_entry): New
11133 (reset_edge_growth_cache): Update to use constructor.
11135 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11137 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11138 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11139 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11141 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11143 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11144 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11146 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11148 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11149 (*-linux-uclibc*): Add t-uclibc tmake_file.
11150 * config/t-musl: New.
11151 * config/t-uclibc: New.
11153 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11155 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11157 (gen_comm_data): Emit architectural setting of arch_prof.
11158 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11160 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11161 (armv8-m.base, armv8-m.main): Likewise.
11162 * arm-protos.h (arm_build_target): Add profile field.
11163 (arch_option): Likewise.
11164 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11166 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11167 arm_active_target.profile.
11169 2017-06-28 Richard Biener <rguenther@suse.de>
11171 PR middle-end/81227
11172 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11173 TYPE_OVERFLOW_WRAPS.
11174 * match.pd (negate_expr_p): Likewise.
11175 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11176 fold_build2, not fold_binary.
11178 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11180 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11181 Convert memory address to Pmode.
11182 (aarch64_print_operand): Assert MEM operands are always Pmode.
11184 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11187 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11188 Remove redundant if.
11189 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11190 * config/arm/aarch-common-protos.h
11191 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11192 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11194 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11197 * multiple_target.c (create_dispatcher_calls): Set the default
11198 clone to be static, not public.
11200 2017-06-28 Richard Biener <rguenther@suse.de>
11202 * tree-vect-loop.c (vectorizable_reduction): Move special
11203 cond reduction IV var creation ...
11204 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11205 parameter. Use STMT_VINFO_VECTYPE.
11206 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11209 2017-06-28 Martin Liska <mliska@suse.cz>
11212 * ipa-visibility.c (non_local_p): Handle visibility.
11214 2017-06-28 Martin Liska <mliska@suse.cz>
11217 * common.opt: Add IntegerRange to various options.
11218 * opt-functions.awk (integer_range_info): New function.
11219 * optc-gen.awk: Add integer_range_info to cl_options struct.
11220 * opts-common.c (decode_cmdline_option): Handle
11221 CL_ERR_INT_RANGE_ARG.
11222 (cmdline_handle_error): Likewise.
11223 * opts.c (print_filtered_help): Show valid interval in
11224 when --help is provided.
11225 * opts.h (struct cl_option): Add range_min and range_max fields.
11226 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11228 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11230 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11231 (x * C EQ/NE y * C): New transformation.
11233 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11235 * genmultilib (combination_space): Accept '+' in option names.
11237 2017-06-28 Martin Liska <mliska@suse.cz>
11240 * asan.c (instrument_derefs): Bail out inner references
11241 that are hard register variables.
11243 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11246 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11247 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11249 2017-06-28 Richard Biener <rguenther@suse.de>
11251 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11252 (vect_get_slp_defs): Adjust.
11253 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11255 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11257 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11258 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11259 (vectorizable_reduction): Adjust.
11260 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11262 (vect_get_slp_defs): Likewise.
11263 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11264 (vectorizable_bswap): Adjust.
11265 (vectorizable_call): Likewise.
11266 (vectorizable_conversion): Likewise.
11267 (vectorizable_assignment): Likewise.
11268 (vectorizable_shift): Likewise.
11269 (vectorizable_operation): Likewise.
11270 (vectorizable_store): Likewise.
11271 (vectorizable_condition): Likewise.
11272 (vectorizable_comparison): Likewise.
11274 2017-06-28 Michael Collison <michael.collison@arm.com>
11277 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11279 (arm_gen_movmemqi): Removed unused variable 'i'.
11280 Convert 'for' loop into 'while' loop.
11281 (arm_expand_prologue): Remove last unnecessary set of insn.
11282 (thumb_pop): Remove unused variable 'pushed_words'.
11283 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11285 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11287 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11288 * config/s390/s390-protos.h: Add prototype of
11289 s390_rel_address_ok_p.
11290 * config/s390/s390.c (s390_got_symbol): New function.
11291 (s390_rel_address_ok_p): New function.
11292 (legitimize_pic_address): Use s390_rel_address_ok_p.
11293 (s390_load_got): Use s390_got_symbol.
11294 (s390_option_override): Issue error if
11295 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11296 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11298 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11300 2017-06-27 Andrew Pinski <apinski@cavium.com>
11302 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11303 (X * copysign (1.0, X)): New pattern.
11304 (X * copysign (1.0, -X)): New pattern.
11305 (copysign (-1.0, CST)): New pattern.
11307 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11309 * genmultilib (combination_space): Remove variable.
11310 Validate reuse rules against regular expression for any sequence
11311 of multilib options in any order.
11313 2017-06-27 Michael Collison <michael.collison@arm.com>
11315 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11316 call aarch64_split_simd_combine.
11317 * (aarch64_combine_internal<mode>): Delete pattern.
11318 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11319 Allow register and subreg operands.
11321 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11323 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11324 specific need, just fallback on defaults.
11325 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11327 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11328 Olivier Hainque <hainque@adacore.com>
11330 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11332 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11333 targets. Pick a default if no particular attempt applied.
11334 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11337 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11339 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11340 (x86_64-wrs-vxworks7): Likewise.
11342 2017-06-27 Marek Polacek <polacek@redhat.com>
11345 * ubsan.c (instrument_null): Check get_base_address's result for null.
11347 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11349 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11351 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11353 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11354 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11355 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11356 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11357 New function types.
11358 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11359 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11360 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11361 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11362 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11363 BUILT_IN_FEUPDATEENV): New builtins.
11364 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11365 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11366 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11367 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11369 (builtin_structptr_types): Adjust size.
11370 * tree.c (builtin_structptr_types): Add four entries.
11372 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11373 Olivier Hainque <hainque@adacore.com>
11375 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11376 (TLS_SYM): New local macro, forcing reference to __tls__ on
11377 link command lines for VxWorks 7 RTPs, triggering initialization
11379 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11380 OS features TLS support, true for RTPs on VxWorks 7.
11381 * config/vxworks.c (vxworks_override_options): Setup emutls
11384 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11386 * predict.c (test_prediction_value_range): Use -1U instead of -1
11387 to avoid narrowing conversion warning.
11388 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11389 to avoid narrowing conversion warning.
11390 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11392 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11394 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11396 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11397 64bit configurations.
11398 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11399 (SIZE_TYPE): Likewise.
11400 * config/vxworks.c (vxworks_emutls_var_fields): Use
11401 long_unsigned_type_node instead of unsigned_type_node as the offset
11402 field type, which is "pointer" mode in emutls.c.
11404 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11407 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11409 PR middle-end/81207
11410 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11411 gimple_vuse copying separately from gimple_vdef copying.
11413 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11415 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11418 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11419 Olivier Hainque <hainque@adacore.com>
11421 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11422 for all vxworks7 targets.
11423 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11424 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11425 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11426 variations for VX6/VX7 and 32/64bits later on in ...
11427 (VXWORKS_LIB_SPEC): Leverage new macros.
11428 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11429 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11431 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11433 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11434 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11436 2017-06-26 Carl Love <cel@us.ibm.com>
11438 * config/rs6000/rs6000-c.c: Add support for built-in functions
11439 vector bool char vec_reve (vector bool char);
11440 vector signed char vec_reve (vector signed char);
11441 vector unsigned char vec_reve (vector unsigned char);
11442 vector bool int vec_reve (vector bool int);
11443 vector signed int vec_reve (vector signed int);
11444 vector unsigned int vec_reve (vector unsigned int);
11445 vector bool long long vec_reve (vector bool long long);
11446 vector signed long long vec_reve (vector signed long long);
11447 vector unsigned long long vec_reve (vector unsigned long long);
11448 vector bool short vec_reve (vector bool short);
11449 vector signed short vec_reve (vector signed short);
11450 vector double vec_reve (vector double);
11451 vector float vec_reve (vector float);
11452 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11453 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11454 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11455 (altivec_vreve): New pattern.
11456 * config/rs6000/altivec.h (vec_reve): New define.
11457 * doc/extend.texi (vec_rev): Update the built-in documentation file
11458 for the new built-in functions.
11460 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11462 PR tree-optimization/71815
11463 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11465 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11467 (slsr_process_phi): Likewise.
11468 (replace_uncond_cands_and_profitable_phis): Don't replace a
11469 multiply candidate with a stride of 1 (copy or cast).
11470 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11472 (lowest_cost_path): Likewise.
11473 (total_savings): Likewise.
11475 2017-06-26 Richard Biener <rguenther@suse.de>
11478 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11479 Use def_builtin_pure for all gather builtins.
11481 2017-06-26 Richard Biener <rguenther@suse.de>
11483 PR tree-optimization/81203
11484 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11485 non-dominating BBs.
11487 2017-06-26 Marek Polacek <polacek@redhat.com>
11490 * doc/invoke.texi: Document -Wmultistatement-macros.
11492 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11494 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11495 arm_neon_ok_no_float_abi effective target.
11497 2017-06-26 Richard Biener <rguenther@suse.de>
11499 PR tree-optimization/80928
11500 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11501 (copy_bbs): Set BB_DUPLICATED flag early.
11502 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11504 (execute_on_shrinking_pred): Likewise.
11505 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11506 BB_DUPLICATED blocks.
11507 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11508 iterate over all PHIs considering removal of *gsi.
11510 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11512 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11515 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11517 * config/rs6000/rs6000-string.c: (expand_block_clear,
11518 do_load_for_compare, select_block_compare_mode,
11519 compute_current_alignment, expand_block_compare,
11520 expand_strncmp_align_check, expand_strn_compare,
11521 expand_block_move, rs6000_output_load_multiple)
11522 Move functions related to string/block move/compare
11523 to a separate file.
11524 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11525 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11526 for this function which is now used in two files.
11527 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11528 * config.gcc: Add rs6000-string.o to extra_objs for
11529 targets powerpc*-*-* and rs6000*-*-*.
11531 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11534 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11535 32-bit, since indexed is not valid for DImode.
11536 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11537 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11538 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11539 (define_peephole2 for Altivec d-form store): Likewise.
11542 * multiple_target.c (create_dispatcher_calls): Only create the
11543 dispatcher call if the function is the default clone of a
11544 versioned function.
11546 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11548 PR middle-end/80902
11549 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11550 a call, force the call to not be a tail call.
11552 2017-06-23 Jeff Law <law@redhat.com>
11554 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11557 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11559 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11560 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11561 handling for early expansion of vector shifts (sl,sr,sra,rl).
11562 (builtin_function_type): Add vector shift right instructions
11563 to the unsigned argument list.
11565 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11567 rtl-optimizatoin/79286
11568 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11569 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11570 trap. PIC register plus a const unspec without offset can never trap.
11572 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11574 * tree.h (builtin_structptr_type): New type.
11575 (builtin_structptr_types): Declare new array.
11576 * tree.c (builtin_structptr_types): New array.
11577 (free_lang_data, build_common_tree_nodes): Use it.
11579 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11582 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11585 2017-06-22 Matt Turner <mattst88@gmail.com>
11587 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11588 Lake models to skylake case. Assume skylake for unknown
11589 models with clflushopt.
11591 2017-06-22 Jeff Law <law@redhat.com>
11593 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11594 frame sizes that do not satisfy aarch64_uimm12_shift.
11596 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11598 * profile-count.h (apply_probability,
11599 apply_scale, probability_in): Fix checks for zero.
11601 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11603 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11604 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11606 2016-06-22 Richard Biener <rguenther@suse.de>
11608 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11609 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11610 REDUC_MAX_EXPR support.
11611 (vectorizable_reduction): Likewise.
11612 (vect_create_epilog_for_reduction): Likewise.
11614 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11616 * match.pd (A / (1 << B) -> A >> B): New.
11617 * generic-match-head.c: Include optabs-tree.h.
11618 * gimple-match-head.c: Likewise.
11619 * optabs-tree.h (target_supports_op_p): New.
11620 * optabs-tree.c (target_supports_op_p): New.
11622 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11624 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11625 $gcc_cv_ld --help output.
11626 (gcc_cv_ld_demangle): Likewise.
11627 (gcc_cv_ld_eh_frame_hdr): Likewise.
11628 (gcc_cv_ld_pie): Likewise.
11629 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11630 (gcc_cv_ld_buildid): Likewise.
11631 (gcc_cv_ld_sysroot): Likewise.
11632 (ld_bndplt_support): Likewise.
11633 (ld_pushpopstate_support): Likewise.
11634 * configure: Regenerate.
11635 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11637 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11640 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11641 operands indexes to avoid gap between operands and match_dups.
11643 2017-06-21 Andrew Pinski <apinski@cavium.com>
11645 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11646 Increment Arith_shift and Arith_shift_reg by 1.
11647 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11649 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11650 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11651 (aarch64_strip_extend): Add new argument and test for it.
11652 (aarch64_cheap_mult_shift_p): New function.
11653 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11654 add a cost if it is true.
11655 Update calls to aarch64_strip_extend.
11656 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11658 2017-06-21 Andrew Pinski <apinski@cavium.com>
11660 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11662 (thunderxt88): Likewise.
11663 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11664 (thunderx_prefetch_tune): New variable.
11665 (thunderx2t99_prefetch_tune): Update for the correct values.
11666 (thunderxt88_tunings): New variable.
11667 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11668 generic_prefetch_tune.
11669 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11671 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11673 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11674 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11675 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11676 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11677 (aarch64_atomic_cas<mode>, GPI): Likewise.
11679 2017-06-21 Martin Liska <mliska@suse.cz>
11681 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11682 statements on cold and hot labels.
11683 * predict.c (tree_estimate_probability_bb): Remove the
11684 prediction from this place.
11686 2017-06-21 Martin Liska <mliska@suse.cz>
11688 PR tree-optimization/79489
11689 * gimplify.c (maybe_add_early_return_predict_stmt): New
11691 (gimplify_return_expr): Call the function.
11692 * predict.c (tree_estimate_probability_bb): Remove handling
11694 * predict.def: Update comment about early return predictor.
11695 * gimple-predict.h (is_gimple_predict): New function.
11696 * predict.def: Change default value of early return to 66.
11697 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11699 * passes.def: Put pass_strip_predict_hints to the beginning of
11702 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11704 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11705 FUNCTION_DECL declarations.
11706 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11708 (dwaf2out_decl): Likewise.
11709 * godump.c (go_early_global_decl): Skip call to the real debug hook
11710 for FUNCTION_DECL declarations.
11711 * passes.c (rest_of_decl_compilation): Skip call to the
11712 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11713 -fdump-go-spec is passed.
11715 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11717 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11718 Reorder for compactness.
11719 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11720 (def_builtin_pure, def_builtin_pure2): New functions.
11721 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11723 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11725 * match.pd (nop_convert): New predicate.
11726 ((A +- CST1) +- CST2): Allow some NOP conversions.
11728 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11731 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11732 with ctors/dtors if GOVD_SHARED is set.
11734 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11736 * config/aarch64/aarch64.md (movti_aarch64):
11737 Emit mov rather than orr.
11738 (movtf_aarch64): Likewise.
11739 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11740 Emit mov rather than orr.
11742 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11744 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11745 Swap alternatives, make integer dup more expensive.
11747 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11749 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11750 Return true for non-tls symbols.
11752 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11754 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11755 (cortex-a75): Likewise.
11756 (cortex-a75.cortex-a55): Likewise.
11757 * config/aarch64/aarch64-tune.md: Regenerate.
11758 * doc/invoke.texi (-mtune): Document new values for -mtune.
11760 2017-06-21 Tom de Vries <tom@codesourcery.com>
11762 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11763 stack_size feature.
11764 (Effective-Target Keywords, Other attributes): Suggest using
11765 dg-add-options stack_size feature to get stack limit in stack_size
11766 effective target documentation.
11768 2017-06-21 Julian Brown <julian@codesourcery.com>
11769 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11771 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11772 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11773 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11775 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11776 attribute type list for neon_multiply.
11777 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11778 attribute type list for neon_multiply.
11779 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11780 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11781 attribute type list for neon_multiply.
11782 * config/arm/types.md (crypto_pmull): Add.
11783 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11784 attribute type list.
11786 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11788 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11791 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11793 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11794 to make sure not to dereference a NULL cost_classes_ptr pointer.
11796 2017-06-20 Carl Love <cel@us.ibm.com>
11798 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11799 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11800 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11801 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11802 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11803 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11804 VMULOSW): New enum "unspec" values.
11805 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11806 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11807 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11808 altivec_vmulosw): New patterns.
11809 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11810 VMULOSW): Add definitions.
11812 2017-06-20 Julia Koval <julia.koval@intel.com>
11814 * config/i386/i386.c: Fix rounding expand for new pattern.
11815 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11817 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11819 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11820 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11822 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11824 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11827 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11829 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11830 architecture, then by implementer ID.
11831 * config/aarch64/aarch64-tune.md: Regenerate.
11833 2017-06-20 Richard Biener <rguenther@suse.de>
11835 PR middle-end/81097
11836 * fold-const.c (split_tree): Fold to type before negating.
11838 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11840 * diagnostic-show-locus.c
11841 (selftest::test_fixit_deletion_affecting_newline): New function.
11842 (selftest::diagnostic_show_locus_c_tests): Call it.
11844 2017-06-20 Andreas Schwab <schwab@suse.de>
11847 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11850 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11852 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11853 __ARM_FEATURE_COPROC according to support.
11855 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11857 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11858 Rewritten to avoid overflow for > 32-bit pointers.
11861 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11862 by removing enum keyword.
11863 (ubsan_type_descriptor): Likewise. Formatting fix.
11866 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11867 splitter): Require TARGET_SSE2 in the condition.
11869 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11872 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11873 for doing vector set of SFmode on ISA 3.0.
11874 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11875 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11877 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11878 SFmode value into a V4SF variable that was extracted from another
11879 V4SF variable without converting the element to double precision
11880 and back to single precision vector format.
11881 (vsx_insert_extract_v4sf_p9_2): Likewise.
11883 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11885 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11886 in UWHI to avoid undefined overflow.
11889 * ubsan.h (enum ubsan_encode_value_phase): New.
11890 (ubsan_encode_value): Change second argument to
11891 enum ubsan_encode_value_phase with default value of
11892 UBSAN_ENCODE_VALUE_GENERIC.
11893 * ubsan.c (ubsan_encode_value): Change second argument to
11894 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11895 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11896 create_tmp_var_raw instead of create_tmp_var and use a
11898 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11899 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11900 ubsan_encode_value callers.
11903 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11904 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11905 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11907 2017-06-19 Richard Biener <rguenther@suse.de>
11909 PR middle-end/81118
11910 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11911 estimates if we changed anything.
11913 2017-06-19 Richard Biener <rguenther@suse.de>
11915 PR tree-optimization/80887
11916 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11917 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11918 simplified lookups, then reset mprts_hook.
11919 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11921 (try_to_simplify): Likewise.
11923 2017-06-19 Martin Liska <mliska@suse.cz>
11926 * gimplify.c (gimplify_switch_expr):
11927 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11929 2017-06-19 Martin Liska <mliska@suse.cz>
11931 * doc/install.texi: Document that PGO runs in 4 stages.
11933 2017-06-19 Martin Liska <mliska@suse.cz>
11936 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11937 to dispatcher function name.
11938 * multiple_target.c (replace_function_decl): New function.
11939 (create_dispatcher_calls): Redirect both edges and references.
11941 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11943 * profile-count.c (profile_count::dump): Dump quality.
11944 (profile_count::differs_from_p): Update for unsigned val.
11945 * profile-count.h (profile_count_quality): New enum.
11946 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11948 2017-06-19 Richard Biener <rguenther@suse.de>
11950 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11951 struct function as arg.
11952 (estimate_numbers_of_iterations): Export overload with loop arg.
11953 (free_numbers_of_iterations_estimates_loop): Use an overload of
11954 free_numbers_of_iterations_estimates instead.
11955 * tree-cfg.c (remove_bb): Adjust.
11956 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11957 * tree-parloops.c (gen_parallel_loop): Likewise.
11958 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11960 (tree_unroll_loops_completely): Likewise.
11961 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11962 Use an overload instead and export.
11963 (estimated_loop_iterations): Adjust.
11964 (max_loop_iterations): Likewise.
11965 (likely_max_loop_iterations): Likewise.
11966 (estimate_numbers_of_iterations): Take struct function as arg
11968 (loop_exits_before_overflow): Adjust.
11969 (free_numbers_of_iterations_estimates_loop): Use an overload.
11970 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11971 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11973 2017-06-19 Richard Biener <rguenther@suse.de>
11976 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11977 RANGE_EXPR conservatively.
11979 2017-06-16 Carl Love <cel@us.ibm.com>
11981 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11982 definitions for vec_float, vec_float2, vec_floato,
11983 vec_floate built-ins.
11984 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11985 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11987 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11988 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11989 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11990 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11991 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11992 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11993 vec_floato): Add builtin defines.
11994 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11995 Update the built-in documentation file for the new built-in
11998 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12000 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12001 (mthumb): Mark as the negative of -marm.
12003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12005 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12007 (ARM Options, -mtune): Document that this accepts the same
12008 extension options as -mcpu.
12009 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12011 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12013 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12014 permitted extensions.
12016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12018 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12019 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12020 (armv8-m.main): Add option +nodsp.
12021 * config/arm/arm-cpu-cdata.h: Regenerated.
12023 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12025 * config/arm/t-fuchsia: New file.
12026 * config.gcc (arm*-*-fuchsia*): Use it.
12028 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12030 * config/arm/t-symbian: Rewrite for new option infrastructure.
12032 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12034 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12035 (MULTILIB_REQUIRED): Likewise.
12037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12039 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12040 (MULTILIB_RESUE): Likewise.
12041 (MULTILIB_MATCHES): Likewise.
12042 (MULTLIB_REQUIRED): Likewise.
12044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12046 * config/arm/t-rtems: Rewrite for new option framework.
12048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12050 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12051 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12052 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12053 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12054 * config/arm/t-multilib: ... here.
12055 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12056 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12057 armv7-a and armv8*-a when A-profile libraries have not been built.
12058 * config/arm/t-rmprofile: Rewrite.
12060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12062 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12063 with a backslash. Remove the backslash after substituting unescaped
12065 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12067 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12069 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12070 the driver to pass through the default setting of -mfloat-abi.
12071 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12073 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12074 (all_feat_combs): New rule.
12075 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12077 * config/arm/t-aprofile: Rewrite.
12079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12081 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12082 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12083 fpu is not specified by the user/command-line.
12084 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12085 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12086 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12087 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12088 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12089 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12093 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12094 * config/arm/t-arm-elf: Rewritten.
12096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12098 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12099 have some floating-point instructions.
12100 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12101 (TARGET_MAYBE_HARD_FLOAT): New macro.
12102 * config/arm/arm-builtins.c (arm_init_builtins): Use
12103 TARGET_MAYBE_HARD_FLOAT.
12104 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12108 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12109 (configargs.h): Include it.
12110 (arm_print_hint_for_fpu_option): New function.
12111 (arm_parse_fpu_option): New function.
12112 (candidate_extension): New class.
12113 (arm_canon_for_multilib): New function.
12114 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12115 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12116 (ARCH_CANONICAL_SPECS): New macro.
12117 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12119 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12121 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12122 are set after handling multilib fragments. Set target_cpu_default2
12125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12127 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12129 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12130 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12132 (check_arch): Likewise.
12133 * config/arm/arm.c (arm_configure_build_target): Handle
12134 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12135 options in the default.
12137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12139 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12140 when an option is an alias of another.
12141 * config/arm/parsecpu.awk (optalias): New parser token.
12142 (gen_comm_data): Mark non-alias options as such. Emit entries
12143 for extension aliases.
12144 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12145 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12146 (armv6kz, armv6zk, armv6t2): Likewise.
12147 (armv7): Make vfpv3-d16 an alias.
12148 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12150 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12151 Sort in canonical order.
12152 (armv8-a): Sort in canonical order.
12153 (armv8.1-a, armv8.2-a): Likewise.
12154 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12156 (cortex-a9): Sort in canonical order.
12157 * config/arm/arm.c (selftests.h): Include it.
12158 (arm_test_cpu_arch_data): New function.
12159 (arm_run_self_tests): New function.
12160 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12161 (targetm): Move declaration to the end of the file.
12162 * arm-cpu-cdata.h: Regenerated.
12164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12166 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12167 call to target_mode_check describing the type of option passed.
12168 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12169 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12170 arm_parse_cpu_option_name to match parameters against list of
12172 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12173 arm_arch_core_flags data structure.
12174 * config/arm/arm-cpu_cdata.h: Regenerated.
12176 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12178 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12180 (arm_print_hint_for_cpu_option): Likewise.
12181 (arm_print_hint_for_arch_option): Likewise.
12182 (arm_parse_cpu_option_name): Likewise.
12183 (arm_parse_arch_option_name): Likewise.
12184 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12185 of entries in the all_fpus list.
12186 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12187 (arm_parse_cpu_option_name): Declare.
12188 (arm_parse_arch_option_name): Declare.
12189 (arm_parse_option_features): Declare.
12190 (arm_intialize_isa): Declare.
12191 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12193 (gen_comm_data): ... here. Make definitions non-static.
12194 * config/arm/arm-cpu-data.h: Regenerated.
12195 * config/arm/arm-cpu-cdata.h: Regenerated.
12197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12199 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12200 (cpu_arch_extension): New structure.
12201 (cpu_arch_option, arch_option, cpu_option): New structures.
12202 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12203 architecture types.
12204 (gen_data): Generate new format data tables.
12205 * config/arm/arm.c (cpu_tune): New structure.
12206 (cpu_option, processors): Delete.
12207 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12208 (arm_print_hint_for_cpu_option): ... this and ...
12209 (arm_print_hint_for_arch_option): ... this.
12210 (arm_parse_arch_cpu_name): Delete. Replace with ...
12211 (arm_parse_cpu_option_name): ... this and ...
12212 (arm_parse_arch_option_name): ... this.
12213 (arm_unrecognized_feature): Change type of target parameter to
12215 (arm_parse_arch_cpu_features): Delete. Replace with ...
12216 (arm_parse_option_features): ... this.
12217 (arm_configure_build_target): Rework to use new configuration data
12219 (arm_print_tune_info): Rework for new configuration data tables.
12220 * config/arm/arm-cpu-data.h: Regenerated.
12221 * config/arm/arm-cpu.h: Regenerated.
12223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12225 * Makefile.in (OBJS): Move sbitmap.o from here ...
12226 (OBJS-libcommon): ... to here.
12228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12230 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12231 (ISA_ALL_CRYPTO): New macro.
12232 (ISA_ALL_SIMD): New macro
12233 (ISA_ALL_FP): New macro.
12234 * config/arm/arm.c (fpu_bitlist): Update initializer.
12235 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12237 (arm9e): Add fpu. Add option for nofp
12238 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12239 (arm926ej-s, arm1026ej-s): Likewise.
12240 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12241 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12242 neon-fp16, neon-vfpv4, nofp and nosimd.
12243 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12244 (cortex-a8): Add fpu. Add option for nofp.
12245 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12246 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12247 (cortex-r4f): Add fpu.
12248 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12249 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12251 (cortex-r8): Likewise.
12252 (cortex-m4): Add fpu. Add option for nofp.
12253 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12254 (cortex-a17.cortex-a7): Likewise.
12255 (cortex-a32): Add fpu. Add options for crypto and nofp.
12256 (cortex-a35, cortex-a53): Likewise.
12257 (cortex-a57): Add fpu. Add option for crypto.
12258 (cortex-a72, cortex-a73): Likewise.
12259 (exynos-m1): Likewise.
12260 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12261 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12262 (cortex-m33): Add fpu. Add option for nofp.
12263 * config/arm/arm-cpu-cdata.h: Regenerated
12264 * config/arm/arm-cpu-data.h: Regenerated.
12266 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12268 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12269 (armv5te, armv5tej): Likewise.
12270 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12271 (armv7): Add options fp and vfpv3-d16.
12272 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12273 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12275 (armv7ve): Likewise.
12276 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12277 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12278 (armv8-a): Add nocrypto option.
12279 (armv8.1-a, armv8.2-a): Likewise.
12280 (armv8-m.main): add options fp, fp.dp and nofp.
12282 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12284 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12286 (armv8-a+crc): Delete.
12287 (armv8.1-a): Add options simd, crypto and nofp.
12288 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12289 (armv8.2-a+fp16): Delete.
12290 (armv8-m.main): Add option dsp.
12291 (armv8-m.main+dsp): Delete.
12292 (cortex-a8): Add fpu. Add nofp option.
12293 (cortex-a9): Add fpu. Add nofp and nosimd options.
12294 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12295 link to main cpu and architecture data structures.
12296 (gen_comm_data): Only put isa attributes from the main architecture
12298 (option): New statement for architecture and CPU entries.
12299 * arm.c (struct cpu_option): New structure.
12300 (struct processors): Add entry for options.
12301 (arm_unrecognized_feature): New function.
12302 (arm_parse_arch_cpu_name): Ignore any characters after the first
12304 (arm_parse_arch_cpu_feature): New function.
12305 (arm_configure_build_target): Separate out any CPU and architecture
12306 features and parse separately. Don't error out if -mfpu=auto is
12307 used with only an architecture string.
12308 (arm_print_asm_arch_directives): New function.
12309 (arm_file_start): Call it.
12310 * config/arm/arm-cpu-cdata.h: Regenerated.
12311 * config/arm/arm-cpu-data.h: Likewise.
12312 * config/arm/arm-tables.opt: Likewise.
12314 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12316 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12317 assembler when it is not -mfpu=auto.
12319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12321 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12322 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12323 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12324 (ASM_CPU_SPEC): Rewrite.
12325 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12326 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12327 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12328 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12329 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12330 copied string is NUL-terminated. Also strip any characters prefixed
12332 (arm_rewrite_selected_arch): New function.
12333 (arm_rewrite_march): New function.
12335 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12337 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12338 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12339 (march, mcpu, mtune): Convert to string-based options.
12340 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12341 (arm_parse_arch_cpu_name): New function.
12342 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12343 identify selected architecture or CPU.
12344 (arm_option_save): New function.
12345 (TARGET_OPTION_SAVE): Redefine.
12346 (arm_option_restore): Restore string options.
12347 (arm_option_print): Print string options.
12349 2017-06-16 Martin Sebor <msebor@redhat.com>
12351 PR tree-optimization/80933
12352 PR tree-optimization/80934
12353 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12354 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12355 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12356 (gimple_fold_builtin): Call them.
12358 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12360 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12361 as unlikely; update profile.
12363 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12365 * predict.c (force_edge_cold): Handle declaring edges impossible
12368 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12370 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12372 (try_unroll_loop_completely): Fix reporting.
12374 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12376 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12378 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12381 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12382 if given a non-constant argument for an intrinsic which requires a
12385 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12387 * profile.c (compare_freqs): New function.
12388 (branch_prob): Sort edge list.
12389 (find_spanning_tree): Assume that the list is priority sorted.
12391 2017-06-16 Richard Biener <rguenther@suse.de>
12393 PR tree-optimization/81090
12394 * passes.def (pass_record_bounds): Remove.
12395 * tree-pass.h (make_pass_record_bounds): Likewise.
12396 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12397 make_pass_record_bounds): Likewise.
12398 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12399 not free niter estimates at the beginning but at the end.
12400 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12402 2017-06-16 Richard Biener <rguenther@suse.de>
12404 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12405 initializer to workaround ICE in host GCC 4.8.
12407 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12409 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12411 (clone_inlined_nodes): Update.
12413 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12415 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12416 prefetch settings, and enable prefetching by default at -O3.
12418 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12420 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12421 Set flag_prefetch_loop_arrays according to tuning data.
12423 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12425 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12426 New tune structure.
12427 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12428 [Unrelated to main purpose of the patch] Place the pointer field last
12429 to enable type checking errors when tune structure are wrongly merged.
12430 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12431 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12432 (thunderx2t99_prefetch_tune): New tune constants.
12433 (tune_params *_tunings): Update all tunings (no functional change).
12434 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12435 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12436 from tunings structures.
12438 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12441 * ubsan.c (instrument_null): Add T argument, use it instead
12442 of computing it based on IS_LHS.
12443 (instrument_object_size): Likewise.
12444 (pass_ubsan::execute): Adjust instrument_null and
12445 instrument_object_size callers to pass gimple_get_lhs or
12446 gimple_assign_rhs1 result to it. Use instrument_null instead of
12447 calling get_base_address and instrument_mem_ref. Handle
12448 aggregate call arguments for object-size sanitization.
12450 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12452 PR tree-optimization/81089
12453 * tree-vrp.c (is_masked_range_test): Validate operands of
12456 2017-06-15 Martin Sebor <msebor@redhat.com>
12459 * dumpfile.c (dump_register): Avoid calling memset to initialize
12460 a class with a default ctor.
12461 * gcc.c (struct compiler): Remove const qualification.
12462 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12463 * hash-table.h: Ditto.
12464 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12466 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12467 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12469 * params.h (struct param_info): Make struct members non-const.
12470 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12471 with default initialization.
12472 * vec.h (vec_copy_construct, vec_default_construct): New helper
12474 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12475 with vec_copy_construct.
12476 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12477 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12478 * doc/invoke.texi (-Wclass-memaccess): Document.
12480 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12482 * emit-rtl.h (is_leaf): Update comment about local
12483 register allocator.
12485 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12488 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12489 for a variable to have a section before checking if the section has a
12491 Set section to.persistent if persistent attribute is set.
12492 Warn if .persistent attribute is used on an automatic variable.
12494 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12496 PR rtl-optimization/80474
12497 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12499 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12501 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12504 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12506 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12507 (casesi): Emit bounds checking as RTL.
12508 (casesi_internal_mips16_<mode>): Remove bounds checking.
12510 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12512 * config/xtensa/xtensa.c (xtensa_option_override): Append
12513 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12514 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12515 xtensa_doloop_hooks): Define unconditionally.
12516 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12518 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12519 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12520 for it in xtensa_option_override.
12521 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12522 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12524 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12526 * doc/cppopts.texi: Document '-' special value to -MF.
12528 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12530 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12531 (cortex_a53_fconst): Likewise.
12532 (cortex_a53_fpmul): Likewise.
12533 (cortex_a53_f_load_64): Likewise.
12534 (cortex_a53_f_load_many): Likewise.
12535 (cortex_a53_advsimd_alu): Likewise.
12536 (cortex_a53_advsimd_alu_q): Likewise.
12537 (cortex_a53_advsimd_mul): Likewise.
12538 (cortex_a53_advsimd_mul_q): Likewise.
12539 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12540 Add missing fmul, r2f_cvt and fconst cases.
12542 2017-06-14 Richard Biener <rguenther@suse.de>
12544 PR middle-end/81088
12545 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12547 (fold_binary_loc): When associating do not treat pre-existing
12548 TREE_OVERFLOW on literal constants as a reason to allow
12549 TREE_OVERFLOW on associated literal constants.
12551 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12553 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12554 (MASK_FEATURES): New macro.
12555 * config/sparc/sparc.c (sparc_option_override): Remove the special
12556 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12558 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12560 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12561 a division of 0 if non-call exceptions are enabled.
12563 2017-06-14 Andrew Pinski <apinski@cavium.com>
12564 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12567 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12568 Improve vector initialization code gen for only variable case.
12570 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12572 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12574 2017-06-14 Richard Biener <rguenther@suse.de>
12576 PR tree-optimization/81083
12577 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12580 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12582 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12583 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12584 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12585 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12586 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12587 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12589 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12591 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12592 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12594 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12596 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12598 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12600 * config/rs6000/t-rtems: Don't handle SPE.
12602 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12604 * config/rs6000/t-linux: Don't handle SPE.
12606 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12608 * config/rs6000/eabispe.h: Delete file.
12610 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12612 * config/rs6000/t-spe: Delete file.
12614 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12616 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12617 (rs6000_legitimate_offset_address_p): Return false for anything in
12618 V2SImode or V2SFmode.
12620 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12622 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12623 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12624 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12626 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12627 (rs6000_legitimate_offset_address_p): Ditto.
12628 (rs6000_emit_move): Ditto.
12629 (rs6000_init_builtins): Remove V4HI_type_node.
12631 2017-06-13 Martin Liska <mliska@suse.cz>
12634 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12635 (gate_asan): Likewise.
12636 * asan.h (asan_no_sanitize_address_p): Remove the function.
12637 (sanitize_flags_p): New function.
12638 * builtins.def: Fix coding style.
12639 * common.opt: Use renamed enum value.
12640 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12641 * doc/extend.texi: Document no_sanitize attribute.
12642 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12643 to SANITIZE_UNDEFINED_NONDEFAULT.
12644 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12645 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12646 Use sanitize_flags_p.
12647 * gimplify.c (gimplify_function_tree): Likewise.
12648 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12649 * opts.c (parse_no_sanitize_attribute): New function.
12650 (common_handle_option): Use renamed enum value.
12651 * opts.h (parse_no_sanitize_attribute): Declare.
12652 * tree.c (sanitize_flags_p): New function.
12653 * tree.h: Declared here.
12654 * tsan.c: Use sanitize_flags_p.
12655 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12656 (instrument_mem_ref): Likewise.
12657 (instrument_bool_enum_load): Likewise.
12658 (do_ubsan_in_current_function): Remove the function.
12659 (pass_ubsan::execute): Use sanitize_flags_p.
12660 * ubsan.h: Remove do_ubsan_in_current_function
12661 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12662 (dump_function_to_file): Use it here.
12664 2017-06-13 Martin Jambor <mjambor@suse.cz>
12666 PR tree-optimization/80803
12667 PR tree-optimization/81063
12668 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12669 (propagate_subaccesses_across_link): Enqueue subtree whenever
12670 necessary instead of relying on the caller.
12672 2017-06-13 Martin Jambor <mjambor@suse.cz>
12674 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12675 that have a first_link.
12676 (sort_and_splice_var_accesses): Do not check first_link before
12678 (subtree_mark_written_and_enqueue): Likewise.
12679 (propagate_all_subaccesses): Likewise and do not stop at first
12680 parent with a first_link.
12682 2017-06-13 Martin Jambor <mjambor@suse.cz>
12684 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12687 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12689 * match.pd: New pattern.
12691 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12693 * tree-vrp.c (is_masked_range_test): New function.
12694 (register_edge_assert_for): Determine ranges for
12697 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12699 PR tree-optimization/67328
12700 * fold-const.c (maskable_range_p): New function.
12701 (build_range_check): Generate bittests if possible.
12703 2017-06-13 Martin Liska <mliska@suse.cz>
12705 * gimple-pretty-print.c (dump_probability): Add new argument.
12706 (dump_edge_probability): Dump both probability and count.
12707 (dump_gimple_label): Likewise.
12708 (dump_gimple_bb_header): Likewise.
12710 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12713 * config/avr/avr-devices.c: Fix indentation.
12714 * config/avr/gen-avr-mmcu-specs.c: Dito.
12716 2017-06-13 Richard Biener <rguenther@suse.de>
12718 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12719 instead get vector type from stmt_info.
12720 (vectorizable_reduction): Adjust. Remove dead code.
12722 2017-06-13 Richard Biener <rguenther@suse.de>
12724 PR middle-end/81065
12725 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12726 case of C * (x * C2 + C3).
12727 (fold_addr_of_array_ref_difference): Properly fold index difference.
12729 2017-06-12 David S. Miller <davem@davemloft.net>
12732 * config/sparc/sparc.md (return expander): Emit frame blockage if
12733 function uses alloca.
12735 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12737 * combine.c (make_field_assignment): Check len rather than the mode
12738 precision when calling force_to_mode.
12740 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12742 Support multilibs and devices that see flash in RAM address range.
12745 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12746 (avr_mcu_t) <flash_pm_offset>: New field.
12747 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12748 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12749 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12750 (AVR_TINY_PM_OFFSET): Remove macro.
12751 * config/avr/avr.opt (-mshort-calls): New option.
12752 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12753 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12754 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12755 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12756 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12757 instead of avr_arch->have_jmp_call.
12758 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12759 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12760 avr_arch->flash_pm_offset to define.
12761 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12762 new field flash_pm_offset. Add entry for avrxmega3.
12763 (avr_texinfo): Add entry for avrxmega3.
12764 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12765 attiny212, attiny214,
12766 attiny412, attiny414, attiny416, attiny417,
12767 attiny814, attiny816, attiny817,
12768 attiny1614, attiny1616, attiny1617,
12769 attiny3214, attiny3216, attiny3217.
12770 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12771 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12772 (avr_print_operand_address) [AVR_TINY]: Same.
12773 (avr_asm_init_sections) <readonly_data_section>: Only patch
12774 callback if avr_arch->flash_pm_offset = 0.
12775 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12776 for rodata if avr_arch->flash_pm_offset != 0.
12777 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12778 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12779 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12780 (m_options): Append opt_rcall.
12781 (m_dirnames): Append dir_rcall.
12782 * config/avr/t-multilib: Regenerate.
12784 * configure.ac [target=avr]: Check whether avrxmega3 default
12785 linker description file works as needed.
12786 * configure: Regenerate.
12787 * doc/avr-mmcu.texi: Regenerate.
12788 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12789 <__AVR_ARCH__>: Document avrxmega3 and 103.
12790 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12791 <__AVR_SHORT_CALLS__>: Document it.
12792 <__AVR_PM_BASE_ADDRESS__>: Document it.
12793 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12794 (AVR Variable Attributes) <progmem>: Document this is
12795 not needed for avrxmega3.
12796 (AVR Named Address Spaces) <__flash>: Dito.
12798 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12800 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12802 2017-06-12 Doug Rupp <rupp@adacore.com>
12804 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12805 Append vxworks-stdint.h to the tm_file list.
12806 * config/vxworks-stdint.h: New file.
12808 2017-06-12 Martin Liska <mliska@suse.cz>
12810 PR tree-optimization/81041
12811 * tree-profile.c (gimple_gen_ic_func_profiler):
12812 Create an extra BB in profile-generate
12813 (gimple_gen_time_profiler): Likewise.
12815 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12817 PR tree-optimization/81003
12818 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12819 (update_range_test): Use it instead of force_gimple_operand_gsi.
12821 2017-06-12 Richard Biener <rguenther@suse.de>
12823 PR tree-optimization/81053
12824 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12825 with backedge value not defined in loop. Simplify def stmt
12828 2017-06-11 Tom de Vries <tom@codesourcery.com>
12831 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12833 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12834 nvptx_cannot_force_const_mem.
12836 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12838 * opts.c (finish_options): Move test for flag_split_stack after
12839 it has been initialized.
12841 2017-06-11 Jason Merrill <jason@redhat.com>
12843 * tree.h (id_equal): New.
12844 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12845 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12846 instead of strcmp of IDENTIFIER_POINTER.
12848 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12850 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12851 (mark_all_inlined_calls_cdtor): Fix formating.
12852 (inline_transform): Rescale profile before inlining.
12854 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12856 * cgraph.h (cgraph_edge::clone): Update prototype.
12857 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12858 (cgraph_node::create_clone): Update.
12859 (cgraph_node::create_version_clone): Update.
12860 * tree-inline.c (copy_bb): Update.
12861 (expand_call_inline): Update.
12863 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12865 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12866 factored out from ...
12867 (rs6000_emit_prologue): ... here.
12869 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12871 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12872 factored out from ...
12873 (rs6000_emit_prologue): ... here.
12875 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12877 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12879 (handle_missing_profiles): Fix computation of tp_first_run.
12880 (counts_to_freqs): Do not touch freqs when count is 0.
12882 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12884 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12887 2017-06-10 Tom de Vries <tom@codesourcery.com>
12889 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12890 attributes): Document signal effective target.
12892 2017-06-10 Tom de Vries <tom@codesourcery.com>
12894 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12895 Document effective target stack_size.
12897 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12899 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12900 to the edit_context if they can be auto-applied.
12902 2017-06-9 Ian Lance Taylor <iant@golang.org>
12904 * opts.c (finish_options): If -fsplit-stack, disable implicit
12905 -forder-blocks-and-partition.
12906 * doc/invoke.texi (Optimize Options): Document that when using
12907 -fsplit-stack -forder-blocks-and-partition is not implicitly
12910 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12912 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12913 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12914 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12915 * builtins.def (abort, trap, unreachable): Declare cold.
12916 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12917 * tree-core.h (ECF_COLD): New.
12918 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12919 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12921 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12923 * predict.c (unlikely_executed_stmt_p): Cleanup.
12925 2017-06-09 Richard Biener <rguenther@suse.de>
12927 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12928 model if the ref is always written to.
12930 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12932 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12934 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12936 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12939 2017-06-09 Tom de Vries <tom@codesourcery.com>
12942 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12943 "target cannot support label values" when encountering LABEL_REF.
12945 2017-06-09 Martin Liska <mliska@suse.cz>
12947 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12948 (gimple_gen_ic_func_profiler): Emit direct comparison
12949 of __gcov_indirect_call_callee with NULL.
12950 (gimple_gen_time_profiler): Change probability from
12951 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12953 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12955 * profile.c (edge_gcov_counts): Turn to pointer.
12956 (compute_branch_probabilities, compute_branch_probabilities): Update.
12957 (branch_prob): Do not clear edge_gcov_count.
12958 * profile.h (edge_gcov_counts): Turn to pointer.
12959 (edge_gcov_count): Update.
12961 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12963 * gimple.h (gimple_check_failed): Mark cold.
12965 2017-06-09 Richard Biener <rguenther@suse.de>
12967 PR tree-optimization/66623
12968 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12969 refactor check_reduction into two parts, properly computing
12970 whether we have to check reduction validity for outer loop
12973 2017-06-09 Richard Biener <rguenther@suse.de>
12975 PR tree-optimization/79483
12976 * graphite-scop-detection.c (order): New global.
12977 (get_order): Compute bb to order mapping that satisfies code
12978 generation constraints.
12979 (cmp_pbbs): New helper.
12980 (build_scops): Start domwalk at entry block, sort generated
12983 2017-06-09 Richard Biener <rguenther@suse.de>
12985 PR middle-end/81007
12986 * ipa-polymorphic-call.c
12987 (ipa_polymorphic_call_context::restrict_to_inner_class):
12988 Skip FIELD_DECLs with error_mark_node type.
12989 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12992 2017-06-09 Martin Liska <mliska@suse.cz>
12994 * predict.c (struct branch_predictor): New struct.
12995 (test_prediction_value_range): New test.
12996 (predict_c_tests): New function.
12997 * selftest-run-tests.c (selftest::run_tests): Run the function.
12998 * selftest.h: Declare new tests.
13000 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13003 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13004 gen_add3_insn did not fail.
13005 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13006 r0, construct that number in a temporary reg and add that reg to r0.
13007 If asked to put the result in r0 as well, fail.
13009 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13011 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13012 for early expansion of vec_eqv.
13014 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13016 PR middle-end/81005
13017 * ubsan.c (instrument_null): Avoid pointless code temporary.
13018 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13020 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13024 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13027 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13030 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13032 * predict.c (unlikely_executed_edge_p): Move ahead.
13033 (probably_never_executed_edge_p): Use it.
13035 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13037 PR middle-end/79988
13038 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13039 gimple_call_builtin_p call.
13041 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13043 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13044 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13045 rtl_check_failed_type2, rtl_check_failed_code1,
13046 rtl_check_failed_code2, rtl_check_failed_code_mode,
13047 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13048 rtvec_check_failed_bounds, rtl_check_failed_flag,
13049 _fatal_insn_not_found, _fatal_insn): Likewise.
13050 * tree.h (tree_contains_struct_check_failed,
13051 tree_check_failed, tree_not_check_failed,
13052 tree_class_check_failed, tree_range_check_failed,
13053 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13054 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13055 tree_operand_check_failed, omp_clause_check_failed,
13056 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13059 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13061 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13062 flag_branch_probabilities.
13063 * ipa-inline.c (edge_badness): Likewise.
13064 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13065 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13066 * predict.c (maybe_hot_frequency_p): Likewise.
13067 (probably_never_executed): Likewise.
13068 * sched-ebb.c (schedule_ebbs): Likewise.
13069 * sched-rgn.c (find_single_block_region): Likewise.
13070 * tracer.c (tail_duplicate): Likewise.
13072 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13074 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13075 longer requires x_flag_profile_use.
13077 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13079 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13080 instead of flag_reorder_blocks_and_partition.
13081 * dbxout.c (dbxout_function_end): Likewise.
13082 * dwarf2out.c (gen_subprogram_die): Likewise.
13083 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13084 * hw-doloop.c (reorg_loops): Likewise.
13085 * varasm.c (assemble_start_function,
13086 assemble_end_function): Likewise.
13087 (decide_function_section): Do not check for
13088 flag_reorder_blocks_and_partition.
13090 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13092 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13094 (chkp_get_hard_register_fake_addr_expr): Ditto.
13095 (chkp_build_addr_expr): Add check for hard reg case.
13096 (chkp_parse_array_and_component_ref): Ditto.
13097 (chkp_find_bounds_1): Ditto.
13098 (chkp_process_stmt): Don't generate bounds store for
13101 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13103 * predict.c (maybe_hot_bb_p): Do not check profile status.
13104 (maybe_hot_edge_p): Likewise.
13105 (probably_never_executed): Check for zero counts even if profile
13107 (unlikely_executed_edge_p): New function.
13108 (unlikely_executed_stmt_p): New function.
13109 (unlikely_executed_bb_p): New function.
13110 (set_even_probabilities): Use unlikely predicates.
13111 (combine_predictions_for_bb): Likewise.
13112 (predict_paths_for_bb): Likewise.
13113 (predict_paths_leading_to_edge): Likewise.
13114 (determine_unlikely_bbs): New function.
13115 (estimate_bb_frequencies): Use it.
13116 (compute_function_frequency): Use zero counts even if profile is
13118 * profile-count.h: Fix typo.
13120 2017-08-08 Julia Koval <julia.koval@intel.com>
13122 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13123 _mm512_mask_cvtsepi16_storeu_epi8,
13124 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13125 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13126 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13127 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13128 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13129 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13130 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13131 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13132 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13133 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13134 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13135 __builtin_ia32_pmovuswb256mem_mask,
13136 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13137 __builtin_ia32_pmovwb512mem_mask): New builtins.
13139 2017-08-08 Julia Koval <julia.koval@intel.com>
13141 PR target/73350,80862
13142 * config/i386/subst.md (round): Fix round pattern.
13143 * config/i386/i386.c (ix86_erase_embedded_rounding):
13144 Fix erasing rounding for the fixed pattern.
13146 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13148 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13150 2017-06-08 Martin Liska <mliska@suse.cz>
13152 PR gcov-profile/80911
13153 * gcov.c (block_info::block_info): New constructor.
13155 2017-06-07 Carl Love <cel@us.ibm.com>
13157 * config/rs6000/rs6000-c: The return type of the following
13158 built-in functions was implemented as int not long long. Fix sign
13159 of return value for the unsigned version of vec_mulo and vec_mule.
13160 vector unsigned long long vec_bperm (vector unsigned long long,
13161 vector unsigned char)
13162 vector signed long long vec_mule (vector signed int,
13164 vector unsigned long long vec_mule (vector unsigned int,
13165 vector unsigned int)
13166 vector signed long long vec_mulo (vector signed int,
13168 vector unsigned long long vec_mulo (vector unsigned int,
13169 vector unsigned int)
13170 * doc/extend.texi: Fix the documentation for the built-in
13173 2017-06-07 Carl Love <cel@us.ibm.com>
13176 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13179 2017-06-07 Carl Love <cel@us.ibm.com>
13181 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13182 support, Generate doublehv for signed int/float for BE case only.
13184 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13186 * doc/invoke.texi (mcx16): Rewrite.
13188 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13190 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13191 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13192 *mov<mode>_softfloat, and an anonymous splitter): Use
13193 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13195 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13197 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13199 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13200 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13201 (rs6000_debug_reg_global): Adjust.
13202 (rs6000_init_hard_regno_mode_ok): Adjust.
13203 (rs6000_dbx_register_number): Adjust.
13204 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13205 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13206 Remove SPE_ACC and SPEFSCR.
13207 (REG_ALLOC_ORDER): Ditto.
13208 (FRAME_POINTER_REGNUM): Change to 111.
13209 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13210 (REG_CLASS_NAMES): Ditto.
13211 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13212 (REGISTER_NAMES): Ditto.
13213 (ADDITIONAL_REG_NAMES): Ditto.
13214 (rs6000_reg_names): Ditto.
13215 * config/rs6000/rs6000.md: Renumber some register number
13218 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13220 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13222 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13223 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13225 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13226 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13227 Delete the SPE high registers.
13228 (REG_ALLOC_ORDER): Ditto.
13229 (enum reg_class): Remove SPE_HIGH_REGS.
13230 (REG_CLASS_NAMES): Ditto.
13231 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13232 (REGISTER_NAMES): Ditto.
13233 (rs6000_reg_names): Ditto.
13234 * doc/tm.texi.in: Remove SPE as example.
13235 * doc/tm.texi: Regenerate.
13237 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13239 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13240 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13241 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13242 * config/rs6000/rs6000.md (type): Remove "brinc".
13244 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13246 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13247 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13248 * config/rs6000/linuxspe.h: Delete file.
13249 * config/rs6000/rs6000.md: Don't include spe.md.
13250 * config/rs6000/spe.h: Delete file.
13251 * config/rs6000/spe.md: Delete file.
13252 * config/rs6000/t-rs6000: Remove spe.md.
13254 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13256 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13257 (reg_or_none500mem_operand): Delete.
13258 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13259 instead of reg_or_none500mem_operand.
13261 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13263 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13264 handling of SPE flags.
13265 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13267 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13269 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13271 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13272 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13273 paired_divv2sf3): Similar.
13274 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13275 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13276 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13278 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13279 Rename the paired_* instruction patterns.
13280 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13282 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13283 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13284 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13285 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13286 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13287 PAIRED_VECTOR_MODE.
13288 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13289 (spe_func_has_64bit_regs_p): Delete.
13290 (spe_expand_predicate_builtin): Delete.
13291 (spe_expand_evsel_builtin): Delete.
13292 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13293 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13294 (invalid_e500_subreg): Delete.
13295 (rs6000_legitimize_address): Always force_reg op2 as well, for
13296 paired single memory accesses.
13297 (rs6000_member_type_forces_blk): Delete.
13298 (rs6000_spe_function_arg): Delete.
13299 (rs6000_expand_unop_builtin): Delete SPE handling.
13300 (rs6000_expand_binop_builtin): Ditto.
13301 (spe_expand_stv_builtin): Delete.
13302 (bdesc_2arg_spe): Delete.
13303 (spe_expand_builtin): Delete.
13304 (spe_expand_predicate_builtin): Delete.
13305 (spe_expand_evsel_builtin): Delete.
13306 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13307 (spe_init_builtins): Delete.
13308 (spe_func_has_64bit_regs_p): Delete.
13309 (savres_routine_name): Delete "info" parameter. Adjust callers.
13310 (rs6000_emit_stack_reset): Ditto.
13311 (rs6000_dwarf_register_span): Delete.
13312 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13313 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13314 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13316 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13318 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13319 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13320 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13321 mulv2sf3, divv2sf3): Delete expanders.
13323 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13325 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13327 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13329 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13330 * config/rs6000/rs6000.c: Ditto.
13332 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13334 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13335 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13336 comparison_operator.
13338 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13340 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13341 * config/rs6000/rs6000.opt: Ditto.
13342 * config/rs6000/t-rtems: Ditto.
13344 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13346 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13347 TARGET_E500_SINGLE by 0, simplify.
13348 * config/rs6000/rs6000.c: Ditto.
13349 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13350 (spe_build_register_parallel): Delete.
13351 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13352 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13353 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13354 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13355 (E500_CONVERT): Delete.
13356 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13358 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13360 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13361 * config/rs6000/dfp.md: Ditto.
13362 (negdd2, *negdd2_fpr): Merge.
13363 (absdd2, *absdd2_fpr): Merge.
13364 (negtd2, *negtd2_fpr): Merge.
13365 (abstd2, *abstd2_fpr): Merge.
13366 * config/rs6000/e500.h: Delete file.
13367 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13368 TARGET_FPRS by 1 and simplify.
13369 * config/rs6000/rs6000-c.c: Ditto.
13370 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13371 TARGET_DF_SPE by 0.
13372 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13374 * config/rs6000/rs6000.md: Ditto.
13375 (floatdidf2, *floatdidf2_fpr): Merge.
13376 (move_from_CR_gt_bit): Delete.
13377 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13378 (E500_CR_IOR_COMPARE): Delete.
13379 (All patterns that require !TARGET_FPRS): Delete.
13380 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13382 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13384 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13386 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13388 * graphds.c (add_edge): Intitialize edge's attached data.
13389 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13390 pointer parameter. Call pointed function on each edge during
13391 graph traversing. Skip traversing the edge when the function
13393 (graphds_dfs, graphds_scc): Ditto.
13394 (for_each_edge): New parameter. Pass the new parameter to callback
13396 * graphds.h (skip_edge_callback): New function pointer type.
13397 (graphds_dfs, graphds_scc): New function pointer parameter.
13398 (graphds_edge_callback, for_each_edge): New parameter.
13400 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13402 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13403 out code checking if runtime alias check is possible to below ...
13404 Call the new function.
13405 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13406 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13408 2017-06-07 Marek Polacek <polacek@redhat.com>
13411 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13412 TYPE_OVERFLOW_WRAPS check.
13414 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13416 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13417 if versioning is required.
13418 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13419 peeling with the check for versioning.
13421 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13423 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13424 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13425 Set true to new parameter if new ssa variable is defined.
13426 (vect_gen_vector_loop_niters): Refactor. Set range information
13427 for the new vector loop bound variable.
13428 (vect_do_peeling): Ditto.
13430 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13432 * tree-affine.c (ssa.h): Include header file.
13433 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13434 has wrapping overflow behavior.
13436 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13438 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13440 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13442 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13443 (tree_to_aff_combination): ... here.
13445 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13447 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13448 reg_pressure model function.
13449 (ivopts_global_cost_for_size): Delete.
13450 (determine_set_costs, iv_ca_recount_cost): Call new model function
13451 ivopts_estimate_reg_pressure.
13453 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13455 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13456 expensive than udiv. Remove floating point cases from mod.
13458 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13460 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13461 Increase idiv cost.
13463 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13465 * config/aarch64/aarch64.md
13466 (copysignsf3): Fix mask generation.
13468 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13470 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13472 (class dump_manager): Add register_dumps method.
13473 * dumpfile.c: Include langhooks.h.
13474 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13475 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13476 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13477 (dump_manager::dump_register): Start with 512 entries instead of 32.
13478 (dump_manager::register_dumps): New method.
13479 * toplev.c (general_init): Instead of invoking register_dumps
13480 langhook, invoke register_dumps method on the dump manager.
13481 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13484 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13486 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13487 Say that # requires an associated define_split to exist, and that
13488 the define_split must be suitable for use after register allocation.
13490 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13492 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13493 (compute_outgoing_frequencies): Also initialize zero counts.
13494 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13495 around loops; preserve more of profile when nothing changes.
13497 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13499 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13501 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13502 * config/arm/arm-cpu-cdata.h: Regenerate.
13503 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13504 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13505 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13506 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13507 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13509 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13511 * config/arm/t-rmprofile: Likewise.
13512 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13514 2017-06-06 David S. Miller <davem@davemloft.net>
13517 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13518 blockage if function uses alloca.
13520 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13522 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13523 New "uid" fields to hold pretty-print IDs of group and ref.
13524 Memory references are now identified as <group_id>:<ref_id>
13525 instead of using [random] addresses.
13526 (dump_mem_details): Simplify, no functional change.
13527 (dump_mem_ref): Simplify and make output more concise.
13528 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13529 (find_or_create_group): Initialize group uid.
13530 (record_ref): Initialize ref uid. Improve debug output.
13531 (prune_group_by_reuse, should_issue_prefetch_p,)
13532 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13533 (mark_nontemporal_store, determine_loop_nest_reuse):
13534 Improve debug output.
13536 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13538 * dbgcnt.def (prefetch): New debug counter.
13539 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13540 (schedule_prefetches): Stop issueing prefetches if debug counter
13543 2017-06-06 Tom de Vries <tom@codesourcery.com>
13545 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13546 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13548 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13550 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13551 Use aarch64_reg_or_zero predicate for operand 4.
13552 (aarch64_compare_and_swap<mode> define_insn_and_split):
13553 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13554 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13556 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13558 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13559 (arm_compute_save_core_reg_mask): This.
13560 (thumb1_compute_save_reg_mask): Rename into ...
13561 (thumb1_compute_save_core_reg_mask): This.
13562 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13563 (arm_compute_frame_layout): Likewise.
13565 2017-06-06 Richard Biener <rguenther@suse.de>
13567 PR tree-optimization/80974
13568 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13569 keep or clear leaders SSA info.
13571 2017-06-06 Tom de Vries <tom@codesourcery.com>
13573 * config/nvptx/nvptx.c (split_mode_p): New function.
13574 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13576 2017-06-06 Tom de Vries <tom@codesourcery.com>
13578 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13580 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13583 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13586 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13588 * shrink-wrap.c (handle_simple_exit): Update profile.
13589 (try_shrink_wrapping): Upate profile.
13591 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13593 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13594 (tree_guess_outgoing_edge_probabilities): New.
13595 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13596 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13598 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13600 * ipa-split.c (split_function): Initialize return bb profile.
13602 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13604 * profile.c (compute_branch_probabilities): Also initialize
13605 EXIT_BLOCK profile.
13607 2017-06-06 Richard Biener <rguenther@suse.de>
13609 PR tree-optimization/80928
13610 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13611 (vect_analyze_loop_operations): Properly guard analysis for
13613 (vect_transform_loop): Likewise.
13614 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13615 (vect_model_induction_cost): Do not cost for pure SLP.
13616 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13617 of induction in inner loop vectorization.
13618 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13619 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13620 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13622 (vect_analyze_slp_cost_1): Cost induction.
13623 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13624 (vect_get_slp_vect_defs): Likewise.
13625 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13626 (vect_transform_stmt): Handle SLP reductions.
13627 * tree-vectorizer.h (vectorizable_induction): Adjust.
13629 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13631 * config/rs6000/rs6000.c (make_resolver_func): Update
13632 init_lowered_empty_function call.
13634 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13636 * doc/invoke.texi: Document the -fprofile-abs-path option.
13637 * common.opt (fprofile-abs-path): New option.
13638 * gcov-io.h (gcov_write_filename): Declare.
13639 * gcov-io.c (gcov_write_filename): New function.
13640 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13641 * profile.c (output_location): Likewise.
13643 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13645 * shring-wrap.c: Revert accidental commit.
13647 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13649 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13651 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13653 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13655 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13656 profile in callgraph edge.
13657 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13658 (apply_scale): Likewise.
13659 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13660 Also scale profile when inlining function with zero profile.
13661 (initialize_cfun): Update exit block profile even when it is zero.
13662 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13663 when profile is read.
13665 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13667 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13668 (CLONE_*): New constants to define the processors we can generate
13669 code for with the target_clone attribute.
13670 (rs6000_clone_map): New array to identify which clone processors
13671 the current program is running on.
13672 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13673 target_clone attribute.
13674 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13675 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13676 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13677 (cpu_expand_builtin): Add support for target_clone attribute.
13678 (rs6000_valid_attribute_p): Allow "default" attribute.
13679 (get_decl_name): New debug function to simplify printing the
13680 current function name in debugging statements.
13681 (rs6000_clone_priority): New functions to support the target_clone
13682 attribute, and be able to generate code to switch between ISA 2.05
13683 through ISA 3.0 (power6 through power9).
13684 (rs6000_compare_version_priority): Likewise.
13685 (rs6000_get_function_versions_dispatcher): Likewise.
13686 (make_resolver_func): Likewise.
13687 (add_condition_to_bb): Likewise.
13688 (dispatch_function_versions): Likewise.
13689 (rs6000_generate_version_dispatcher_body): Likewise.
13690 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13691 (fusion_gpr_load_p): Fix a spacing issue.
13692 * doc/extend.texi (Common Function Attributes): Document that the
13693 PowerPC supports the target_clone attribute.
13695 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13697 * config/arm/arm.h: explain F symbol found in description of ARM
13698 register allocation in its legend.
13700 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13702 * config/mips/frame-header-opt.c: Include profile-count.h.
13703 * config/riscv/riscv.c: Include profile-count.h
13705 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13707 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13709 (sm_set_flag_if_changed): Add bbs field.
13710 (execute_sm_if_changed_flag_set): Pass BBS.
13711 (execute_sm): Update.
13713 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13715 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13718 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13720 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13721 (peephole2): New peephole2 to emit the above.
13722 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13724 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13726 * config/aarch64/aarch64.c (define_peephole2 above
13727 *sub_<shift>_<mode>): New peephole.
13729 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13731 * config/i386/i386.c (make_resolver_func): Update.
13732 * Makefile.in: Add profile-count.h and profile-count.o
13733 * auto-profile.c (afdo_indirect_call): Update to new API.
13734 (afdo_set_bb_count): Update.
13735 (afdo_propagate_edge): Update.
13736 (afdo_propagate_circuit): Update.
13737 (afdo_calculate_branch_prob): Update.
13738 (afdo_annotate_cfg): Update.
13739 * basic-block.h: Include profile-count.h
13740 (struct edge_def): Turn count to profile_count.
13741 (struct basic_block_def): Likewie.
13742 (REG_BR_PROB_BASE): Move to profile-count.h
13743 (RDIV): Move to profile-count.h
13744 * bb-reorder.c (max_entry_count): Turn to profile_count.
13745 (find_traces): Update.
13746 (rotate_loop):Update.
13747 (connect_traces):Update.
13748 (sanitize_hot_paths):Update.
13749 * bt-load.c (migrate_btr_defs): Update.
13750 * cfg.c (RDIV): Remove.
13751 (init_flow): Use alloc_block.
13752 (alloc_block): Uninitialize count.
13753 (unchecked_make_edge): Uninitialize count.
13754 (check_bb_profile): Update.
13755 (dump_edge_info): Update.
13756 (dump_bb_info): Update.
13757 (update_bb_profile_for_threading): Update.
13758 (scale_bbs_frequencies_int): Update.
13759 (scale_bbs_frequencies_gcov_type): Update.
13760 (scale_bbs_frequencies_profile_count): New.
13761 * cfg.h (update_bb_profile_for_threading): Update.
13762 (scale_bbs_frequencies_profile_count): Declare.
13763 * cfgbuild.c (compute_outgoing_frequencies): Update.
13764 (find_many_sub_basic_blocks): Update.
13765 * cfgcleanup.c (try_forward_edges): Update.
13766 (try_crossjump_to_edge): Update.
13767 * cfgexpand.c (expand_gimple_tailcall): Update.
13768 (construct_exit_block): Update.
13769 * cfghooks.c (verify_flow_info): Update.
13770 (dump_bb_for_graph): Update.
13771 (split_edge): Update.
13772 (make_forwarder_block): Update.
13773 (duplicate_block): Update.
13774 (account_profile_record): Update.
13775 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13776 (get_estimated_loop_iterations): Update.
13777 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13778 (single_likely_exit): Update.
13779 * cfgloopmanip.c (scale_loop_profile): Update.
13781 (set_zero_probability): Update.
13782 (lv_adjust_loop_entry_edge): Update.
13783 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13784 (purge_dead_edges): Update.
13785 (rtl_account_profile_record): Update.
13786 * cgraph.c (cgraph_node::create): Uninitialize count.
13787 (symbol_table::create_edge): Uninitialize count.
13788 (cgraph_update_edges_for_call_stmt_node): Update.
13789 (cgraph_edge::dump_edge_flags): Update.
13790 (cgraph_node::dump): Update.
13791 (cgraph_edge::maybe_hot_p): Update.
13792 * cgraph.h: Include profile-count.h
13793 (create_clone), create_edge, create_indirect_edge): Update.
13794 (cgraph_node): Turn count to profile_count.
13795 (cgraph_edge0: Likewise.
13796 (make_speculative, clone): Update.
13797 (create_edge): Update.
13798 (init_lowered_empty_function): Update.
13799 * cgraphclones.c (cgraph_edge::clone): Update.
13800 (duplicate_thunk_for_node): Update.
13801 (cgraph_node::create_clone): Update.
13802 * cgraphunit.c (cgraph_node::analyze): Update.
13803 (cgraph_node::expand_thunk): Update.
13804 * final.c (dump_basic_block_info): Update.
13805 * gimple-streamer-in.c (input_bb): Update.
13806 * gimple-streamer-out.c (output_bb): Update.
13807 * graphite.c (print_global_statistics): Update.
13808 (print_graphite_scop_statistics): Update.
13809 * hsa-brig.c: Include basic-block.h.
13810 * hsa-dump.c: Include basic-block.h.
13811 * hsa-gen.c (T sum_slice): Update.
13812 (convert_switch_statements):Update.
13813 * hsa-regalloc.c: Include basic-block.h.
13814 * ipa-chkp.c (chkp_produce_thunks): Update.
13815 * ipa-cp.c (struct caller_statistics): Update.
13816 (init_caller_stats): Update.
13817 (gather_caller_stats): Update.
13818 (ipcp_cloning_candidate_p): Update.
13819 (good_cloning_opportunity_p): Update.
13820 (get_info_about_necessary_edges): Update.
13821 (dump_profile_updates): Update.
13822 (update_profiling_info): Update.
13823 (update_specialized_profile): Update.
13824 (perhaps_add_new_callers): Update.
13825 (decide_about_value): Update.
13826 (ipa_cp_c_finalize): Update.
13827 * ipa-devirt.c (struct odr_type_warn_count): Update.
13828 (struct decl_warn_count): Update.
13829 (struct final_warning_record): Update.
13830 (possible_polymorphic_call_targets): Update.
13831 (ipa_devirt): Update.
13832 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13833 * ipa-icf.c (sem_function::merge): Update.
13834 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13835 * ipa-inline.c (compute_uninlined_call_time): Update.
13836 (compute_inlined_call_time): Update.
13837 (want_inline_small_function_p): Update.
13838 (want_inline_self_recursive_call_p): Update.
13839 (edge_badness): Update.
13840 (lookup_recursive_calls): Update.
13841 (recursive_inlining): Update.
13842 (inline_small_functions): Update.
13843 (dump_overall_stats): Update.
13844 (dump_inline_stats): Update.
13845 * ipa-profile.c (ipa_profile_generate_summary): Update.
13846 (ipa_propagate_frequency): Update.
13847 (ipa_profile): Update.
13848 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13849 * ipa-utils.c (ipa_merge_profiles): Update.
13850 * loop-doloop.c (doloop_modify): Update.
13851 * loop-unroll.c (report_unroll): Update.
13852 (unroll_loop_runtime_iterations): Update.
13853 * lto-cgraph.c (lto_output_edge): Update.
13854 (lto_output_node): Update.
13855 (input_node): Update.
13856 (input_edge): Update.
13857 (merge_profile_summaries): Update.
13858 * lto-streamer-in.c (input_cfg): Update.
13859 * lto-streamer-out.c (output_cfg): Update.
13860 * mcf.c (create_fixup_graph): Update.
13861 (adjust_cfg_counts): Update.
13862 (sum_edge_counts): Update.
13863 * modulo-sched.c (sms_schedule): Update.
13864 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13865 * predict.c (maybe_hot_count_p): Update.
13866 (probably_never_executed): Update.
13867 (dump_prediction): Update.
13868 (combine_predictions_for_bb): Update.
13869 (propagate_freq): Update.
13870 (handle_missing_profiles): Update.
13871 (counts_to_freqs): Update.
13872 (rebuild_frequencies): Update.
13873 (force_edge_cold): Update.
13874 * predict.h: Include profile-count.h
13875 (maybe_hot_count_p, counts_to_freqs): UPdate.
13876 * print-rtl-function.c: Do not include cfg.h
13877 * print-rtl.c: Include basic-block.h
13878 * profile-count.c: New file.
13879 * profile-count.h: New file.
13880 * profile.c (is_edge_inconsistent): Update.
13881 (correct_negative_edge_counts): Update.
13882 (is_inconsistent): Update.
13883 (set_bb_counts): Update.
13884 (read_profile_edge_counts): Update.
13885 (compute_frequency_overlap): Update.
13886 (compute_branch_probabilities): Update; Initialize and deinitialize
13888 (branch_prob): Update.
13889 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13890 (edge_gcov_count): New.
13891 (bb_gcov_count): New.
13892 * shrink-wrap.c (try_shrink_wrapping): Update.
13893 * tracer.c (better_p): Update.
13894 * trans-mem.c (expand_transaction): Update.
13895 (ipa_tm_insert_irr_call): Update.
13896 (ipa_tm_insert_gettmclone_call): Update.
13897 * tree-call-cdce.c: Update.
13898 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13899 (gimple_duplicate_sese_tail): Update.
13900 (gimple_account_profile_record): Update.
13901 (execute_fixup_cfg): Update.
13902 * tree-inline.c (copy_bb): Update.
13903 (copy_edges_for_bb): Update.
13904 (initialize_cfun): Update.
13905 (freqs_to_counts): Update.
13906 (copy_cfg_body): Update.
13907 (expand_call_inline): Update.
13908 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13909 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13910 (try_unroll_loop_completely): Update.
13911 (try_peel_loop): Update.
13912 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13913 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13914 * tree-ssa-loop-split.c (connect_loops): Update.
13915 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13916 * tree-ssa-reassoc.c (branch_fixup): Update.
13917 * tree-ssa-tail-merge.c (replace_block_by): Update.
13918 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13919 (compute_path_counts): Update.
13920 (update_profile): Update.
13921 (recompute_probabilities): Update.
13922 (update_joiner_offpath_counts): Update.
13923 (estimated_freqs_path): Update.
13924 (freqs_to_counts_path): Update.
13925 (clear_counts_path): Update.
13926 (ssa_fix_duplicate_block_edges): Update.
13927 (duplicate_thread_path): Update.
13928 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13929 (struct switch_conv_info): Update.
13930 * tree-tailcall.c (decrease_profile): Update.
13931 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13932 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13933 * value-prof.c (check_counter): Update.
13934 (gimple_divmod_fixed_value): Update.
13935 (gimple_mod_pow2): Update.
13936 (gimple_mod_subtract): Update.
13937 (gimple_ic_transform): Update.
13938 (gimple_stringop_fixed_value): Update.
13939 * value-prof.h (gimple_ic): Update.
13941 2017-06-02 Carl Love <cel@us.ibm.com>
13943 * config/rs6000/rs6000-c: Add support for built-in functions
13944 vector double vec_doublee (vector signed int);
13945 vector double vec_doublee (vector unsigned int);
13946 vector double vec_doublee (vector float);
13947 vector double vec_doubleh (vector signed int);
13948 vector double vec_doubleh (vector unsigned int);
13949 vector double vec_doubleh (vector float);
13950 vector double vec_doublel (vector signed int);
13951 vector double vec_doublel (vector unsigned int);
13952 vector double vec_doublel (vector float);
13953 vector double vec_doubleo (vector signed int);
13954 vector double vec_doubleo (vector unsigned int);
13955 vector double vec_doubleo (vector float);.
13956 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13957 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13959 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13960 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13961 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13963 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13964 vec_doublel, vec_doubleh.
13965 * doc/extend.texi: Update the built-in documentation file for the
13966 new built-in functions.
13968 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13971 * ipa-inline-analysis.c (free_growth_caches): Set
13972 edge_removal_hook_holder to NULL after removing it.
13974 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13976 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13977 comparision with zero.
13979 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13981 for early expansion of vec_min and vec_max builtins.
13982 (builtin_function_type): Add min/max unsigned variants to those
13983 identified as having unsigned arguments.
13985 2017-06-02 Olivier Hainque <hainque@adacore.com>
13987 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13989 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13991 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13992 Use VALL_F16 iterator rather than VALL.
13994 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13996 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13997 Emit CBNZ inside loop when doing a strong exchange and comparing
13998 against zero. Generate the CC flags after the loop.
14000 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14002 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14003 (dl_section_ref): New.
14004 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14005 On AIX, append an expression to subtract the size of the
14006 section length to dl_section_ref.
14008 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14011 for early expansion of vector absolute builtins.
14013 2017-06-02 Richard Biener <rguenther@suse.de>
14015 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14016 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14018 2017-06-02 Richard Biener <rguenther@suse.de>
14020 PR tree-optimization/80948
14021 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14022 stmt order as well.
14024 2017-06-02 Richard Biener <rguenther@suse.de>
14026 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14028 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14029 for inductions as relevant.
14031 2017-06-02 Richard Biener <rguenther@suse.de>
14033 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14034 (vectorizable_induction): ... this. Remove dead code.
14036 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14038 * builtins. (expand_builtin_alloca): Remove second parameter and
14039 infer its value from the first parameter instead.
14040 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14042 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14044 PR rtl-optimization/80903
14045 * loop-doloop.c (add_test): Unshare sequence.
14047 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14049 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14051 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14053 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14055 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14056 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14057 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14058 (xlogue_layout::get_stub_rtx): Make static.
14059 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14060 (xlogue_layout::compute_stub_managed_regs): Rename to...
14061 (xlogue_layout::count_stub_managed_regs): ...this.
14062 (xlogue_layout::is_stub_managed_reg): New function.
14063 (xlogue_layout::m_stub_names): Rename to...
14064 (xlogue_layout::s_stub_names): ...this, make static.
14065 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14066 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14067 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14068 xlogue_layout::s_stub_names): Instantiate statics.
14069 (stub_managed_regs): Remove.
14070 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14071 (disable_call_ms2sysv_xlogues): Rename to...
14072 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14073 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14075 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14076 change after reload_completed.
14077 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14079 (ix86_expand_prologue): Likewise.
14080 (ix86_expand_epilogue): Likewise.
14081 (ix86_expand_split_stack_prologue): Likewise.
14082 (ix86_compute_frame_layout): Remove frame parameter ...
14083 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14084 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14086 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14087 (ix86_frame): Move from here ...
14088 * config/i386/i386.h (ix86_frame): ... to here.
14089 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14090 complete ix86_frame data structure instead. Remove some_ld_name.
14092 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14094 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14095 symbols that hold a DECL_VALUE_EXPR.
14097 2017-06-01 Martin Jambor <mjambor@suse.cz>
14099 PR tree-optimization/80898
14100 * tree-sra.c (process_subtree_disqualification): Removed.
14101 (disqualify_candidate): Do not acll
14102 process_subtree_disqualification.
14103 (subtree_mark_written_and_enqueue): New function.
14104 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14105 RHS has been disqualified and re-queue LHS if necessary. Apart
14106 from that, ignore disqualified RHS.
14108 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14110 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14111 address fetch for z10 or later.
14113 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14115 * config/arc/arc.md (tst_movb): Add guard when splitting.
14117 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14119 * config/arc/arc.c (arc_can_eliminate): Test against
14120 arc_frame_pointer_needed.
14122 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14124 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14125 to prevent store reordering.
14126 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14127 (type): Add block type.
14128 (stack_tie): Define special instruction to be used in
14131 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14133 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14134 constraint. It is not valid for the pattern.
14135 (noncommutative_binary_comparison): Likewise.
14137 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14139 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14142 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14144 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14145 be used by the reg-alloc.
14147 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14149 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14150 reg-alloc when having mul64 or mul32x16 instructions.
14151 (mulsidi3): Likewise.
14152 (umulsidi3): Likewise.
14153 (mulsi32x16): New pattern.
14154 (mulsi64): Likewise.
14155 (mulsidi64): Likewise.
14156 (umulsidi64): Likewise.
14157 (MUL32x16_REG): Define.
14158 (mul64_600): Use MUL32x16_REG.
14159 (mac64_600): Likewise.
14160 (umul64_600): Likewise.
14161 (umac64_600): Likewise.
14163 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14165 * config/arc/arc.md (mulsi3_700): Make it commutative.
14167 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14169 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14171 (*sign_extendsidi2_insn): Likewise for movstosw.
14173 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14175 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14176 the type of the input discriminant value. Convert the
14177 discriminant value of signedness vary.
14179 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14181 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14182 Add to -Wall section.
14184 2017-06-01 Richard Biener <rguenther@suse.de>
14186 PR middle-end/66313
14187 * fold-const.c (fold_plusminus_mult_expr): If the factored
14188 factor may be zero use a wrapping type for the inner operation.
14189 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14190 and handle moved defs.
14191 (process_assignment): Properly guard the unary op case. Return a
14192 tri-state indicating that moving the stmt before the call may allow
14193 to continue. Pass through to_move.
14194 (find_tail_calls): Handle moving unrelated defs before
14197 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14200 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14201 splitter result in the canonical way.
14203 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14205 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14206 also for 32bit target. Update insn attributes.
14207 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14209 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14211 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14212 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14213 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14214 (_mm_maskz_min_ss): New intrinsics.
14216 2017-05-31 Martin Liska <mliska@suse.cz>
14218 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14219 Change comment style to one we normally use.
14220 (vectorizable_reduction): Likewise.
14221 (vectorizable_induction): Likewise.
14222 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14223 (vectorizable_call): Likewise.
14224 (vectorizable_simd_clone_call): Likewise.
14225 (vectorizable_conversion): Likewise.
14226 (vectorizable_assignment): Likewise.
14227 (vectorizable_shift): Likewise.
14228 (vectorizable_operation): Likewise.
14229 (vectorizable_store): Likewise.
14230 (vectorizable_load): Likewise.
14231 * tree-vectorizer.h: Likewise.
14233 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14235 * passes.c (emergency_dump_function): New.
14236 * tree-pass.h (emergency_dump_function): Declare.
14237 * plugin.c (plugins_internal_error_function): Remove.
14238 * plugin.h (plugins_internal_error_function): Remove declaration.
14239 * toplev.c (internal_error_function): New static function. Use it...
14240 (general_init): ...here.
14242 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14244 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14245 (arc_rtx_costs): Add costs for new patterns.
14246 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14247 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14249 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14251 * tree-ssa-strlen.c (get_next_strinfo): New function.
14252 (get_stridx_plus_constant): Use it.
14253 (zero_length_string): Likewise.
14254 (adjust_related_strinfos): Likewise.
14255 (adjust_last_stmt): Likewise.
14257 2017-05-31 Richard Biener <rguenther@suse.de>
14260 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14261 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14263 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14265 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14266 loop_vinfo argument and use of dependence distance vectors.
14267 Check instead whether the two references differ only in their
14268 initial value and assume that they have the same alignment if the
14269 difference is a multiple of the vector alignment.
14270 (vect_analyze_data_refs_alignment): Update call accordingly.
14272 2017-05-31 Martin Liska <mliska@suse.cz>
14275 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14277 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14279 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14280 (create_intersect_range_checks): Move from ...
14281 * tree-data-ref.c (create_intersect_range_checks_index)
14282 (create_intersect_range_checks): ... to here.
14283 (create_runtime_alias_checks): New function factored from ...
14284 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14285 here. Call above function.
14286 * tree-data-ref.h (create_runtime_alias_checks): New function.
14288 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14290 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14291 segment length for dr_b and compute it in wide_int.
14293 2017-05-31 Richard Biener <rguenther@suse.de>
14295 PR tree-optimization/80906
14296 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14297 and pass through iv_map.
14298 (copy_bb_and_scalar_dependences): Adjust.
14299 (translate_pending_phi_nodes): Likewise.
14300 (copy_loop_close_phi_args): Handle code-generating IVs instead
14303 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14305 * diagnostic-color.c (color_dict): Add "type-diff".
14306 (parse_gcc_colors): Update comment.
14307 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14308 -fdiagnostics-show-template-tree and -fno-elide-type.
14309 (GCC_COLORS): Add type-diff to example.
14311 (-fdiagnostics-show-template-tree): New.
14312 (-fno-elide-type): New.
14313 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14314 the pp_format_decoder callback. Call any m_format_postprocessor's
14316 (pretty_printer::pretty_printer): Initialize
14317 m_format_postprocessor.
14318 (pretty_printer::~pretty_printer): Delete any
14319 m_format_postprocessor.
14320 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14321 (class format_postprocessor): New class.
14322 (struct pretty_printer::format_decoder): Document the new parameters.
14323 (struct pretty_printer::m_format_postprocessor): New field.
14324 * tree-diagnostic.c (default_tree_printer): Update for new
14325 bool and const char ** params.
14326 * tree-diagnostic.h (default_tree_printer): Likewise.
14328 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14330 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14331 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14332 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14333 rs6000_gen_cell_microcode code.
14334 (rs6000_final_prescan_insn): Delete.
14335 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14336 "warn-cell-microcode" entries.
14337 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14338 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14339 throughout. Change cc_reg_not_micro_cr0_operand to
14340 cc_reg_not_cr0_operand throughout.
14341 (*extendhi<mode>2_noload): Delete.
14342 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14343 (mwarn-cell-microcode): Delete.
14344 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14345 -mgen-cell-microcode and -mwarn-cell-microcode.
14347 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14350 * config/i386/constraints.md (Yd): New constraint.
14352 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14353 and (?Yd, r) alternatives. Update insn attributes.
14354 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14355 and (?*Yd, r) alternatives. Update insn attributes.
14356 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14358 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14360 * gimplify.c (gimplify_modify_expr): Don't create a
14361 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14364 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14366 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14368 2017-05-30 Richard Biener <rguenther@suse.de>
14370 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14371 and reduc_def fields.
14372 (STMT_VINFO_REDUC_TYPE): New define.
14373 (STMT_VINFO_REDUC_DEF): Likewise.
14374 (vect_force_simple_reduction): Adjust prototype.
14375 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14376 (vect_is_simple_reduction): Remove check_reduction argument.
14377 (vect_force_simple_reduction): Adjust and set
14378 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14379 (vectorizable_reduction): Do not re-do reduction analysis
14380 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14381 * tree-parloops.c (gather_scalar_reductions): Adjust.
14383 2017-05-30 Richard Biener <rguenther@suse.de>
14385 PR middle-end/80901
14386 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14389 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14391 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14392 Introduce unknown_misalignment parameter and remove vf.
14393 (vect_peeling_hash_get_lowest_cost):
14394 Pass unknown_misalignment parameter.
14395 (vect_enhance_data_refs_alignment):
14396 Fix unsupportable data ref treatment.
14398 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14400 * tree-vect-data-refs.c (vect_get_data_access_cost):
14401 Workaround for SLP handling.
14402 (vect_enhance_data_refs_alignment):
14403 Compute costs for doing no peeling at all, compare to the best
14404 peeling costs so far and avoid peeling if cheaper.
14406 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14408 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14409 Return peeling info and set costs to zero for unlimited cost
14411 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14412 with unknown misalignment. Compute and costs for unknown
14413 misalignment, compare them to the costs for known misalignment
14414 and choose the cheapest for peeling.
14416 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14418 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14419 (vect_get_peeling_costs_all_drs): Create function.
14420 (vect_peeling_hash_get_lowest_cost):
14421 Use vect_get_peeling_costs_all_drs.
14422 (vect_peeling_supportable): Create function.
14423 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14425 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14427 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14428 DR_HAS_NEGATIVE_STEP.
14429 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14430 (vect_enhance_data_refs_alignment): Use.
14431 (vect_duplicate_ssa_name_ptr_info): Use.
14432 * tree-vectorizer.h (dr_misalignment): Use.
14433 (known_alignment_for_access_p): Use.
14435 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14438 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14440 (has_section_name): New function.
14442 2017-05-30 Martin Liska <mliska@suse.cz>
14445 * auto-profile.c (get_function_decl_from_block): Fix
14448 2017-05-30 Richard Biener <rguenther@suse.de>
14450 PR middle-end/80876
14451 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14453 2017-05-30 Martin Liska <mliska@suse.cz>
14455 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14456 * dumpfile.h (struct dump_file_info): Remove ctors.
14458 2017-05-30 Martin Liska <mliska@suse.cz>
14460 * predict.def: Fix GNU coding style.
14462 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14464 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14465 Mark 'to' argument with ATTRIBUTE_UNUSED.
14467 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14469 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14470 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14471 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14474 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14476 * doc/install.texi (Options specification): Restore entry of
14477 --enable-sjlj-exceptions.
14479 2017-05-27 Michael Eager <eager@eagercon.com>
14482 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14484 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14486 * config/microblaze/microblaze.h
14487 (FIXED_REGISTERS): Update in macro.
14488 (CALL_USED_REGISTERS): Update in macro.
14490 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14492 * doc/install.texi: Add links to macOS binary distributions.
14494 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14498 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14500 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14502 2017-05-26 Martin Liska <mliska@suse.cz>
14504 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14506 2017-05-26 Martin Liska <mliska@suse.cz>
14508 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14509 always leading ';; '.
14510 (dump_bb_info): Likewise.
14511 (brief_dump_cfg): Likewise.
14512 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14513 * dumpfile.c: Remove usage of TDF_VERBOSE.
14514 * dumpfile.h (enum dump_kind): Likewise.
14515 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14516 * print-tree.c (debug_verbose): Remove.
14517 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14518 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14519 * tree-diagnostic.c (default_tree_printer): Replace
14520 TDF_DIAGNOSTIC with TDF_SLIM.
14522 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14524 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14525 in parameter loop, rather than loop_vinfo.
14526 (create_intersect_range_checks): Ditto.
14527 (vect_create_cond_for_alias_checks): Update call to above functions.
14529 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14531 PR tree-optimization/80815
14532 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14533 for merging runtime alias checks. Handle negative DR_STEPs.
14535 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14537 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14539 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14540 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14541 out code pruning runtime alias checks.
14542 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14543 factored out from above.
14544 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14546 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14548 (prune_runtime_alias_test_list): New decalaration.
14550 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14552 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14553 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14554 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14555 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14556 (operator==, comp_dr_with_seg_len_pair): Ditto.
14557 (vect_prune_runtime_alias_test_list): Ditto.
14559 2017-05-26 Martin Liska <mliska@suse.cz>
14562 * params.def: Bound partial-inlining-entry-probability param.
14564 2017-05-26 Marek Polacek <polacek@redhat.com>
14567 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14570 2017-05-26 Richard Biener <rguenther@suse.de>
14572 PR tree-optimization/80842
14573 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14576 2017-05-26 Richard Biener <rguenther@suse.de>
14578 PR tree-optimization/80844
14579 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14581 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14583 * doc/md.texi (Machine Constraints): Update x86 family
14584 machine constraints section to match 'config/i386/constraints.md'.
14586 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14588 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14590 2017-05-25 Nathan Sidwell <nathan@acm.org>
14592 * doc/invoke.texi (--enable-languages): Update documentation.
14594 2017-05-25 Martin Liska <mliska@suse.cz>
14596 * dumpfile.c: Add TDF_FOLDING.
14597 * dumpfile.h (enum dump_kind): Likewise.
14598 * genmatch.c (dt_simplify::gen_1): Use it.
14600 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14602 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14604 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14606 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14607 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14609 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14611 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14612 * match.pd (X == C): Rewrite it here.
14613 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14614 with_certain_nonzero_bits2): New predicates.
14615 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14617 2017-05-24 Nathan Sidwell <nathan@acm.org>
14619 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14622 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14625 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14627 * config/powerpcspe: New port. Files are copied from the rs6000
14628 port, with "rs6000" in filenames replaced by "powerpcspe".
14630 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14632 PR rtl-optimization/80754
14633 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14635 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14637 * config/sparc/sparc.md (length): Return the correct value for -mflat
14638 sibcalls to match output_sibcall.
14640 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14644 * config/rs6000/rs6000.c (struct machine_function): Add new field
14646 (rs6000_get_separate_components): Init that field, use it.
14647 (rs6000_components_for_bb): Use the field.
14649 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14651 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14653 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14655 PR middle-end/80823
14656 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14658 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14661 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14662 address against address_operand predicate.
14663 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14665 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14667 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14668 parameters passed indirectly.
14670 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14672 * config/i386/i386.md (*movdi_internal): Remove SSE4
14673 alternative 18 (?r, *v). Update insn attributes.
14674 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14675 Update insn attributes.
14676 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14677 Update insn attributes.
14678 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14679 alternative 1 (r, v). Remove isa attribute.
14680 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14681 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14682 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14684 2017-05-23 Tom de Vries <tom@codesourcery.com>
14686 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14689 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14691 * cgraphunit.c (symbol_table::process_new_functions): Update.
14692 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14693 (inline_generate_summary): Rename to ...
14694 (ipa_fn_summary_generate): ... this one.
14695 (inline_read_summary): Rename to ...
14696 (ipa_fn_summary_read): ... this one.
14697 (inline_write_summary): Rename to ...
14698 (ipa_fn_summary_write): ... this one.
14699 (inline_free_summary): Rename to ...
14700 (ipa_free_fn_summary): ... this one.
14701 (pass_data_local_fn_summary, pass_local_fn_summary,
14702 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14703 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14704 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14705 make_pass_ipa_fn_summary): New.
14706 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14707 inline_write_summary, inline_free_summary): Remove.
14708 (ipa_free_fn_summary) : New.
14709 * ipa-inline.c (ipa_inline): Update.
14710 (pass_ipa_inline): Do not generate summaries.
14711 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14713 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14714 and add pass_ipa_fn_summary.
14715 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14717 (make_pass_inline_parameters): Remove.
14719 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14721 * omp-low.c (struct omp_context): Remove "default_kind" member.
14724 * omp-offload.c (execute_oacc_device_lower): Remove the
14725 parallelism dimensions function attributes for unparallelized
14726 OpenACC kernels constructs.
14728 2017-05-23 Martin Liska <mliska@suse.cz>
14730 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14732 (cgraph_edge::make_speculative): Likewise.
14733 (cgraph_edge::resolve_speculation): Likewise.
14734 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14735 (cgraph_node::dump): Likewise.
14736 * cgraph.h: Likewise.
14737 * cgraphunit.c (analyze_functions): Likewise.
14738 (symbol_table::compile): Likewise.
14739 * ipa-cp.c (print_all_lattices): Likewise.
14740 (determine_versionability): Likewise.
14741 (initialize_node_lattices): Likewise.
14742 (ipcp_verify_propagated_values): Likewise.
14743 (estimate_local_effects): Likewise.
14744 (update_profiling_info): Likewise.
14745 (create_specialized_node): Likewise.
14746 (perhaps_add_new_callers): Likewise.
14747 (decide_about_value): Likewise.
14748 (decide_whether_version_node): Likewise.
14749 (identify_dead_nodes): Likewise.
14750 (ipcp_store_bits_results): Likewise.
14751 * ipa-devirt.c (dump_targets): Likewise.
14752 (ipa_devirt): Likewise.
14753 * ipa-icf.c (sem_item::dump): Likewise.
14754 (sem_function::equals): Likewise.
14755 (sem_variable::equals): Likewise.
14756 (sem_item_optimizer::read_section): Likewise.
14757 (sem_item_optimizer::execute): Likewise.
14758 (congruence_class::dump): Likewise.
14759 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14760 (dump_inline_summary): Likewise.
14761 (estimate_node_size_and_time): Likewise.
14762 (inline_analyze_function): Likewise.
14763 * ipa-inline-transform.c (inline_call): Likewise.
14764 * ipa-inline.c (report_inline_failed_reason): Likewise.
14765 (want_early_inline_function_p): Likewise.
14766 (edge_badness): Likewise.
14767 (update_edge_key): Likewise.
14768 (inline_small_functions): Likewise.
14769 * ipa-profile.c (ipa_profile): Likewise.
14770 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14771 (ipa_make_edge_direct_to_target): Likewise.
14772 (remove_described_reference): Likewise.
14773 (ipa_impossible_devirt_target): Likewise.
14774 (propagate_controlled_uses): Likewise.
14775 (ipa_print_node_params): Likewise.
14776 (ipcp_transform_function): Likewise.
14777 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14778 (propagate_pure_const): Likewise.
14779 * ipa-reference.c (generate_summary): Likewise.
14780 (read_write_all_from_decl): Likewise.
14781 (propagate): Likewise.
14782 (ipa_reference_read_optimization_summary): Likewise.
14783 * ipa-utils.c (ipa_merge_profiles): Likewise.
14784 * ipa.c (walk_polymorphic_call_targets): Likewise.
14785 (symbol_table::remove_unreachable_nodes): Likewise.
14786 (ipa_single_use): Likewise.
14787 * passes.c (execute_todo): Likewise.
14788 * predict.c (drop_profile): Likewise.
14789 * symtab.c (symtab_node::get_dump_name): New function.
14790 (symtab_node::dump_name): Likewise.
14791 (symtab_node::dump_asm_name): Likewise.
14792 (symtab_node::dump_references): Likewise.
14793 (symtab_node::dump_referring): Likewise.
14794 (symtab_node::dump_base): Likewise.
14795 (symtab_node::debug_symtab): Likewise.
14796 * tree-sra.c (convert_callers_for_node): Likewise.
14797 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14798 * value-prof.c (init_node_map): Likewise.
14800 2017-05-23 Martin Liska <mliska@suse.cz>
14802 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14803 and symtab_node::debug_symtab to symbol_table::debug.
14804 * cgraphunit.c (analyze_functions): Use the renamed function.
14805 (symbol_table::compile): Likewise.
14806 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14807 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14808 * passes.c (execute_todo): Likewise.
14809 * symtab.c (symbol_table::dump): New function.
14810 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14812 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14814 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14815 that nonconst implies exec.
14817 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14819 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14820 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14821 (inline_edge_summary_vec): Turn into ...
14822 (ipa_call_summaries): ... this one.
14823 (redirect_to_unreachable, edge_set_predicate,
14824 evaluate_properties_for_edge, inline_summary_alloc,
14825 reset_ipa_call_summary, reset_inline_summary,
14826 inline_summary_t::duplicate): Update.
14827 (inline_edge_duplication_hook): Turn to ...
14828 (ipa_call_summary_t::duplicate): ... this one.
14829 (inline_edge_removal_hook): Turn to ...
14830 (ipa_call_summary_t::remove): ... this one.
14831 (dump_inline_edge_summary): Turn to ...
14832 (dump_ipa_call_summary): ... this one.
14833 (estimate_function_body_sizes): Update.
14834 (inline_update_callee_summaries): Update.
14835 (remap_edge_change_prob): Update.
14836 (remap_edge_summaries): Update.
14837 (inline_merge_summary): Update.
14838 (do_estimate_edge_time): Update.
14839 (inline_generate_summary): Update.
14840 (inline_read_section): Update.
14841 (inline_read_summary): Update.
14842 (inline_free_summary): Update.
14843 * ipa-inline.c (can_inline_edge_p): Update.
14844 (compute_inlined_call_time): Update.
14845 (want_inline_small_function_p): Update.
14846 (edge_badness): Update.
14847 (early_inliner): Update.
14848 * ipa-inline.h (inline_edge_summary): Turn to ...
14849 (ipa_call_summary): ... this one.
14850 (ipa_call_summary_t): New class.
14851 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14852 (ipa_call_summaries): New.
14853 (inline_edge_summary): Remove.
14854 (estimate_edge_growth): Update.
14855 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14856 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14857 * ipa-split.c (execute_split_functions): Update.
14858 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14860 2017-05-23 Tom de Vries <tom@codesourcery.com>
14862 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14863 attributes): Document rdrand effective target.
14865 2017-05-23 Tom de Vries <tom@codesourcery.com>
14867 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14868 attributes): Sort alphabetically.
14870 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14872 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14874 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14877 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14878 V2DF/V2DI splat into two separate patterns, one that handles
14879 registers, and the other that only handles memory. Drop support
14880 for splatting from a GPR on ISA 2.07 and then splitting the
14881 splat into direct move and splat.
14882 (vsx_splat_<mode>_reg): Likewise.
14883 (vsx_splat_<mode>_mem): Likewise.
14885 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14887 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14889 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14891 PR middle-end/80809
14892 * omp-low.c (finish_taskreg_remap): New function.
14893 (finish_taskreg_scan): If unit size of ctx->record_type
14894 is non-constant, unshare the size expression and replace
14895 decls in it with possible outer var refs.
14897 PR middle-end/80809
14898 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14899 GOVD_SHARED rather than GOVD_PRIVATE with it.
14900 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14901 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14903 PR middle-end/80853
14904 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14905 as last argument to build_outer_var_ref for pointer bases of array
14906 section reductions.
14908 2017-05-19 Martin Sebor <msebor@redhat.com>
14910 * print-tree.c (print_node): Print DECL_READ_P flag.
14912 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14914 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14915 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14916 * cgraph.c: Likewise.
14917 * cgraphunit.c: Likewise.
14918 * gengtype.c: Likewise.
14919 * ipa-cp.c: Likewise.
14920 * ipa-devirt.c: Likewise.
14921 * ipa-icf.c: Likewise.
14922 * ipa-predicate.c: Likewise.
14923 * ipa-profile.c: Likewise.
14924 * ipa-prop.c: Likewise.
14925 * ipa-split.c: Likewise.
14927 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14928 edge_predicate_pool, dump_inline_hints,
14929 inline_summary::account_size_time, redirect_to_unreachable,
14930 edge_set_predicate, set_hint_predicate,
14931 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14932 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14933 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14934 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14935 ipa_call_summary_t::remove, initialize_growth_caches,
14936 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14937 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14938 mark_modified, unmodified_parm_1, unmodified_parm,
14939 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14940 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14941 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14942 will_be_nonconstant_predicate, record_modified_bb_info,
14943 get_minimal_bb, record_modified, param_change_prob,
14944 phi_result_unknown_predicate, predicate_for_phi_result,
14945 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14946 estimate_function_body_sizes, compute_inline_parameters,
14947 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14948 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14949 inline_update_callee_summaries, remap_edge_change_prob,
14950 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14951 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14952 inline_analyze_function, inline_summary_t::insert,
14953 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14954 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14955 inline_free_summary): Move to ipa-fnsummary.h
14956 (predicate_t): Remove.
14957 * ipa-fnsummary.c: New file.
14958 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14959 (enum inline_hints_vals, inline_hints, agg_position_info,
14960 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14961 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14962 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14963 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14964 inline_read_summary, inline_write_summary, inline_free_summary,
14965 inline_analyze_function, initialize_inline_failed,
14966 inline_merge_summary, inline_update_overall_summary,
14967 compute_inline_parameters): Move to ipa-fnsummary.h
14968 * ipa-fnsummary.h: New file.
14969 * ipa-inline-transform.h: Include ipa-inline.h.
14970 * ipa-inline.c: LIkewise.
14972 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14974 * ipa-inline.c (edge_badness): Use inlined_time instead of
14975 inline_summaries->get.
14977 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14979 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14981 2017-05-22 Nathan Sidwell <nathan@acm.org>
14983 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14984 (fdump-lang): Document 'raw' option.
14985 * dumpfile.h (TDI_tu): Delete.
14986 * dumpfile.c (dump_files): Remove translation-unit.
14987 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14989 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14991 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14992 command option from $(AWK) call.
14993 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14994 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14995 [FORMAT]: Remove handling of variable.
14996 * config/avr/t-multilib: Regenerate.
14998 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15000 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15002 (dump_inline_summary): Do not print self_time.
15003 (estimate_function_body_sizes): Do not set self_time.
15004 (compute_inline_parameters): Likewise.
15005 (inline_read_section, inline_write_summary): Do not stream self_time.
15006 * ipa-inline.h (inline_summary): Drop self_time.
15008 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15010 * ipa-inline-analysis.c (account_size_time): Rename to ...
15011 (inline_summary::account_size_time): ... this one.
15012 (reset_ipa_call_summary): Turn to ...
15013 (ipa_call_summary::reset): ... this one.
15014 (reset_inline_summary): Turn to ...
15015 (inline_summary::reset): ... this one.
15016 (inline_summary_t::remove): Update.
15017 (inline_summary_t::duplicate): Update.
15018 (ipa_call_summary_t::remove): Update.
15019 (dump_inline_summary): Update.
15020 (estimate_function_body_sizes): Update.
15021 (compute_inline_parameters): Update.
15022 (estimate_node_size_and_time): Update.
15023 (inline_merge_summary): Update.
15024 (inline_update_overall_summary): Update.
15025 (inline_read_section): Update.
15026 (inline_write_summary): Update.
15027 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15028 add account_size_time and reset member functions.
15029 (ipa_call_summary): Add reset function.
15030 * ipa-predicate.h (predicate::operator &): Constify.
15032 2017-05-22 Richard Biener <rguenther@suse.de>
15034 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15036 2017-05-19 Jason Merrill <jason@redhat.com>
15038 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15040 2017-05-19 Marek Polacek <polacek@redhat.com>
15043 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15044 TYPE_OVERFLOW_WRAPS checks.
15046 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15048 * tree-core.h (enum omp_clause_default_kind): Add
15049 "OMP_CLAUSE_DEFAULT_PRESENT".
15050 * tree-pretty-print.c (dump_omp_clause): Handle it.
15051 * gimplify.c (enum gimplify_omp_var_data): Add
15052 "GOVD_MAP_FORCE_PRESENT".
15053 (gimplify_adjust_omp_clauses_1): Map it to
15054 "GOMP_MAP_FORCE_PRESENT".
15055 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15057 * gimplify.c (oacc_default_clause): Clarify.
15059 2017-05-19 Nathan Sidwell <nathan@acm.org>
15061 LANG_HOOK_REGISTER_DUMPS
15062 * toplev.c (general_init): Call register dump lang hook.
15063 * doc/invoke.texi: Document -fdump-lang option family.
15064 * dumpfile.c (dump_files): Remove class dump here.
15065 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15066 * dumpfile.h (tree_dump_index): Remove TDI_class.
15067 * langhooks-def.h (lhd_register_dumps): Declare.
15068 (LANG_HOOKS_REGISTER_DUMPS): Define.
15069 (LANG_HOOKS_INITIALIZER): Add it.
15070 * langhooks.c (lhd_register_dumps): Define.
15071 * langhooks.h (struct lang_hooks): Add register_dumps.
15073 2017-05-19 Nathan Sidwell <nathan@acm.org>
15075 * context.h (context::set_passes): New.
15076 * context.c (context::context): Do not create pass manager.
15077 * toplev.c (general_init): Create pass manager here.
15079 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15081 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15082 use this splitter if two add or or instructions would also work for
15083 the constant we want to generate.
15085 2017-05-19 Richard Biener <rguenther@suse.de>
15088 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15089 predicate evaluation.
15091 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15093 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15095 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15098 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15100 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15101 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15102 (inline_edge_summary_vec): Turn into ...
15103 (ipa_call_summaries): ... this one.
15104 (redirect_to_unreachable, edge_set_predicate,
15105 evaluate_properties_for_edge, inline_summary_alloc,
15106 reset_ipa_call_summary, reset_inline_summary,
15107 inline_summary_t::duplicate): Update.
15108 (inline_edge_duplication_hook): Turn to ...
15109 (ipa_call_summary_t::duplicate): ... this one.
15110 (inline_edge_removal_hook): Turn to ...
15111 (ipa_call_summary_t::remove): ... this one.
15112 (dump_inline_edge_summary): Turn to ...
15113 (dump_ipa_call_summary): ... this one.
15114 (estimate_function_body_sizes): Update.
15115 (inline_update_callee_summaries): Update.
15116 (remap_edge_change_prob): Update.
15117 (remap_edge_summaries): Update.
15118 (inline_merge_summary): Update.
15119 (do_estimate_edge_time): Update.
15120 (inline_generate_summary): Update.
15121 (inline_read_section): Update.
15122 (inline_read_summary): Update.
15123 (inline_free_summary): Update.
15124 * ipa-inline.c (can_inline_edge_p): Update.
15125 (compute_inlined_call_time): Update.
15126 (want_inline_small_function_p): Update.
15127 (edge_badness): Update.
15128 (early_inliner): Update.
15129 * ipa-inline.h (inline_edge_summary): Turn to ...
15130 (ipa_call_summary): ... this one.
15131 (ipa_call_summary_t): New class.
15132 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15133 (ipa_call_summaries): New.
15134 (inline_edge_summary): Remove.
15135 (estimate_edge_growth): Update.
15136 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15137 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15138 * ipa-split.c (execute_split_functions): Update.
15139 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15141 2017-05-19 Richard Biener <rguenther@suse.de>
15143 PR middle-end/80764
15144 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15146 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15148 * config/rs6000/rs6000.c (struct machine_function): Add field
15149 fpr_is_wrapped_separately.
15150 (rs6000_get_separate_components): Use 64 components. Handle the
15151 new FPR components.
15152 (rs6000_components_for_bb): Handle the FPR components.
15153 (rs6000_emit_prologue_components): Handle the FPR components.
15154 (rs6000_emit_epilogue_components): Handle the FPR components.
15155 (rs6000_set_handled_components): Handle the FPR components.
15156 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15157 that are already separately shrink-wrapped.
15158 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15159 that are already separately shrink-wrapped.
15161 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15164 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15167 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15168 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15169 cases where the register allocator uses a move and an offsettable
15170 memory operation to/from a FPR register on ISA 2.06/2.07.
15171 (define_peephole2 for Altivec d-form store): Likewise.
15173 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15176 * config/i386/mmx.md (*mov<mode>_internal): Enable
15177 alternatives 11, 12, 13 and 14 also for 32bit targets.
15178 Remove alternatives 15, 16, 17 and 18.
15179 * config/i386/sse.md (vec_concatv2di): Change
15180 alternative (!x, *y) to (x, ?!*Yn).
15182 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15184 * dumpfile.h (enum dump_kind): Remove stray comma.
15186 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15188 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15189 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15190 predicate::num_conditions
15191 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15192 (CHANGED): turn into predicate::changed.
15193 (agg_position_info): Move to ipa-predicate.h
15194 (add_condition, predicate::add_clause, predicate::operator &=,
15195 predicate::or_with, predicate::evaluate, predicate::probability,
15196 dump_condition, dump_clause, predicate::dump,
15197 predicate::remap_after_duplication, predicate::remap_after_inlining,
15198 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15199 (evaluate_conditions_for_known_args): Update.
15200 (set_cond_stmt_execution_predicate): Update.
15201 * ipa-inline.h: Include ipa-predicate.h
15202 (condition, inline_param_summary, conditions, agg_position_info,
15203 predicate): Move to ipa-predicate.h
15204 * ipa-predicate.c: New file.
15205 * ipa-predicate.h: New file.
15207 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15209 * final.c (leaf_function_p): Check we are not in a sequence.
15211 2017-05-18 Martin Liska <mliska@suse.cz>
15213 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15214 * dumpfile.c (dump_register): Use new enum dump_kind.
15215 (get_dump_file_name): Likewise.
15216 (dump_enable_all): Likewise.
15217 (dump_switch_p_1): Likewise.
15218 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15219 * dumpfile.h (enum dump_kind): New enum type.
15220 (struct dump_file_info): Create constructor and
15221 format fields and comments.
15222 * passes.c (pass_manager::register_one_dump_file):
15224 * statistics.c (statistics_early_init): Likewise.
15225 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15226 TDF_TREE with TDF_SLIM.
15227 (gather_memory_references_ref): Likewise.
15229 2017-05-18 Martin Liska <mliska@suse.cz>
15231 * vec.h (struct vnull): Use it.
15233 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15235 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15236 (true_predicate, false_predicate, true_predicate_p,
15237 false_predicate_p): Remove.
15238 (single_cond_predicate, not_inlined_predicate): Turn to member function
15240 (add_condition): Update.
15241 (add_clause): Turn to...
15242 (predicate::add_clause): ... this one; update; allow passing NULL
15244 (and_predicates): Turn to ...
15245 (predicate::operator &=): ... this one.
15246 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15247 (or_predicates): Turn to ...
15248 (predicate::or_with): ... this one.
15249 (evaluate_predicate): Turn to ...
15250 (predicate::evaluate): ... this one.
15251 (predicate_probability): Turn to ...
15252 (predicate::probability): ... this one.
15253 (dump_condition): Update.
15254 (dump_predicate): Turn to ...
15255 (predicate::dump): ... this one.
15256 (account_size_time): Update.
15257 (edge_set_predicate): Update.
15258 (set_hint_predicate): UPdate.
15259 (evaluate_conditions_for_known_args): Update.
15260 (evaluate_properties_for_edge): Update.
15261 (remap_predicate_after_duplication): Turn to...
15262 (predicate::remap_after_duplication): ... this one.
15263 (remap_hint_predicate_after_duplication): Update.
15264 (inline_summary_t::duplicate): UPdate.
15265 (dump_inline_edge_summary): Update.
15266 (dump_inline_summary): Update.
15267 (set_cond_stmt_execution_predicate): Update.
15268 (set_switch_stmt_execution_predicate): Update.
15269 (compute_bb_predicates): Update.
15270 (will_be_nonconstant_expr_predicate): Update.
15271 (will_be_nonconstant_predicate): Update.
15272 (phi_result_unknown_predicate): Update.
15273 (predicate_for_phi_result): Update.
15274 (array_index_predicate): Update.
15275 (estimate_function_body_sizes): Update.
15276 (estimate_node_size_and_time): Update.
15277 (estimate_ipcp_clone_size_and_time): Update.
15278 (remap_predicate): Rename to ...
15279 (predicate::remap_after_inlining): ... this one.
15280 (remap_hint_predicate): Update.
15281 (inline_merge_summary): Update.
15282 (inline_update_overall_summary): Update.
15283 (estimate_size_after_inlining): Update.
15284 (read_predicate): Rename to ...
15285 (predicate::stream_in): ... this one.
15286 (read_inline_edge_summary): Update.
15287 (write_predicate): Rename to ...
15288 (predicate::stream_out): ... this one.
15289 (write_inline_edge_summary): Update.
15290 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15291 (clause_t): Turn to uint32_t
15292 (predicate): Turn to class; implement constructor and operators
15294 (size_time_entry): Update.
15295 (inline_summary): Update.
15296 (inline_edge_summary): Update.
15298 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15300 * fold-const.c (fold_binary_loc): Move transformation...
15301 * match.pd (C - X CMP X): ... here.
15303 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15305 * config/sparc/sparc.c (sparc_option_override): Set function
15306 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15307 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15309 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15311 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15313 2017-05-18 Marek Polacek <polacek@redhat.com>
15316 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15317 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15318 gimple_assign_load_p.
15320 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15322 PR middle-end/80692
15323 * real.c (do_compare): Give decimal_do_compare preference over
15324 comparing just the signs.
15326 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15328 * doc/md.texi (Canonicalization of Instructions): Describe the
15329 canonical form of instructions that inherently set a condition
15332 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15334 PR middle-end/80775
15335 * tree-cfg.c: Move deletion of unreachable case statements to after
15336 the merging of consecutive case labels.
15338 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15340 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15341 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15342 restoring of callee-saved registers.
15344 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15346 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15347 * config/visium/visium.c (single_set_and_flags): Likewise.
15348 * config/visium/visium.md (Substitutions): Likewise.
15350 2017-05-17 Martin Liska <mliska@suse.cz>
15352 * cfg.c: Introduce dump_flags_t type and
15353 use it instead of int type.
15355 * cfghooks.c: Likewise.
15356 * cfghooks.h (struct cfg_hooks): Likewise.
15357 * cfgrtl.c: Likewise.
15358 * cfgrtl.h: Likewise.
15359 * cgraph.c (cgraph_node::get_body): Likewise.
15360 * coretypes.h: Likewise.
15361 * domwalk.c: Likewise.
15362 * domwalk.h: Likewise.
15363 * dumpfile.c (struct dump_option_value_info): Likewise.
15364 (dump_enable_all): Likewise.
15365 (dump_switch_p_1): Likewise.
15366 (opt_info_switch_p): Likewise.
15367 * dumpfile.h (enum tree_dump_index): Likewise.
15368 (struct dump_file_info): Likewise.
15369 * genemit.c: Likewise.
15370 * generic-match-head.c: Likewise.
15371 * gengtype.c (open_base_files): Likewise.
15372 * gimple-pretty-print.c: Likewise.
15373 * gimple-pretty-print.h: Likewise.
15374 * graph.c (print_graph_cfg): Likewise.
15375 * graphite-scop-detection.c (dot_all_sese): Likewise.
15376 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15377 * loop-unroll.c (report_unroll): Likewise.
15378 * passes.c (pass_manager::register_one_dump_file): Likewise.
15379 * print-tree.c: Likewise.
15380 * statistics.c: Likewise.
15381 * tree-cfg.c: Likewise.
15382 * tree-cfg.h: Likewise.
15383 * tree-dfa.c: Likewise.
15384 * tree-dfa.h: Likewise.
15385 * tree-dump.c (dump_function): Likewise.
15386 * tree-dump.h (struct dump_info): Likewise.
15387 * tree-pretty-print.c: Likewise.
15388 * tree-pretty-print.h: Likewise.
15389 * tree-ssa-live.c: Likewise.
15390 * tree-ssa-live.h: Likewise.
15391 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15392 * tree-vect-loop.c: Likewise.
15393 * tree-vect-slp.c: Likewise.
15395 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15396 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15398 PR tree-optimization/80457
15399 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15400 of all arguments to a statement as scalar_to_vec operations.
15401 (vectorizable_call): Adjust call to vect_model_simple_cost for
15403 (vectorizable_conversion): Likewise.
15404 (vectorizable_assignment): Likewise.
15405 (vectorizable_shift): Likewise.
15406 (vectorizable_operation): Likewise.
15407 (vectorizable_comparison): Likewise.
15408 (vect_is_simple_cond): Record the def types for operands.
15409 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15410 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15411 for statement argument count.
15413 2017-05-16 Carl Love <cel@us.ibm.com>
15415 * config/rs6000/rs6000-c: Add support for built-in functions
15416 vector unsigned long long vec_bperm (vector unsigned long long,
15417 vector unsigned char)
15418 vector signed long long vec_mule (vector signed int,
15420 vector unsigned long long vec_mule (vector unsigned int,
15421 vector unsigned int)
15422 vector signed long long vec_mulo (vector signed int,
15424 vector unsigned long long vec_mulo (vector unsigned int,
15425 vector unsigned int)
15426 vector signed char vec_sldw (vector signed char,
15427 vector signed char,
15429 vector unsigned char vec_sldw (vector unsigned char,
15430 vector unsigned char,
15432 vector signed short vec_sldw (vector signed short,
15433 vector signed short,
15435 vector unsigned short vec_sldw (vector unsigned short,
15436 vector unsigned short,
15438 vector signed int vec_sldw (vector signed int,
15441 vector unsigned int vec_sldw (vector unsigned int,
15442 vector unsigned int,
15444 vector signed long long vec_sldw (vector signed long long,
15445 vector signed long long,
15447 vector unsigned long long vec_sldw (vector unsigned long long,
15448 vector unsigned long long,
15450 * config/rs6000/rs6000-c: Add support for built-in functions
15451 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15452 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15453 * doc/extend.texi: Update the built-in documentation for the
15454 new built-in functions.
15456 2017-05-16 Marek Polacek <polacek@redhat.com>
15460 * tree.c (save_expr): Don't fold the expression.
15462 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15464 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15465 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15466 and (?*y,m). Update insn attributes.
15468 2017-05-16 Martin Liska <mliska@suse.cz>
15470 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15471 flags argument of print_gimple_stmt, print_gimple_expr,
15472 print_generic_stmt and print_generic_expr.
15473 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15474 * coretypes.h: Likewise.
15475 * except.c (dump_eh_tree): Likewise.
15476 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15477 * gimple-pretty-print.h: Likewise.
15478 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15479 (backprop::push_to_worklist): Likewise.
15480 (backprop::pop_from_worklist): Likewise.
15481 (backprop::process_use): Likewise.
15482 (backprop::intersect_uses): Likewise.
15483 (note_replacement): Likewise.
15484 * gimple-ssa-store-merging.c
15485 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15486 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15487 (pass_store_merging::execute): Likewise.
15488 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15489 (ssa_base_cand_dump_callback): Likewise.
15490 (dump_incr_vec): Likewise.
15491 (replace_refs): Likewise.
15492 (replace_mult_candidate): Likewise.
15493 (create_add_on_incoming_edge): Likewise.
15494 (create_phi_basis): Likewise.
15495 (insert_initializers): Likewise.
15496 (all_phi_incrs_profitable): Likewise.
15497 (introduce_cast_before_cand): Likewise.
15498 (replace_one_candidate): Likewise.
15499 * gimplify.c (gimplify_expr): Likewise.
15500 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15501 (set_rename): Likewise.
15502 (rename_uses): Likewise.
15503 (copy_loop_phi_nodes): Likewise.
15504 (add_close_phis_to_merge_points): Likewise.
15505 (copy_loop_close_phi_args): Likewise.
15506 (copy_cond_phi_args): Likewise.
15507 (graphite_copy_stmts_from_block): Likewise.
15508 (translate_pending_phi_nodes): Likewise.
15509 * graphite-poly.c (print_pdr): Likewise.
15510 (dump_gbb_cases): Likewise.
15511 (dump_gbb_conditions): Likewise.
15512 (print_scop_params): Likewise.
15513 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15514 (build_cross_bb_scalars_use): Likewise.
15515 (gather_bbs::before_dom_children): Likewise.
15516 * hsa-dump.c (dump_hsa_immed): Likewise.
15517 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15518 (get_replacement_map): Likewise.
15519 * ipa-inline-analysis.c (dump_condition): Likewise.
15520 (estimate_function_body_sizes): Likewise.
15521 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15522 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15523 * ipa-prop.c (ipa_dump_param): Likewise.
15524 (ipa_print_node_jump_functions_for_edge): Likewise.
15525 (ipa_modify_call_arguments): Likewise.
15526 (ipa_modify_expr): Likewise.
15527 (ipa_dump_param_adjustments): Likewise.
15528 (ipa_dump_agg_replacement_values): Likewise.
15529 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15530 * ipa-pure-const.c (check_stmt): Likewise.
15531 (pass_nothrow::execute): Likewise.
15532 * ipa-split.c (execute_split_functions): Likewise.
15533 * omp-offload.c (dump_oacc_loop_part): Likewise.
15534 (dump_oacc_loop): Likewise.
15535 * trans-mem.c (tm_log_emit): Likewise.
15536 (tm_memopt_accumulate_memops): Likewise.
15537 (dump_tm_memopt_set): Likewise.
15538 (dump_tm_memopt_transform): Likewise.
15539 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15540 (print_loop): Likewise.
15541 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15542 (chkp_gather_checks_info): Likewise.
15543 (chkp_get_check_result): Likewise.
15544 (chkp_remove_check_if_pass): Likewise.
15545 (chkp_use_outer_bounds_if_possible): Likewise.
15546 (chkp_reduce_bounds_lifetime): Likewise.
15547 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15548 (chkp_mark_completed_bounds): Likewise.
15549 (chkp_register_incomplete_bounds): Likewise.
15550 (chkp_mark_invalid_bounds): Likewise.
15551 (chkp_maybe_copy_and_register_bounds): Likewise.
15552 (chkp_build_returned_bound): Likewise.
15553 (chkp_get_bound_for_parm): Likewise.
15554 (chkp_build_bndldx): Likewise.
15555 (chkp_get_bounds_by_definition): Likewise.
15556 (chkp_generate_extern_var_bounds): Likewise.
15557 (chkp_get_bounds_for_decl_addr): Likewise.
15558 * tree-chrec.c (chrec_apply): Likewise.
15559 * tree-data-ref.c (dump_data_reference): Likewise.
15560 (dump_subscript): Likewise.
15561 (dump_data_dependence_relation): Likewise.
15562 (analyze_overlapping_iterations): Likewise.
15563 * tree-inline.c (expand_call_inline): Likewise.
15564 (tree_function_versioning): Likewise.
15565 * tree-into-ssa.c (dump_defs_stack): Likewise.
15566 (dump_currdefs): Likewise.
15567 (dump_names_replaced_by): Likewise.
15568 (dump_update_ssa): Likewise.
15569 (update_ssa): Likewise.
15570 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15571 * tree-parloops.c (build_new_reduction): Likewise.
15572 (try_create_reduction_list): Likewise.
15573 (ref_conflicts_with_region): Likewise.
15574 (oacc_entry_exit_ok_1): Likewise.
15575 (oacc_entry_exit_single_gang): Likewise.
15576 * tree-pretty-print.h: Likewise.
15577 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15578 (get_scalar_evolution): Likewise.
15579 (add_to_evolution): Likewise.
15580 (get_loop_exit_condition): Likewise.
15581 (analyze_evolution_in_loop): Likewise.
15582 (analyze_initial_condition): Likewise.
15583 (analyze_scalar_evolution): Likewise.
15584 (instantiate_scev): Likewise.
15585 (number_of_latch_executions): Likewise.
15586 (gather_chrec_stats): Likewise.
15587 (final_value_replacement_loop): Likewise.
15588 (scev_const_prop): Likewise.
15589 * tree-sra.c (dump_access): Likewise.
15590 (disqualify_candidate): Likewise.
15591 (create_access): Likewise.
15592 (reject): Likewise.
15593 (maybe_add_sra_candidate): Likewise.
15594 (create_access_replacement): Likewise.
15595 (analyze_access_subtree): Likewise.
15596 (analyze_all_variable_accesses): Likewise.
15597 (sra_modify_assign): Likewise.
15598 (initialize_constant_pool_replacements): Likewise.
15599 (find_param_candidates): Likewise.
15600 (decide_one_param_reduction): Likewise.
15601 (replace_removed_params_ssa_names): Likewise.
15602 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15603 * tree-ssa-copy.c (dump_copy_of): Likewise.
15604 (copy_prop_visit_cond_stmt): Likewise.
15605 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15606 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15607 (record_equivalences_from_stmt): Likewise.
15608 * tree-ssa-dse.c (compute_trims): Likewise.
15609 (delete_dead_call): Likewise.
15610 (delete_dead_assignment): Likewise.
15611 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15612 (forward_propagate_into_cond): Likewise.
15613 (pass_forwprop::execute): Likewise.
15614 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15615 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15617 (move_computations_worker): Likewise.
15618 (execute_sm): Likewise.
15619 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15620 (remove_exits_and_undefined_stmts): Likewise.
15621 (remove_redundant_iv_tests): Likewise.
15622 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15623 (adjust_iv_update_pos): Likewise.
15624 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15625 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15626 (value_replacement): Likewise.
15627 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15628 * tree-ssa-pre.c (print_pre_expr): Likewise.
15629 (get_representative_for): Likewise.
15630 (create_expression_by_pieces): Likewise.
15631 (insert_into_preds_of_block): Likewise.
15632 (eliminate_insert): Likewise.
15633 (eliminate_dom_walker::before_dom_children): Likewise.
15634 (eliminate): Likewise.
15635 (remove_dead_inserted_code): Likewise.
15636 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15637 * tree-ssa-reassoc.c (get_rank): Likewise.
15638 (eliminate_duplicate_pair): Likewise.
15639 (eliminate_plus_minus_pair): Likewise.
15640 (eliminate_not_pairs): Likewise.
15641 (undistribute_ops_list): Likewise.
15642 (eliminate_redundant_comparison): Likewise.
15643 (update_range_test): Likewise.
15644 (optimize_range_tests_var_bound): Likewise.
15645 (optimize_vec_cond_expr): Likewise.
15646 (rewrite_expr_tree): Likewise.
15647 (rewrite_expr_tree_parallel): Likewise.
15648 (linearize_expr): Likewise.
15649 (break_up_subtract): Likewise.
15650 (linearize_expr_tree): Likewise.
15651 (attempt_builtin_powi): Likewise.
15652 (attempt_builtin_copysign): Likewise.
15653 (transform_stmt_to_copy): Likewise.
15654 (transform_stmt_to_multiply): Likewise.
15655 (dump_ops_vector): Likewise.
15656 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15657 (print_scc): Likewise.
15658 (set_ssa_val_to): Likewise.
15659 (visit_reference_op_store): Likewise.
15660 (visit_use): Likewise.
15661 (sccvn_dom_walker::before_dom_children): Likewise.
15662 (run_scc_vn): Likewise.
15663 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15665 (expr_hash_elt::print): Likewise.
15666 (const_and_copies::pop_to_marker): Likewise.
15667 (const_and_copies::record_const_or_copy_raw): Likewise.
15668 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15669 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15670 (dump_predicates): Likewise.
15671 (find_uninit_use): Likewise.
15672 (warn_uninitialized_phi): Likewise.
15673 (pass_late_warn_uninitialized::execute): Likewise.
15674 * tree-ssa.c (verify_vssa): Likewise.
15675 (verify_ssa): Likewise.
15676 (maybe_optimize_var): Likewise.
15677 * tree-vrp.c (dump_value_range): Likewise.
15678 (dump_all_value_ranges): Likewise.
15679 (dump_asserts_for): Likewise.
15680 (register_edge_assert_for_2): Likewise.
15681 (vrp_visit_cond_stmt): Likewise.
15682 (vrp_visit_switch_stmt): Likewise.
15683 (vrp_visit_stmt): Likewise.
15684 (vrp_visit_phi_node): Likewise.
15685 (simplify_cond_using_ranges_1): Likewise.
15686 (fold_predicate_in): Likewise.
15687 (evrp_dom_walker::before_dom_children): Likewise.
15688 (evrp_dom_walker::push_value_range): Likewise.
15689 (evrp_dom_walker::pop_value_range): Likewise.
15690 (execute_early_vrp): Likewise.
15692 2017-05-16 Richard Biener <rguenther@suse.de>
15694 * dwarf2out.c (loc_list_from_tree_1): Do not create
15695 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15697 2017-05-16 Richard Biener <rguenther@suse.de>
15699 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15701 (note_variable_value_in_expr): If we resolved the decl ref
15702 do not push to the stack.
15704 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15706 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15707 operations in fast-math mode.
15708 (vaddq_f16): Likewise.
15709 (vmul_f16): Likewise.
15710 (vmulq_f16): Likewise.
15711 (vsub_f16): Likewise.
15712 (vsubq_f16): Likewise.
15713 * config/arm/neon.md (add<mode>3): New.
15715 (fma:<VH:mode>3): New. Also remove outdated comment.
15718 2017-05-16 Martin Liska <mliska@suse.cz>
15722 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15723 (odr_types_equivalent_p): Likewise.
15725 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15727 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15729 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15732 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15733 non-interunit SSE move alternatives with '?'.
15734 (zero-extendsidi peephole2): New peephole to skip intermediate
15735 general register in SSE zero-extend sequence.
15737 2017-05-15 Jeff Law <law@redhat.com>
15739 * reorg.c (relax_delay_slots): Create a new variable to hold
15740 the temporary target rather than clobbering TARGET_LABEL.
15742 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15743 missing argument to extract_bit_field call.
15744 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15746 2017-05-15 Martin Liska <mliska@suse.cz>
15749 * gcc.c (process_command): Do not allow empty argument of -o option.
15751 2017-05-15 Renlin Li <renlin.li@arm.com>
15753 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15754 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15755 * config/aarch64/constraints.md (Usf): Add long call check.
15756 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15757 (call_value): Likewise.
15758 (sibcall): Likewise.
15759 (sibcall_value): Likewise.
15761 (call_value_insn): New.
15762 (sibcall_insn): Update rtx pattern.
15763 (sibcall_value_insn): Likewise.
15764 (call_internal): Remove.
15765 (call_value_internal): Likewise.
15766 (sibcall_internal): Likewise.
15767 (sibcall_value_internal): Likewise.
15768 (call_reg): Likewise.
15769 (call_symbol): Likewise.
15770 (call_value_reg): Likewise.
15771 (call_value_symbol): Likewise.
15773 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15776 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15778 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15780 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15781 compatible with CCGOCmode and with CCZmode.
15783 2017-05-14 Martin Sebor <msebor@redhat.com>
15785 PR middle-end/77671
15786 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15787 (gimple_fold_builtin_snprintf): Same.
15788 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15789 (gimple_fold_builtin_snprintf): Same.
15790 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15791 of character types.
15792 (is_call_safe): New function.
15793 (try_substitute_return_value): Call it.
15794 (try_simplify_call): New function.
15795 (pass_sprintf_length::handle_gimple_call): Call it.
15797 2017-05-14 Martin Sebor <msebor@redhat.com>
15799 PR middle-end/80669
15800 * builtins.c (expand_builtin_stpncpy): Simplify.
15802 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15804 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15805 * config/i386/i386.h
15806 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15807 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15808 (struct machine_function): Add new members call_ms2sysv,
15809 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15810 (struct machine_frame_state): New fields sp_realigned and
15811 sp_realigned_offset.
15812 * config/i386/i386.c
15813 (enum xlogue_stub): New enum.
15814 (enum xlogue_stub_sets): New enum.
15815 (class xlogue_layout): New class.
15816 (struct ix86_frame): New fields stack_realign_allocate_offset,
15817 stack_realign_offset and outlined_save_offset. Modify comments to
15818 detail stack layout when using out-of-line stubs.
15819 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15820 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15821 -mcall-ms2sysv-xlogues.
15822 (stub_managed_regs): New static variable.
15823 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15824 registers managed by out-of-line stub.
15825 (disable_call_ms2sysv_xlogues): New function.
15826 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15827 m->call_ms2sysv when appropriate and compute frame layout for
15829 (sp_valid_at, fp_valid_at): New inline functions.
15830 (choose_basereg): New function.
15831 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15833 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15834 Use align parameter of choose_baseaddr to generated aligned SSE movs
15836 (pro_epilogue_adjust_stack): Modify to track
15837 machine_frame_state::sp_realigned.
15838 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15839 (ix86_nsaved_sseregs): Likewise.
15840 (ix86_emit_save_regs): Likewise.
15841 (ix86_emit_save_regs_using_mov): Likewise.
15842 (ix86_emit_save_sse_regs_using_mov): Likewise.
15843 (get_scratch_register_on_entry): Likewise.
15844 (gen_frame_set): New function.
15845 (gen_frame_load): Likewise.
15846 (gen_frame_store): Likewise.
15847 (emit_outlined_ms2sysv_save): Likewise.
15848 (emit_outlined_ms2sysv_restore): Likewise.
15849 (ix86_expand_prologue): Modify stack re-alignment code and call
15850 emit_outlined_ms2sysv_save when appropriate.
15851 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15852 parameter rtx_insn *insn, which allows the function to be used to only
15853 generate the notes.
15854 (ix86_expand_epilogue): Modify validity checks of frame and stack
15855 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15856 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15857 * config/i386/predicates.md
15858 (save_multiple): New predicate.
15859 (restore_multiple): Likewise.
15860 * config/i386/sse.md
15861 (save_multiple<mode>): New pattern.
15862 (save_multiple_realign<mode>): Likewise.
15863 (restore_multiple<mode>): Likewise.
15864 (restore_multiple_and_return<mode>): Likewise.
15865 (restore_multiple_leave_return<mode>): Likewise.
15866 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15868 2017-05-14 Julia Koval <julia.koval@intel.com>
15870 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15871 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15872 (__builtin_ia32_xsetbv): New builtins.
15873 * config/i386/i386.c (ix86_expand_special_args_builtin):
15875 (ix86_expand_builtin): Special expand for new intrinsics.
15876 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15877 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15878 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15880 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15882 * cfganal.c (inverted_post_order_compute): Change argument type
15884 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15885 * df-core.c (rest_of_handle_df_initialize): Adjust.
15886 (rest_of_handle_df_finish): Likewise.
15887 (df_analyze_1): Likewise.
15888 (df_analyze): Likewise.
15889 (loop_inverted_post_order_compute): Change argument to be a vec *.
15890 (df_analyze_loop): Adjust.
15891 (df_get_n_blocks): Likewise.
15892 (df_get_postorder): Likewise.
15893 * df.h (struct df_d): Change field to be a vec.
15894 * lcm.c (compute_laterin): Adjust.
15895 (compute_available): Likewise.
15896 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15897 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15898 * tree-ssa-pre.c (compute_antic): Likewise.
15900 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15902 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15903 (depth_first_search::depth_first_search): Change structure init
15904 function to this constructor.
15905 (depth_first_search::add_bb): Rename function to this member.
15906 (depth_first_search::execute): Likewise.
15907 (flow_dfs_compute_reverse_finish): Adjust.
15909 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15911 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15912 (longest_simple_path): Likewise.
15913 * shrink-wrap.c (spread_components): Likewise.
15914 (disqualify_problematic_components): Likewise.
15915 (emit_common_heads_for_components): Likewise.
15916 (emit_common_tails_for_components): Likewise.
15917 (insert_prologue_epilogue_for_components): Likewise.
15919 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15921 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15924 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15926 * df-core.c (df_set_blocks): Start using auto_bitmap.
15927 (df_compact_blocks): Likewise.
15928 * df-problems.c (df_rd_confluence_n): Likewise.
15929 * df-scan.c (df_insn_rescan_all): Likewise.
15930 (df_process_deferred_rescans): Likewise.
15931 (df_update_entry_block_defs): Likewise.
15932 (df_update_exit_block_uses): Likewise.
15933 (df_entry_block_bitmap_verify): Likewise.
15934 (df_exit_block_bitmap_verify): Likewise.
15935 (df_scan_verify): Likewise.
15936 * lra-constraints.c (lra_constraints): Likewise.
15937 (undo_optional_reloads): Likewise.
15938 (lra_undo_inheritance): Likewise.
15939 * lra-remat.c (calculate_gen_cands): Likewise.
15940 (do_remat): Likewise.
15941 * lra-spills.c (assign_spill_hard_regs): Likewise.
15942 (spill_pseudos): Likewise.
15943 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15944 (bitmap_set_subtract_values): Likewise.
15946 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15948 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15949 management with auto_bitmap.
15950 (fix_inter_tick): Likewise.
15951 (fix_recovery_deps): Likewise.
15952 * ira.c (add_store_equivs): Likewise.
15953 (find_moveable_pseudos): Likewise.
15954 (split_live_ranges_for_shrink_wrap): Likewise.
15955 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15956 (rtx_reuse_manager::seen_def_p): Likewise.
15957 (rtx_reuse_manager::set_seen_def): Likewise.
15958 * print-rtl.h (class rtx_reuse_manager): Likewise.
15960 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15962 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15964 (migrate_btr_def): Likewise.
15965 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15966 * df-core.c (loop_post_order_compute): Likewise.
15967 (loop_inverted_post_order_compute): Likewise.
15968 * hsa-common.h: Likewise.
15969 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15970 * init-regs.c (initialize_uninitialized_regs): Likewise.
15971 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15972 (inline_small_functions): Likewise.
15973 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15974 * ira.c (combine_and_move_insns): Likewise.
15975 (build_insn_chain): Likewise.
15976 * loop-invariant.c (find_invariants): Likewise.
15977 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15978 * predict.c (tree_predict_by_opcode): Likewise.
15979 (predict_paths_leading_to): Likewise.
15980 (predict_paths_leading_to_edge): Likewise.
15981 (estimate_loops_at_level): Likewise.
15982 (estimate_loops): Likewise.
15983 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15984 (spread_components): Likewise.
15985 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15986 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15987 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15988 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15989 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15990 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15991 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15992 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15993 (mark_threaded_blocks): Likewise.
15994 (thread_through_all_blocks): Likewise.
15995 * tree-ssa.c (verify_ssa): Likewise.
15996 (execute_update_addresses_taken): Likewise.
15997 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15999 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16001 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16003 (post_order_compute): Likewise.
16004 (inverted_post_order_compute): Likewise.
16005 (pre_and_rev_post_order_compute_fn): Likewise.
16007 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16009 * genrecog.c (int_set::int_set): Explicitly construct our
16010 auto_vec base class.
16011 * vec.h (auto_vec::auto_vec): New constructor.
16013 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16015 * bitmap.h (class auto_bitmap): New constructor taking
16016 bitmap_obstack * argument.
16018 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16020 * bitmap.h (class auto_bitmap): Change type of m_bits to
16021 bitmap_head, and adjust ctor / dtor and member operators.
16023 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16025 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16026 when returned register mode doesn't match original mode.
16028 2017-05-12 Jeff Law <law@redhat.com>
16029 Jakub Jelinek <jakub@redhat.com>
16031 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16032 we look for cc setter after the compare-elim changes.
16033 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16034 within the vector to match what compare-elim now expects.
16035 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16036 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16038 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16039 after the compare-elim changes.
16040 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16041 the vector to match what compare-elim now expects.
16042 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16043 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16044 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16045 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16046 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16048 * config/visium/visium.c (single_set_and_flags): Fix where
16049 we look for cc setter after the compare-elim changes.
16050 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16051 with the vector to match what compare-elim now expects.
16052 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16053 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16054 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16055 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16056 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16057 (neg<mode>2_insn_set_overflow): Likewise.
16059 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16061 PR middle-end/79794
16062 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16063 maybe_expand_insn call, set ops[0].target. If still set after call,
16064 set alt_rtl. Add extra arg to recursive calls.
16065 (extract_bit_field): Add alt_rtl argument. Pass to
16067 * expmed.h (extract_bit_field): Fix prototype.
16068 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16069 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16070 to extract_bit_field_calls.
16071 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16072 Pass alt_rtl to extract_bit_field calls.
16073 * calls.c (store_unaligned_arguments_into_psuedos)
16074 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16075 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16077 * optabs.h (struct expand_operand): Add target bitfield.
16079 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16081 * compare-elim.c (try_eliminate_compare): Canonicalize
16082 operation with embedded compare to
16083 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16084 (set (reg) (operation)].
16086 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16088 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16091 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16092 cost of adding a carry flag for ADC instruction.
16093 [case MINUS]: Ignore the cost of subtracting a carry flag
16094 for SBB instruction.
16096 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16098 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16100 * config/rs6000/bmiintrin.h: New file.
16101 * config/rs6000/bmi2intrin.h: New file.
16102 * config/rs6000/x86intrin.h: New file.
16104 2017-05-12 Jeff Law <law@redhat.com>
16106 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16109 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16111 PR middle-end/80707
16112 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16114 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16116 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16117 early expansion of vector divide builtins.
16118 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16119 builtins identified as having unsigned arguments.
16121 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16123 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16124 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16125 expansion of vector logical operations (and, andc, or, xor,
16128 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16130 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16131 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16133 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16135 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16136 early GIMPLE expansion of vector multiplies.
16138 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16140 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16141 TARGET_HAVE_MOVT conditional.
16142 (movt splitter): Likewise.
16144 2017-05-12 Richard Biener <rguenther@suse.de>
16146 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16147 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16148 Fold all stmts not inplace.
16150 2017-05-12 Richard Biener <rguenther@suse.de>
16152 PR tree-optimization/80713
16153 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16154 inserted_exprs bit for not removed stmts.
16156 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16158 PR middle-end/69921
16159 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16160 parallelized" attribute for parallelized OpenACC kernels.
16161 * omp-offload.c (execute_oacc_device_lower): Use it.
16163 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16164 Set "oacc kernels" attribute.
16165 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16166 parameter. Adjust all users.
16167 (oacc_fn_attrib_kernels_p): Remove function.
16168 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16169 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16170 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16171 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16172 assert "oacc kernels" attribute is set.
16174 2017-05-11 Carl Love <cel@us.ibm.com>
16176 * config/rs6000/rs6000-c: Add support for built-in functions
16177 vector unsigned char vec_popcnt (vector signed char)
16178 vector unsigned char vec_popcnt (vector unsigned char)
16179 vector unsigned short vec_popcnt (vector signed short)
16180 vector unsigned short vec_popcnt (vector unsigned short)
16181 vector unsigned int vec_popcnt (vector signed int)
16182 vector unsigned int vec_popcnt (vector unsigned int)
16183 vector unsigned long long vec_popcnt (vector signed long long)
16184 vector unsigned long long vec_popcnt (vector unsigned long long)
16185 vector signed long long vec_slo (vector signed long long,
16186 vector signed char)
16187 vector signed long long vec_slo (vector signed long long,
16188 vector unsigned char)
16189 vector unsigned long long vec_slo (vector unsigned long long,
16190 vector signed char)
16191 vector unsigned long long vec_slo (vector unsigned long long,
16192 vector unsigned char)
16193 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16194 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16195 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16196 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16197 * doc/extend.texi: Update the built-in documentation file for the
16198 new built-in functions.
16200 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16202 * attribs.h (sorted_attr_string): Move machine independent
16203 functions for target clone support from the i386 port to common
16204 code. Rename ix86_function_versions to common_function_versions.
16205 Rename make_name to make_unique_name.
16206 (common_function_versions): Likewise.
16207 (make_unique_name): Likewise.
16208 (make_dispatcher_decl): Likewise.
16209 (is_function_default_version): Likewise.
16210 * attribs.c (attr_strcmp): Likewise.
16211 (sorted_attr_string): Likewise.
16212 (common_function_versions): Likewise.
16213 (make_unique_name): Likewise.
16214 (make_dispatcher_decl): Likewise.
16215 (is_function_default_version): Likewise.
16216 * config/i386/i386.c (attr_strcmp): Likewise.
16217 (sorted_attr_string): Likewise.
16218 (ix86_function_versions): Likewise.
16219 (make_name): Likewise.
16220 (make_dispatcher_decl): Likewise.
16221 (is_function_default_version): Likewise.
16222 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16224 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16227 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16228 Account for direct move costs for vec_construct of integer
16231 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16234 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16235 (UNSPEC_STX_ATOMIC): Ditto.
16236 (loaddi_via_sse): New insn.
16237 (storedi_via_sse): Ditto.
16238 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16239 Update corresponding peephole2 patterns.
16240 (atomic_storedi_fpu): Ditto.
16242 2017-05-11 Julia Koval <julia.koval@intel.com>
16244 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16245 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16247 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16248 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16249 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16251 2017-05-11 Nathan Sidwell <nathan@acm.org>
16253 * graphite-poly.c: Include dumpfile.h.
16255 * dumpfle.h (dump_function): Declare here ...
16256 * tree-dump.h (dump_function): ... not here.
16257 * dumpfile.c: #include tree-cfg.h.
16258 (dump_function): Move here from ...
16259 * tree-dump.c (dump_function): ... here.
16260 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16261 * graphite-poly.c: Don't include tree-dump.h.
16262 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16263 * print-tree.c: Likewise.
16264 * stor-layout.c: Likewise.
16265 * tree-nested.c: Likewise.
16267 * dumpfile.c (dump_start): Use TDF_FLAGS.
16268 (dump_enable_all): Fix TDF_KIND check thinko.
16270 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16272 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16273 array entries to represent two legal parameterizations of the
16274 overloaded __builtin_cmpb function, as represented by the
16275 P6_OV_BUILTIN_CMPB constant.
16276 (altivec_resolve_overloaded_builtin): Add special case handling
16277 for the __builtin_cmpb function, as represented by the
16278 P6_OV_BUILTIN_CMPB constant.
16279 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16280 (BU_P6_64BIT_2): New macro.
16281 (BU_P6_OVERLOAD_2): New macro
16282 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16283 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16284 (CMPB): Add overload support to represent both 32-bit and 64-bit
16285 compare-bytes function.
16286 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16287 support for TARGET_CMPB.
16288 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16289 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16290 documentation of the __builtin_cmpb overloaded built-in function.
16292 2017-05-11 Richard Biener <rguenther@suse.de>
16294 PR tree-optimization/80705
16295 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16296 bases are not vectorizable.
16298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16300 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16301 when counting register pressure.
16303 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16305 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16306 (iv_ca_more_deps): Renamed to ...
16307 (iv_ca_compare_deps): ... this.
16308 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16310 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16312 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16314 (determine_group_iv_costs): ... here.
16315 (find_inv_vars_cb): Record inv var if it's not recorded before.
16317 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16319 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16320 (get_shiftadd_cost): Ditto.
16322 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16324 * tree-ssa-address.c: Include header file.
16325 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16327 (add_to_parts): Refactor.
16328 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16329 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16332 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16334 PR tree-optimization/53090
16335 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16337 (extract_cond_operands): Detect condition with IV on both sides
16338 and return COMP_IV_EXPR_2.
16339 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16340 (rewrite_use_compare): Simplify by removing call to function
16341 extract_cond_operands.
16343 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16345 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16346 (extract_cond_operands): Detect condition comparing against non-
16347 invariant bound and return appropriate enum value.
16348 (find_interesting_uses_cond): Update use of extract_cond_operands.
16349 Handle its return value accordingly.
16350 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16352 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16354 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16355 nonlinear iv_use computation in loop invariant sensitive way.
16357 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16359 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16360 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16362 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16364 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16365 (dump_cand): Support iv_cand.inv_exprs.
16366 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16368 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16371 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16373 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16375 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16376 as local function. Include necessary header files.
16377 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16381 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16383 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16385 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16386 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16387 RSHIFT_EXPR and BIT_NOT_EXPR.
16389 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16391 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16392 (adjust_setup_cost): New parameter supporting round up adjustment.
16393 (struct address_cost_data): Delete.
16394 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16395 (split_address_cost, ptr_difference_cost): Delete.
16396 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16397 (struct ainc_cost_data): New struct.
16398 (get_address_cost_ainc): New function.
16399 (get_address_cost, get_computation_cost): Reimplement.
16400 (determine_group_iv_cost_address): Record inv_expr for all uses of
16402 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16403 (iv_ca_has_deps): Reimplemented to ...
16404 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16406 (iv_ca_extend): Call iv_ca_more_deps.
16408 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16410 * tree-ssa-address.c (struct mem_address): Move to header file.
16411 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16412 * tree-ssa-address.h (struct mem_address): Move from C file.
16413 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16415 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16417 * tree-affine.h (aff_combination_type): New interface.
16418 (aff_combination_zero_p): Remove static.
16419 (aff_combination_const_p): New interface.
16420 (aff_combination_singleton_var_p): New interfaces.
16422 2017-05-11 Richard Biener <rguenther@suse.de>
16424 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16425 Skip unreachable blocks and destinations.
16426 (eliminate): Move stmt removal and fixup ...
16427 (fini_eliminate): ... here. Skip inserted exprs.
16428 (pass_pre::execute): Move fini_pre after fini_eliminate.
16429 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16430 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16431 PRE to get rid of dead code that has invalid SSA form and
16432 split critical edges again.
16434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16436 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16438 2017-05-11 Richard Biener <rguenther@suse.de>
16440 * passes.c (execute_function_todo): Verify loops if they are
16441 said to be up-to-date.
16442 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16443 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16445 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16448 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16449 handle calling assemble_external ourself.
16452 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16453 modes with zero size. Enhance comment.
16455 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16457 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16458 built-ins for vec_xl and vec_xst with short and char pointer
16461 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16463 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16464 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16465 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16466 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16467 (_mm_maskz_min_round_ss): New intrinsics.
16468 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16469 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16470 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16471 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16472 (__builtin_ia32_minss_mask_round): New builtins.
16473 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16474 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16475 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16477 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16478 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16480 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16483 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16485 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16486 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16487 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16488 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16489 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16490 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16491 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16492 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16493 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16494 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16495 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16496 (__builtin_ia32_mulss_mask_round): New builtins.
16497 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16498 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16499 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16501 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16502 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16504 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16507 2017-05-10 Julia Koval <julia.koval@intel.com>
16509 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16510 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16511 (_mm256_setr_m128i): New intrinsics.
16513 2017-05-10 Julia Koval <julia.koval@intel.com>
16515 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16516 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16517 (_mm_maskz_rcp14_ss): New intrinsics.
16518 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16519 (__builtin_ia32_rcp14ss_mask): New builtins.
16520 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16522 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16524 PR tree-optimization/51513
16525 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16526 (assert_unreachable_fallthru_edge_p): Use it.
16527 (group_case_labels_stmt): Likewise.
16528 * tree-cfg.h: Prototype it.
16529 * stmt.c: Include cfghooks.h and tree-cfg.h.
16530 (emit_case_dispatch_table) <gap_label>: New local variable.
16531 Use it to fill dispatch table gaps.
16532 Test for default_label before updating probabilities.
16533 (expand_case) <default_label>: Remove unneeded initialization.
16534 Test for unreachable default case statement and remove its edge.
16535 Set default_label accordingly.
16536 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16538 2017-05-10 Carl Love <cel@us.ibm.com>
16540 * config/rs6000/rs6000-c: Add support for built-in functions
16541 vector signed char vec_neg (vector signed char)
16542 vector signed short int vec_neg (vector short int)
16543 vector signed int vec_neg (vector signed int)
16544 vector signed long long vec_neg (vector signed long long)
16545 vector float vec_neg (vector float)
16546 vector double vec_neg (vector double)
16547 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16549 * config/rs6000/altivec.h: Add define for vec_neg
16550 * doc/extend.texi: Update the built-in documentation for the
16551 new built-in functions.
16553 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16555 PR tree-optimization/77644
16556 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16558 2017-05-10 Nathan Sidwell <nathan@acm.org>
16560 * dumpfile.h (TDI_lang_all): New.
16561 (TDF_KIND): New. Renumber others
16562 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16564 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16566 (get_dump_file_name): Adjust suffix generation.
16567 (dump_enable_all): Use TDF_KIND.
16568 * doc/invoke.texi (-fdump-lang-all): Document.
16570 * dumpfile.h: Tabify.
16572 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16575 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16576 Move member access before delete.
16578 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16580 * tree-inline.c (expand_call_inline): Split block at stmt
16583 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16586 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16587 are now unused after splitting mov{sf,sd}_hardfloat.
16588 (f32_lr2): Likewise.
16589 (f32_lm): Likewise.
16590 (f32_lm2): Likewise.
16591 (f32_li): Likewise.
16592 (f32_li2): Likewise.
16593 (f32_lv): Likewise.
16594 (f32_sr): Likewise.
16595 (f32_sr2): Likewise.
16596 (f32_sm): Likewise.
16597 (f32_sm2): Likewise.
16598 (f32_si): Likewise.
16599 (f32_si2): Likewise.
16600 (f32_sv): Likewise.
16601 (f32_dm): Likewise.
16602 (f32_vsx): Likewise.
16603 (f32_av): Likewise.
16604 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16605 For movsf, order stores so the VSX stores occur before the GPR
16606 store which encourages the register allocator to use a traditional
16607 FPR instead of a GPR. For movsd, order the stores so that the GPR
16608 store comes before the VSX stores to allow the power6 to work.
16609 This is due to the power6 not having a 32-bit integer store
16610 instruction from a FPR.
16611 (movsf_hardfloat): Likewise.
16612 (movsd_hardfloat): Likewise.
16614 2017-05-09 Martin Sebor <msebor@redhat.com>
16616 PR translation/80280
16617 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16620 PR translation/80280
16621 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16622 data member added in r247778.
16623 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16625 2017-05-09 Nathan Sidwell <nathan@acm.org>
16627 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16629 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16632 2017-05-09 Marek Polacek <polacek@redhat.com>
16634 * doc/invoke.texi: Fix typo.
16636 2017-05-09 Richard Biener <rguenther@suse.de>
16638 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16639 (vrp_val_is_min): Likewise.
16640 (set_value_range_to_value): Likewise.
16641 (set_value_range_to_nonnegative): Likewise.
16642 (gimple_assign_nonzero_p): Likewise.
16643 (gimple_stmt_nonzero_p): Likewise.
16644 (vrp_int_const_binop): Likewise. Remove unreachable case.
16645 (adjust_range_with_scev): Adjust comments.
16646 (compare_range_with_value): Likewise.
16647 (extract_range_from_phi_node): Likewise.
16648 (test_for_singularity): Likewise.
16650 2017-05-09 Richard Biener <rguenther@suse.de>
16652 * tree-vrp.c (get_single_symbol): Add assert that we don't
16653 get overflowed constants as invariant part.
16654 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16655 checks. Use wi::cmp instead of recursing for integer constants.
16656 (compare_values): Just ignore whether we assumed undefined
16657 overflow instead of failing the compare.
16658 (extract_range_for_var_from_comparison_expr): Add comment before the
16659 TREE_NO_WARNING sets.
16660 (test_for_singularity): Likewise.
16661 (extract_range_from_comparison): Do not disable optimization
16662 when we assumed undefined overflow.
16663 (extract_range_basic): Remove init of unused var.
16665 2017-05-09 Richard Biener <rguenther@suse.de>
16667 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16668 (extract_range_from_multiplicative_op_1): Adjust.
16669 (extract_range_from_binary_expr_1): Use int_const_binop.
16671 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16674 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16675 rs6000_store_data_bypass_p in seven define_bypass directives and
16676 in several comments.
16677 * config/rs6000/rs6000-protos.h: Add prototype for
16678 rs6000_store_data_bypass_p function.
16679 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16680 function implements slightly different (rs6000-specific) semantics
16681 than store_data_bypass_p, returning false rather than aborting
16682 with assertion error when arguments do not satisfy the
16683 requirements of store data bypass.
16684 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16685 rs6000_store_data_bypass_p.
16687 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16689 * config/xtensa/xtensa-protos.h
16690 (xtensa_initial_elimination_offset): New declaration.
16691 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16692 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16693 macro definition, add case for FRAME_POINTER_REGNUM when
16694 FRAME_GROWS_DOWNWARD.
16695 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16696 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16697 xtensa_initial_elimination_offset.
16699 2017-05-08 Nathan Sidwell <nathan@acm.org>
16701 * doc/invoke.texi: Alphabetize -fdump options.
16703 2017-05-08 Martin Sebor <msebor@redhat.com>
16705 PR translation/80280
16706 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16708 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16710 * target.def (compute_frame_layout): New optional target hook.
16711 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16712 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16713 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16715 * reload1.c (verify_initial_elim_offsets): Likewise.
16716 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16717 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16718 (arm_get_frame_offsets): Split up into this ...
16719 (arm_compute_frame_layout): ... and this function.
16721 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16723 * config/aarch64/constraints.md (Usa): New constraint.
16724 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16726 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16728 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16729 with_multilib_list after it has been checked.
16731 2017-05-08 Richard Biener <rguenther@suse.de>
16733 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16734 (bitmap_set_subtract_values): Likewise.
16736 2017-05-08 Richard Biener <rguenther@suse.de>
16738 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16739 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16741 (gimple_stmt_nonzero_warnv_p): Rename to ...
16742 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16744 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16745 (extract_range_basic): Adjust, do not disable propagation on
16746 strict overflow sensitive simplification.
16747 (vrp_visit_cond_stmt): Likewise.
16749 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16751 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16752 body size unconditionally.
16754 2017-05-07 Jeff Law <law@redhat.com>
16757 2017-05-06 Jeff Law <law@redhat.com>
16758 PR tree-optimization/78496
16759 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16762 PR tree-optimization/78496
16763 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16764 (simplify_stmt_using_ranges): Call it.
16765 (vrp_dom_walker::before_dom_children): Extract equivalences
16766 from an ASSERT_EXPR with an equality comparison against a
16769 2017-05-06 Jeff Law <law@redhat.com>
16771 PR tree-optimization/78496
16772 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16775 PR tree-optimization/78496
16776 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16777 (simplify_stmt_using_ranges): Call it.
16778 (vrp_dom_walker::before_dom_children): Extract equivalences
16779 from an ASSERT_EXPR with an equality comparison against a
16782 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16784 * lra-constraints.c (lra_copy_reg_equiv): New function.
16785 (split_reg): Use it to copy equivalence information from the
16786 original register to the spill register.
16788 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16790 PR rtl-optimization/75964
16791 * simplify-rtx.c (simplify_const_relational_operation): Remove
16792 invalid handling of comparisons of integer ABS.
16794 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16796 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16797 initialize to zero.
16798 (init_regs): Remove declaration.
16799 (function_arg_advance_32): Initialize error_p as boolean variable.
16801 2017-05-05 Nathan Sidwell <nathan@acm.org>
16803 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16804 lines. Use for (;;).
16806 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16808 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16809 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16810 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16811 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16812 VF=2 that require versioning.
16814 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16816 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16819 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16821 * diagnostic.h (diagnostic_override_option_index): Convert from
16822 macro to inline function.
16824 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16826 * diagnostic.c (last_module_changed_p): New function.
16827 (set_last_module): New function.
16828 (diagnostic_report_current_module): Convert macro usage to
16829 the above functions.
16830 * diagnostic.h (diagnostic_context::last_module): Strengthen
16831 from const line_map * to const line_map_ordinary *.
16832 (diagnostic_last_module_changed): Delete macro.
16833 (diagnostic_set_last_module): Delete macro.
16835 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16837 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16838 with diagnostic_report_diagnostic.
16839 (diagnostic_n_impl_richloc): Likewise.
16840 * diagnostic.h (report_diagnostic): Delete macro.
16841 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16842 with diagnostic_report_diagnostic.
16843 * substring-locations.c (format_warning_va): Likewise.
16845 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16847 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16848 save/restor of format_spec. Move option-printing code to...
16849 (print_option_information): ...this new function, and
16850 reimplement by simply printing to the pretty_printer,
16851 rather than appending to the format string.
16853 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16855 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16856 handling logic into...
16857 (update_effective_level_from_pragmas): ...this new function.
16859 2017-05-04 Andrew Waterman <andrew@sifive.com>
16861 * config/riscv/riscv.opt (mstrict-align): New option.
16862 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16863 (SLOW_UNALIGNED_ACCESS): Define.
16864 (riscv_slow_unaligned_access): Declare.
16865 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16867 (riscv_slow_unaligned_access): New variable.
16868 (rocket_tune_info): Set slow_unaligned_access to true.
16869 (optimize_size_tune_info): Set slow_unaligned_access to false.
16870 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16871 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16872 (riscv_option_override): Set riscv_slow_unaligned_access.
16873 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16875 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16877 * config/riscv/riscv.md: Unify indentation.
16879 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16884 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16886 (extendsi<mode>2): Add support for doing sign extension via
16887 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16888 don't have ISA 3.0 instructions.
16889 (extendsi<mode>2 splitter): Likewise.
16890 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16891 generate the normal insns since SImode can now go in vector
16892 registers. Disallow the special UNSPECs needed for previous
16893 machines to hide SImode being used. Add new insns
16894 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16895 (fix_trunc<mode>si2_stfiwx): Likewise.
16896 (fix_trunc<mode>si2_internal): Likewise.
16897 (fixuns_trunc<mode>si2): Likewise.
16898 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16899 (fctiw<u>z_<mode>_smallint): Likewise.
16900 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16901 of floating point to 32-bit integer from doing a direct move to
16902 the GPR registers to do a store.
16903 (fctiwz_<mode>): Break long line.
16905 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16907 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16908 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16909 (addr_list, addr_offset_valid_p): New.
16910 (split_address_groups): Check offset validity with above function.
16911 (gt-tree-ssa-loop-ivopts.h): Include header file.
16913 2017-05-05 Nathan Sidwell <nathan@acm.org>
16915 * config.gcc (arm*-*-*): Add missing 'fi'.
16917 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16919 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16920 included in -fopt-info does not matter.
16921 * doc/optinfo.texi (-fopt-info): Fix description of default
16922 behavour. Explicitly say order of options included in -fopt-info
16925 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16927 * config.gcc: Allow combinations of aprofile and rmprofile values for
16928 --with-multilib-list.
16929 * config/arm/t-multilib: New file.
16930 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16931 variables. Remove setting of ISA and floating-point ABI in
16932 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16933 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16934 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16935 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16937 * config/arm/t-rmprofile: Likewise except for the matches changes.
16938 * doc/install.texi (--with-multilib-list): Document the combination of
16939 aprofile and rmprofile values and warn about pitfalls in doing that.
16941 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16943 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16944 (movdi_aarch64): Likewise.
16946 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16948 PR tree-optimization/80632
16949 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16951 (build_arrays): Initialize it for virtual phis.
16952 (fix_phi_nodes): Use it for virtual phis.
16954 PR tree-optimization/80558
16955 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16956 [x, y] op z into [x op, y op z] for op & or | if conditions
16959 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16960 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16963 * config/arm/arm.md (use_literal_pool): Remove.
16964 (64-bit immediate split): No longer takes cost into consideration
16965 if arm_disable_literal_pool is enabled.
16966 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16967 used when arm_disable_literal_pool is enabled.
16968 (arm_max_const_double_inline_cost): Remove use of
16969 arm_disable_literal_pool.
16970 (push_minipool_fix): Add assert.
16971 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16972 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16973 (no_literal_pool_sf_immediate): New.
16975 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16977 PR tree-optimization/80613
16978 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16979 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16981 2017-05-05 Richard Biener <rguenther@suse.de>
16983 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16985 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16987 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16988 of this flag from insn conditions due to removal from r247495.
16990 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16992 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16994 (arm_early_store_addr_dep_ptr): Likewise.
16995 * config/arm/aarch-common-protos.h
16996 (arm_early_load_addr_dep_ptr): Add prototype.
16997 (arm_early_store_addr_dep_ptr): Likewise.
16998 * config/arm/cortex-a53.md: Add new bypasses.
17000 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17002 * tree.c (next_type_uid): Change type to unsigned.
17003 (type_hash_canon): Decrement back next_type_uid if
17004 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17005 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17008 2017-05-04 Martin Sebor <msebor@redhat.com>
17010 * builtins.c: Fix a trivial typo in a comment.
17012 PR middle-end/79234
17013 * builtins.c (check_sizes): Adjust to handle reading past the end.
17014 Avoid printing excessive upper bound of ranges. Use %E to print
17015 tree nodes instead of converting them to %wu.
17016 (expand_builtin_memchr): New function.
17017 (compute_dest_size): Rename...
17018 (compute_objsize): ...to this.
17019 (expand_builtin_memcpy): Adjust.
17020 (expand_builtin_mempcpy): Adjust.
17021 (expand_builtin_strcat): Adjust.
17022 (expand_builtin_strcpy): Adjust.
17023 (check_strncat_sizes): Adjust.
17024 (expand_builtin_strncat): Adjust.
17025 (expand_builtin_strncpy): Adjust and simplify.
17026 (expand_builtin_memset): Adjust.
17027 (expand_builtin_bzero): Adjust.
17028 (expand_builtin_memcmp): Adjust.
17029 (expand_builtin): Handle memcmp.
17030 (maybe_emit_chk_warning): Check strncat just once.
17032 2017-05-04 Martin Sebor <msebor@redhat.com>
17034 PR preprocessor/79214
17035 PR middle-end/79222
17036 PR middle-end/79223
17037 * builtins.c (check_sizes): Add inlining context and issue
17038 warnings even when -Wno-system-headers is set.
17039 (check_strncat_sizes): Same.
17040 (expand_builtin_strncat): Same.
17041 (expand_builtin_memmove): New function.
17042 (expand_builtin_stpncpy): Same.
17043 (expand_builtin): Handle memmove and stpncpy.
17045 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17047 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17048 which is not used any more.
17050 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17052 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17054 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17056 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17057 (cortexa53_tunings): Likewise.
17058 (cortexa57_tunings): Likewise.
17059 (cortexa72_tunings): Likewise.
17060 (cortexa73_tunings): Likewise.
17062 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17064 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17065 Set loop alignment to 8.
17067 2017-05-04 Martin Sebor <msebor@redhat.com>
17069 PR translation/80280
17070 * builtins.c (expand_builtin_object_size): Add missing quoting to
17071 %D and like directives.
17072 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17073 (hsa_type_for_tree_type): Same.
17074 (verify_function_arguments): Same.
17075 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17076 * varasm.c (get_section): Same.
17079 2017-05-04 Martin Sebor <msebor@redhat.com>
17081 PR translation/80280
17082 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17084 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17086 * config/aarch64/aarch64.c (generic_addrcost_table):
17087 Change HI/TI mode setting.
17089 2017-05-04 Martin Jambor <mjambor@suse.cz>
17091 PR tree-optimization/80622
17092 * tree-sra.c (comes_initialized_p): New function.
17093 (build_accesses_from_assign): Only set write lazily when
17094 comes_initialized_p is false.
17095 (analyze_access_subtree): Use comes_initialized_p.
17096 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17097 instead of testing for PARM_DECL.
17099 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17101 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17102 constraint on operand 0 to allow more general addressing modes.
17103 Adjust output template.
17104 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17106 * config/aarch64/aarch64-protos.h
17107 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17108 * config/aarch64/constraints.md (Dp): New address constraint.
17109 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17112 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17114 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17115 update use of estimate_ipcp_clone_size_and_time.
17116 (estimate_local_effects): Update use of
17117 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17118 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17119 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17120 Return nonspecialized time.
17122 2017-05-04 Richard Biener <rguenther@suse.de>
17124 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17125 for the last VUSE which def dominates the PHI. Directly call
17127 (get_continuation_for_phi_1): Remove.
17129 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17131 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17132 to explain the use of truncating division. Cap the number of
17133 iterations to the maximum given by nb_iterations_upper_bound,
17136 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17138 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17139 * configure: Regenerate.
17140 * config.in: Regenerate.
17141 * config/i386/driver-mingw32.c: new file.
17142 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17143 * config.host: Link driver-mingw32.o on MinGW host.
17144 * doc/install.texi: Document new --enable-mingw-wildcard configure
17147 2017-05-04 Marek Polacek <polacek@redhat.com>
17149 PR tree-optimization/80612
17150 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17152 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17153 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17155 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17156 (movt splitter): Likewise.
17157 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17158 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17159 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17160 block for Thumb-1 with MOVT.
17161 (thumb2_legitimate_address_p): Move code block ...
17162 (can_avoid_literal_pool_for_label_p): ... into this new function.
17163 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17165 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17166 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17167 "M-profile targets with the MOVT instruction".
17169 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17171 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17172 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17173 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17175 2017-05-04 Martin Liska <mliska@suse.cz>
17177 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17178 variable cond_code.
17180 2017-05-04 Richard Biener <rguenther@suse.de>
17182 * tree.c (array_at_struct_end_p): Handle arrays at struct
17183 end with flexarrays more conservatively. Refactor and treat
17184 arrays of arrays or aggregates more strict. Fix
17185 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17186 * tree.c (array_at_struct_end_p): Adjust prototype.
17187 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17188 * gimple-fold.c (get_range_strlen): Likewise.
17189 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17191 2017-05-04 Richard Biener <rguenther@suse.de>
17193 PR tree-optimization/31130
17194 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17196 (supports_overflow_infinity): Likewise.
17197 (is_negative_overflow_infinity): Likewise.
17198 (is_positive_overflow_infinity): Likewise.
17199 (is_overflow_infinity): Likewise.
17200 (stmt_overflow_infinity): Likewise.
17201 (overflow_infinity_range_p): Likewise.
17202 (usable_range_p): Remove as always returning true.
17203 (make_overflow_infinity): Remove.
17204 (negative_overflow_infinity): Likewise.
17205 (positive_overflow_infinity): Likewise.
17206 (avoid_overflow_infinity): Likewise.
17207 (set_value_range): Adjust accordingly.
17208 (set_value_range_to_nonnegative): Likewise, remove now unused
17209 overflow_infinity arg.
17210 (vrp_operand_equal_p): Adjust.
17211 (update_value_range): Likewise.
17212 (range_int_cst_singleton_p): Likewise.
17213 (operand_less_p): Likewise.
17214 (compare_values_warnv): Likewise.
17215 (extract_range_for_var_from_comparison_expr): Likewise.
17216 (vrp_int_const_binop): Likewise.
17217 (zero_nonzero_bits_from_vr): Likewise.
17218 (extract_range_from_multiplicative_op_1): Likewise.
17219 (extract_range_from_binary_expr_1): Likewise.
17220 (extract_range_from_unary_expr): Likewise.
17221 (extract_range_from_comparison): Likewise.
17222 (extract_range_basic): Likewise.
17223 (adjust_range_with_scev): Likewise.
17224 (compare_ranges): Likewise.
17225 (compare_range_with_value): Likewise.
17226 (dump_value_range): Likewise.
17227 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17229 (simplify_cond_using_ranges): Adjust.
17231 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17233 * brig-builtins.def: Added a builtin for class_f64.
17234 * builtin-types.def: Added a builtin type needed by class_f64.
17236 2017-05-03 Jason Merrill <jason@redhat.com>
17238 * timevar.def: Add TV_CONSTEXPR.
17240 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17242 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17244 2017-05-03 Martin Jambor <mjambor@suse.cz>
17246 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17247 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17248 * ipa-cp.c (ipcp_propagate_stage): Do not call
17249 ipa_update_after_lto_read.
17250 * ipa-inline.c (ipa_inline): Likewise.
17252 2017-05-03 Martin Jambor <mjambor@suse.cz>
17254 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17255 tag. Added a default constructor and a destructor.
17256 (ipa_edge_args_sum_t): New class;
17257 (ipa_edge_args_sum): Declare.
17258 (ipa_edge_args_vector): Remove declaration.
17259 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17260 (ipa_free_edge_args_substructures): Remove declaration.
17261 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17262 (ipa_edge_args_info_available_for_edge_p): Likewise.
17263 * ipa-prop.c (ipa_edge_args_vector): Removed.
17264 (edge_removal_hook_holder): Likewise.
17265 (edge_duplication_hook_holder): Likewise.
17266 (ipa_edge_args_sum): New variable.
17267 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17268 ipa_edge_args_vector.
17269 (ipa_free_edge_args_substructures): Likewise.
17270 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17271 ipa_edge_args_vector.
17272 (ipa_edge_removal_hook): Turned into method
17273 ipa_edge_args_sum_t::remove.
17274 (ipa_edge_duplication_hook): Turned into method
17275 ipa_edge_args_sum_t::duplicate.
17276 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17277 registering edge hooks.
17278 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17279 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17280 ipa_edge_args_sum instead of ipa_edge_args_vector.
17281 * ipa-profile.c (ipa_profile): Likewise.
17283 2017-05-03 Martin Jambor <mjambor@suse.cz>
17285 * symbol-summary.h (function_summary): New method exists.
17286 (function_summary::symtab_removal): Deallocate through release.
17287 (call_summary): New class.
17288 (gt_ggc_mx): New overload.
17289 (gt_pch_nx): Likewise.
17290 (gt_pch_nx): Likewise.
17292 2017-05-03 Jeff Law <law@redhat.com>
17294 PR tree-optimization/78496
17295 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17296 from simplify_cond_using_ranges. Split off code to walk
17297 backwards through casts into ...
17298 (simplify_cond_using_ranges_2): New function.
17299 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17300 (execute_vrp): After identifying jump threads, call
17301 simplify_cond_using_ranges_2.
17303 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17306 * ipa-inline.h (inline_summary): Add ctor.
17307 (create_ggc): Do not use ggc_cleared_alloc.
17309 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17310 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17312 * gcc.c (handle_braces): Support escaping in switch matching
17314 * doc/invoke.texi (Spec Files): Document it.
17315 Remove superfluous @code markup in items.
17317 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17319 * diagnostic-show-locus.c (struct column_range): New struct.
17320 (get_affected_columns): New function.
17321 (get_printed_columns): New function.
17322 (struct correction): New struct.
17323 (correction::ensure_capacity): New function.
17324 (correction::ensure_terminated): New function.
17325 (struct line_corrections): New struct.
17326 (line_corrections::~line_corrections): New dtor.
17327 (line_corrections::add_hint): New function.
17328 (layout::print_trailing_fixits): Reimplement in terms of the new
17330 (selftest::test_overlapped_fixit_printing): New function.
17331 (selftest::diagnostic_show_locus_c_tests): Call it.
17333 2017-05-03 Nathan Sidwell <nathan@acm.org>
17335 Canonicalize canonical type hashing
17336 * tree.h (type_hash_canon_hash): Declare.
17337 * tree.c (type_hash_list, attribute_hash_list): Move into
17338 type_hash_canon_hash.
17339 (build_type_attribute_qual_variant): Break out hash code calc into
17340 type_hash_canon_hash.
17341 (type_hash_canon_hash): New. Generic type hash computation.
17342 (build_range_type_1, build_array_type_1, build_function_type,
17343 build_method_type_directly, build_offset_type, build_complex_type,
17344 make_vector_type): Call it.
17346 2017-05-03 Richard Biener <rguenther@suse.de>
17348 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17349 When all DRs have unknown misaligned do not always peel
17350 when there is a store but apply the same costing model as if
17351 there were only loads.
17353 2017-05-03 Richard Biener <rguenther@suse.de>
17356 PR tree-optimization/80492
17357 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17358 compare_base_decls returning dont-know properly.
17360 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17362 * config/arm/iterators.md (CCSI): New mode iterator.
17363 (arch): New mode attribute.
17364 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17365 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17366 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17367 code iterator for success result mode.
17368 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17369 the corresponding new insn generators.
17371 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17374 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17375 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17377 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17379 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17380 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17381 (DDR_A): Wrap DDR argument in brackets.
17382 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17383 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17384 (DDR_REVERSED_P): Likewise.
17386 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17388 PR tree-optimization/79472
17389 * tree-switch-conversion.c (struct switch_conv_info): Add
17390 contiguous_range and default_case_nonstandard fields.
17391 (collect_switch_conv_info): Compute contiguous_range and
17392 default_case_nonstandard fields, don't clear final_bb if
17393 contiguous_range and only the default case doesn't have the required
17395 (check_all_empty_except_final): Set default_case_nonstandard instead
17396 of failing if contiguous_range and the default case doesn't have empty
17398 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17399 and only the default case doesn't have the required constants. Skip
17401 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17402 if default_case_nonstandard.
17403 (build_constructors): Build constant 1 just once. Assert that default
17404 values aren't inserted in between cases if contiguous_range. Skip
17406 (build_arrays): Skip virtual phis.
17407 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17408 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17409 Handle virtual phis.
17410 (gen_inbound_check): Handle default_case_nonstandard case.
17411 (process_switch): Adjust check_final_bb caller. Call
17412 gather_default_values with the first non-default case instead of
17413 default case if default_case_nonstandard.
17415 2017-05-02 Nathan Sidwell <nathan@acm.org>
17417 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17418 check. Fix formatting.
17420 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17422 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17423 errors when comparing specialized and unspecialized times.
17425 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17427 * diagnostic-show-locus.c
17428 (layout::should_print_annotation_line_p): Make private.
17429 (layout::print_annotation_line): Make private.
17430 (layout::annotation_line_showed_range_p): Make private.
17431 (layout::show_ruler): Make private.
17432 (layout::print_source_line): Make private. Pass in line and
17433 line_width, rather than calling location_get_source_line. Drop
17435 (layout::print_leading_fixits): New method.
17436 (layout::print_any_fixits): Rename to...
17437 (layout::print_trailing_fixits): ...this, and make private.
17438 Don't print newline fixits.
17439 (diagnostic_show_locus): Move logic for printing one row into...
17440 (layout::print_line): ...this new function. Move the
17441 location_get_source_line call and error-handling from
17442 print_source_line to here. Call print_leading_fixits, and rename
17443 print_any_fixits to print_trailing_fixits.
17444 (selftest::test_fixit_insert_containing_newline): Update now that
17445 newlines are partially supported.
17446 (selftest::test_fixit_insert_containing_newline_2): New test.
17447 (selftest::test_fixit_replace_containing_newline): Update comments.
17448 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17449 * edit-context.c (class added_line): New class.
17450 (class edited_line): Describe newline handling in comment.
17451 (edited_line::actually_edited_p): New method.
17452 (edited_line::print_content): Delete redundant decl.
17453 (edited_line::m_predecessors): New field.
17454 (edited_file::print_content): Call edited_line::print_content.
17455 (edited_file::print_diff): Update to support newlines.
17456 (edited_file::print_diff_hunk): Likewise.
17457 (edited_file::print_run_of_changed_lines): New function.
17458 (edited_file::print_diff_line): Convert to...
17459 (print_diff_line): ...this.
17460 (edited_file::get_effective_line_count): New function.
17461 (edited_line::edited_line): Initialize new field m_predecessors.
17462 (edited_line::~edited_line): Clean up m_predecessors.
17463 (edited_line::apply_fixit): Handle newlines.
17464 (edited_line::get_effective_line_count): New function.
17465 (edited_line::print_content): New function.
17466 (edited_line::print_diff_lines): New function.
17467 (selftest::test_applying_fixits_insert_containing_newline): New
17469 (selftest::test_applying_fixits_replace_containing_newline): New
17471 (selftest::insert_line): New function.
17472 (selftest::test_applying_fixits_multiple_lines): Add example of
17474 (selftest::edit_context_c_tests): Call the new tests.
17476 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17478 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17479 parameter cand. Update dump information.
17480 (get_computation_cost): Update uses.
17482 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17484 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17485 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17486 (get_computation_at, rewrite_use_address): Update use of
17487 get_computation_aff.
17489 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17491 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17492 (get_computation): Delete.
17493 (get_computation_cost): Implement like get_computation_cost_at.
17494 Use get_computation_at.
17495 (get_computation_cost_at): Delete.
17496 (rewrite_use_nonlinear_expr): Use get_computation_at.
17497 (rewrite_use_compare, remove_unused_ivs): Ditto.
17499 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17501 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17503 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17505 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17506 (ivopts_global_cost_for_size): Rename parameter and update uses.
17507 (iv_ca_recount_cost): Update uses.
17508 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17509 candidates seperately in n_invs and n_cands.
17510 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17512 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17514 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17515 (find_inv_vars_cb): New.
17516 (find_depends): Renamed to ...
17517 (find_inv_vars): ... this.
17518 (add_candidate_1, force_var_cost): Call find_inv_vars.
17519 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17521 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17523 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17524 inv_vars. Add inv_exprs.
17525 (struct iv_cand): Rename depends_on to inv_vars.
17526 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17527 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17528 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17529 (dump_cand): Dump inv_vars.
17530 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17531 (record_invariant, find_depends, add_candidate_1): Ditto.
17532 (set_group_iv_cost, force_var_cost): Ditto.
17533 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17534 (get_computation_cost_at, get_computation_cost): Ditto.
17535 (determine_group_iv_cost_generic): Ditto.
17536 (determine_group_iv_cost_address): Ditto.
17537 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17538 (determine_group_iv_costs): Ditto.
17539 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17540 (iv_ca_set_remove_invariants): Renamed to ...
17541 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17542 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17543 (iv_ca_set_add_invariants): Renamed to ...
17544 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17545 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17546 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17547 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17548 (create_new_ivs): Remove useless dump.
17550 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17552 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17554 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17555 (iv_ca_set_no_cp, create_new_iv): Ditto.
17557 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17559 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17561 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17563 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17564 function tree_check2.
17566 2017-05-02 Martin Liska <mliska@suse.cz>
17568 * doc/gcov.texi: Add missing preposition.
17569 * gcov.c (function_info::function_info): Properly fill up
17570 all member variables.
17572 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17574 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17576 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17578 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17580 2017-05-02 Martin Liska <mliska@suse.cz>
17583 * lto-streamer-in.c (lto_read_tree_1): Remove
17584 LTO_STREAMER_DEBUG.
17585 * lto-streamer.c (struct tree_hash_entry): Likewise.
17586 (struct tree_entry_hasher): Likewise.
17587 (tree_entry_hasher::hash): Likewise.
17588 (tree_entry_hasher::equal): Likewise.
17589 (lto_streamer_init): Likewise.
17590 (lto_orig_address_map): Likewise.
17591 (lto_orig_address_get): Likewise.
17592 (lto_orig_address_remove): Likewise.
17593 * lto-streamer.h: Likewise.
17594 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17595 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17597 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17599 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17600 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17601 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17602 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17603 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17604 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17605 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17606 (mm_maskz_sub_ss): New intrinsics.
17607 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17608 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17609 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17610 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17611 (__builtin_ia32_subss_mask_round): New builtins.
17612 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17613 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17614 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17616 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17617 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17619 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17622 2017-05-02 Martin Jambor <mjambor@suse.cz>
17624 PR tree-optimization/78687
17625 * tree-sra.c (access): New field parent.
17626 (process_subtree_disqualification): New function.
17627 (disqualify_candidate): Call it.
17628 (build_accesses_from_assign): Reset write flag if creating an
17630 (build_access_subtree): Fill in parent field and also prpagate
17631 down grp_write flag.
17632 (create_artificial_child_access): New parameter set_grp_write, set
17633 grp_write to its value.
17634 (propagate_subaccesses_across_link): Also propagate grp_write flag
17636 (propagate_all_subaccesses): Push the closest parent back to work
17637 queue if add_access_to_work_queue returned true.
17639 2017-05-02 Richard Biener <rguenther@suse.de>
17641 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17642 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17643 -fstrict-overflow documentation.
17644 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17645 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17646 flag_strict_overflow.
17647 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17648 * lto-opts.c (lto_write_options): Do not stream it.
17649 * lto-wrapper.c (merge_and_complain): Do not handle it.
17650 * opts.c (default_options_table): Do not set -fstrict-overflow.
17651 (finish_options): Likewise do not clear it when sanitizing.
17652 * simplify-rtx.c (simplify_const_relational_operation): Do not
17653 test flag_strict_overflow.
17655 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17657 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17658 using enabled attribute.
17659 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17660 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17661 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17662 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17663 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17664 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17665 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17666 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17667 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17668 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17670 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17672 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17674 2017-05-02 Richard Biener <rguenther@suse.de>
17676 PR tree-optimization/80591
17678 2017-04-10 Richard Biener <rguenther@suse.de>
17680 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17683 2017-05-02 Richard Biener <rguenther@suse.de>
17685 PR tree-optimization/80549
17686 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17687 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17688 headers if they do not have a preheader.
17690 2017-05-02 Martin Liska <mliska@suse.cz>
17693 * common.opt: Fix typo.
17694 * doc/invoke.texi: Likewise.
17696 2017-05-01 Jan Beulich <jbeulich@suse.com>
17698 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17699 swapping, add (x,x,m,x,n) alternative.
17701 2017-05-01 Nathan Sidwell <nathan@acm.org>
17703 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17704 unnecessary unadjusted_alignment check.
17706 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17709 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17711 * gimplify.c (gimplify_cilk_detach): New function.
17712 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17713 * tree-core.h: Document EXPR_CILK_SPAWN.
17714 * tree.h (EXPR_CILK_SPAWN): Define.
17716 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17718 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17719 to use new fixit_hint representation, using the "replace" logic.
17720 (get_line_span_for_fixit_hint): Likewise.
17721 (layout::print_any_fixits): Likewise.
17722 (selftest::test_one_liner_many_fixits): Rename to...
17723 (selftest::test_one_liner_many_fixits_1): ...this, and update
17724 comment and expected output to reflect that the multiple fix-it
17725 hints are now consolidated into one insertion.
17726 (selftest::test_one_liner_many_fixits_2): New test.
17727 (selftest::test_diagnostic_show_locus_one_liner): Update for
17729 (selftest::test_fixit_consolidation): Update for fix-it API
17731 * diagnostic.c (print_parseable_fixits): Likewise.
17732 * edit-context.c (edited_line::m_line_events): Convert from
17733 auto_vec <line_event *> to auto_vec <line_event>.
17734 (class line_event): Convert from abstract base class to a concrete
17735 class, taking over the role of replace_event.
17736 (class insert_event): Delete.
17737 (class replace_event): Rename to class line_event. Convert to
17739 (edit_context::add_fixits): Reimplement.
17740 (edit_context::apply_insert): Delete.
17741 (edit_context::apply_replace): Rename to...
17742 (edit_context::apply_fixit): ...this. Convert to half-open range.
17743 (edited_file::apply_insert): Delete.
17744 (edited_file::apply_replace): Rename to...
17745 (edited_file::apply_fixit): ...this.
17746 (edited_line::~edited_line): Drop deletion of events.
17747 (edited_line::apply_insert): Delete.
17748 (edited_line::apply_replace): Rename to...
17749 (edited_line::apply_fixit): ...this. Convert to half-open range.
17750 Update for change to type of m_line_events.
17751 * edit-context.h (edit_context::apply_insert): Delete.
17752 (edit_context::apply_replace): Rename to...
17753 (edit_context::apply_fixit): ...this.
17755 2017-05-01 Martin Sebor <msebor@redhat.com>
17757 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17760 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17763 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17764 __get_cpuid_max returns 0.
17765 (__get_cpuid_count): Ditto.
17767 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17769 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17770 replacement expression is another instance of one of its arguments.
17772 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17775 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17776 check for stack push/pop autoinc.
17777 * config/i386/i386.c (ix86_agi_dependent): Return false
17778 if the only reason why modified_in_p returned true is that
17779 addr is SP based and set_insn is a push or pop.
17781 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17783 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17786 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17789 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17790 (account_size_time): Use two predicates - exec_pred and
17792 (evaluate_conditions_for_known_args): Compute both clause and
17794 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17795 (inline_summary_t::duplicate): Update.
17796 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17798 (compute_inline_parameters): Likewise.
17799 (estimate_edge_size_and_time): Update caluclation of time.
17800 (estimate_node_size_and_time): Compute both time and nonspecialized
17802 (estimate_ipcp_clone_size_and_time): Update.
17803 (inline_merge_summary): Update.
17804 (do_estimate_edge_time): Update.
17805 (do_estimate_edge_size): Update.
17806 (do_estimate_edge_hints): Update.
17807 (inline_read_section, inline_write_summary): Stream both new predicates.
17808 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17810 (compute_inlined_call_time): Cleanup.
17811 (big_speedup_p): Update.
17812 (edge_badness): Update.
17813 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17814 (size_time_entry): Replace predicate by exec_predicate and
17815 nonconst_predicate.
17816 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17817 (estimate_edge_time): Return also nonspec_time.
17818 (reset_edge_growth_cache): Update.
17820 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17822 PR rtl-optimization/80491
17823 * ifcvt.c (noce_process_if_block): When looking for x setter
17824 with missing else_bb, don't check only the insn right before
17825 cond_earliest, but look for the last insn that x is modified in
17826 within the same bb.
17828 PR rtl-optimization/80491
17829 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17831 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17833 PR tree-optimization/80487
17834 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17836 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17838 PR tree-optimization/79697
17839 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17840 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17841 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17843 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17844 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17846 2017-04-28 Martin Sebor <msebor@redhat.com>
17848 PR tree-optimization/80523
17849 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17850 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17852 (maybe_warn, format_directive, parse_directive): Use new functions.
17853 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17855 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17857 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17859 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17861 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17862 target_header_dir): Set correctly.
17863 * configure: Regenerated.
17864 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17865 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17866 instead of SYSTEM_HEADER_DIR.
17868 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17870 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17871 (estimate_local_effects): Likewise.
17872 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17873 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17874 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17875 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17876 do_estimate_edge_time, estimate_edge_time): Likewise.
17877 * ipa-inline-analysis.c (estimate_node_size_and_time,
17878 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17879 (estimate_time_after_inlining): Remove.
17881 2017-04-28 Martin Liska <mliska@suse.cz>
17883 * doc/gcov.texi: Enhance documentation of gcov.
17885 2017-04-28 Martin Liska <mliska@suse.cz>
17887 * doc/gcov.texi: Sort options in alphabetic order.
17888 * doc/gcov-dump.texi: Likewise.
17889 * doc/gcov-tool.texi: Likewise.
17890 * gcov.c (print_usage): Likewise.
17891 * gcov-dump.c (print_usage): Likewise.
17892 * gcov-tool.c (print_merge_usage_message): Likewise.
17893 (print_rewrite_usage_message): Likewise.
17894 (print_overlap_usage_message): Likewise.
17896 2017-04-28 Martin Liska <mliska@suse.cz>
17898 PR gcov-profile/53915
17899 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17901 2017-04-28 Martin Liska <mliska@suse.cz>
17903 PR gcov-profile/79891
17904 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17905 is marked by compiler as living on a line.
17906 (get_cycles_count): Remove usage of the union.
17907 (output_intermediate_file): Likewise.
17908 (find_source): Fix GNU coding style.
17909 (accumulate_line_counts): Remove old non-all block mode.
17910 (output_lines): Remove usage of the union.
17911 * profile.c (output_location): Include all BBs, even if
17912 belonging to a same line (and file) as a previous BB.
17914 2017-04-28 Martin Liska <mliska@suse.cz>
17916 * gcov.c (process_args): Handle new argument 'w'.
17917 (read_graph_file): Assign ID to BBs.
17918 (output_branch_count): Display BB # if verbose flag is set.
17919 (output_lines): Likewise for arcs.
17920 (print_usage): Add '--verbose' option help.
17921 * doc/gcov.texi: Document --verbose (-w) option.
17923 2017-04-28 Martin Liska <mliska@suse.cz>
17925 * gcov.c (struct block_location_info): New struct.
17926 (process_file): Fill up the new structure.
17927 (read_graph_file): Replace usage of encoding by the newly added
17929 (add_line_counts): Likewise.
17930 (accumulate_line_counts): Remove usage of the union.
17931 (function_info::function_info): New function.
17932 (function_info::~function_info): Likewise.
17933 (process_file): Call delete instead of release_function.
17934 (release_function): Release the function.
17935 (release_structures): Call delete instead of release_function.
17936 (solve_flow_graph): Replace usage of num_blocks.
17937 (find_exception_blocks): Likewise.
17938 (output_lines): Fix GNU coding style.
17940 2017-04-28 Martin Liska <mliska@suse.cz>
17943 * coverage.c (coverage_remove_note_file): New function.
17944 * coverage.h: Declare the function.
17945 * toplev.c (finalize): Clean if an error has been seen.
17947 2017-04-28 Martin Liska <mliska@suse.cz>
17949 PR gcov-profile/80031
17950 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17951 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17952 * gcov.c (read_graph_file): Read just number of blocks.
17953 * profile.c (branch_prob): Do not stream 0 flags per a basic
17956 2017-04-28 Martin Liska <mliska@suse.cz>
17958 * gcov-dump.c (tag_*): Add new argument to declarations.
17959 (dump_gcov_file): Likewise.
17960 (tag_blocks): Add and use new argument depth.
17961 (tag_arcs): Likewise.
17962 (tag_lines): Likewise.
17963 (tag_counters): Likewise.
17964 (tag_summary): Likewise.
17965 (dump_working_sets): Use depth to do a proper indentation.
17967 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17970 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17971 * symtab.c (symtab_node::debug_symtab): Move definition here.
17973 2017-04-28 Richard Biener <rguenther@suse.de>
17975 * lto-streamer.h (LTO_major_version): Bump to 7.
17977 2017-04-28 Richard Biener <rguenther@suse.de>
17979 * tree-vrp.c (assert_info): New struct.
17980 (add_assert_info): New helper.
17981 (register_edge_assert_for_2): Refactor to add asserts to a vector
17983 (register_edge_assert_for_1): Likewise.
17984 (register_edge_assert_for): Likewise.
17985 (finish_register_edge_assert_for): New helper actually registering
17986 asserts where live on edge.
17987 (find_conditional_asserts): Adjust.
17988 (find_switch_asserts): Likewise.
17989 (evrp_dom_walker::try_find_new_range): Generalize.
17990 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17992 2017-04-27 Marek Polacek <polacek@redhat.com>
17995 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17996 arg10 and arg11 to itype.
17998 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18000 * doc/extend.texi (Object Size Checking): Improve grammar.
18002 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18005 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18006 that the logic for permitting reciprocal estimates matches that
18009 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18012 * tree.c (type_cache_hasher::equal): Only compare
18013 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18014 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18015 non-aggregate element types.
18016 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18017 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18019 2017-04-27 Richard Biener <rguenther@suse.de>
18021 PR middle-end/80533
18022 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18023 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18024 keeping a reference to a trailing array.
18026 2017-04-27 Richard Biener <rguenther@suse.de>
18028 PR middle-end/80539
18029 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18030 being in loop-closed SSA form conservatively.
18031 (chrec_fold_multiply_poly_poly): Likewise.
18033 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18035 PR middle-end/79665
18036 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18037 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18039 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18042 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18043 (aarch64_function_arg_alignment): Return unsigned int again, but still
18044 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18045 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18046 Don't emit -Wpsabi note.
18047 (aarch64_function_arg_boundary): Likewise.
18048 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18051 2017-04-26 Nathan Sidwell <nathan@acm.org>
18053 * tree.h (crc32_unsigned_n): Declare.
18054 (crc32_unsigned, crc32_unsigned): Make inline.
18055 * tree.c (crc32_unsigned_bits): Replace with ...
18056 (crc32_unsigned_n): ... this.
18057 (crc32_unsigned, crc32_byte): Remove.
18058 (crc32_string): Remove unnecessary braces.
18060 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18062 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18063 * ipa-inline-analysis.c (MAX_TIME): Remove.
18064 (account_size_time): Use sreal for time.
18065 (dump_inline_summary): Update.
18066 (estimate_function_body_sizes): Update.
18067 (estimate_edge_size_and_time): Update.
18068 (estimate_calls_size_and_time): Update.
18069 (estimate_node_size_and_time): Update.
18070 (inline_merge_summary): Update.
18071 (inline_update_overall_summary): Update.
18072 (estimate_time_after_inlining): Update.
18073 (inline_read_section): Update.
18074 (inline_write_summary): Update.
18075 * ipa-inline.c (compute_uninlined_call_time): Update.
18076 (compute_inlined_call_time): Update.
18077 (recursive_inlining): Update.
18078 (inline_small_functions): Update.
18079 (dump_overall_stats): Update.
18080 * ipa-inline.h: Include sreal.h.
18081 (size_time_entry): Turn time to sreal.
18082 (inline_summary): Turn self_time nad time to sreal.
18084 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18086 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18088 (sreal::stream_out, sreal::stream_in): New.
18089 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18091 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18093 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18096 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18099 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18101 (dimode_scalar_chain::compute_convert_gain): Ditto.
18102 (dimode_scalar_chain::make_vector_copies): Ditto.
18103 (dimode_scalar_chain::convert_reg): Ditto.
18104 (dimode_scalar_chain::convert_insn): Ditto.
18105 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18106 (VI248_AVX512BW_1): New mode iterator.
18107 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18108 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18111 2017-04-25 Martin Sebor <msebor@redhat.com>
18113 PR tree-optimization/80497
18114 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18115 constants are representable in HOST_WIDE_INT.
18116 (parse_directive): Ditto.
18118 2017-04-25 Martin Sebor <msebor@redhat.com>
18121 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18122 (new_zero_array): Adjust signature.
18123 (dom_info::dom_init): Used unsigned rather that size_t.
18124 (dom_info::dom_info): Same.
18126 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18127 Jakub Jelinek <jakub@redhat.com>
18130 * config/arm/arm.c: Include gimple.h.
18131 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18132 returns negative, increment ncrn only if it returned positive.
18133 (arm_needs_doubleword_align): Return int instead of bool,
18134 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18135 members, but if there is any such non-FIELD_DECL
18136 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18137 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18138 returns negative, increment nregs only if it returned positive.
18139 (arm_setup_incoming_varargs): Likewise.
18140 (arm_function_arg_boundary): Emit -Wpsabi note if
18141 arm_needs_doubleword_align returns negative, return
18142 DOUBLEWORD_ALIGNMENT only if it returned positive.
18144 2017-04-25 Marek Polacek <polacek@redhat.com>
18147 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18148 first argument to type.
18150 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18153 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18154 type checks to test for compatibility instead of equality.
18156 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18157 Jakub Jelinek <jakub@redhat.com>
18160 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18162 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18163 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18164 the alignment computation, but return their maximum in warn_alignment.
18165 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18166 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18168 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18169 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18172 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18174 * config/arc/simdext.md (dmpyh): Fix typo.
18176 2017-04-25 Richard Biener <rguenther@suse.de>
18178 PR tree-optimization/80492
18179 * alias.c (compare_base_decls): Handle registers with asm
18180 specification conservatively.
18181 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18182 compare_base_decls returning dont-know properly.
18184 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18186 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18187 (legitimate_offset_address_p): New function.
18188 (arc_legitimate_address_p): Use above function.
18190 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18192 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18194 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18196 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18197 ACCH registers whenever they are available.
18199 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18201 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18202 double regs fix when not used.
18204 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18206 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18208 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18209 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18211 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18213 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18214 when emitting short ADD instructions.
18216 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18218 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18220 (cmpsi_cc_c_insn): Likewise.
18221 (cbranchsi4_scratch): Compute proper instruction length using
18222 compact_hreg_operand.
18223 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18225 2017-04-25 Richard Biener <rguenther@suse.de>
18227 PR middle-end/80509
18228 * passes.c (pass_manager::pass_manager): Initialize
18229 m_name_to_pass_map.
18231 2017-04-25 Richard Biener <rguenther@suse.de>
18233 PR tree-optimization/79201
18234 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18236 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18239 * config/s390/vector.md: Split MEM->GPR vector moves for
18240 non-s_operand addresses.
18242 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18245 * config/s390/predicates.md (reload_const_wide_int_operand): New
18247 * config/s390/s390.md ("movti"): Remove d/P alternative.
18248 ("movti_bigconst"): New pattern definition.
18250 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18253 * s390-protos.h (s390_expand_cs_hqi): Removed.
18254 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18255 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18256 modes as well as CCZ1mode and CCZmode.
18257 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18258 signature of s390_emit_compare_and_swap.
18259 (s390_expand_cs_hqi): Likewise, make static.
18260 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18261 compare-and-swap, in some cases.
18262 (s390_expand_cs): Wrapper function.
18263 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18265 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18266 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18267 patterns for small and large integers. Forbid symref memory operands.
18268 Move expander to s390.c. Require cc register.
18269 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18270 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18271 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18272 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18273 symref memory operands. Remove CC mode and call s390_match_ccmode
18275 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18277 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18279 * config/s390/s390.md (define_peephole2): New peephole to help
18280 combining the load-and-test pattern with volatile memory.
18282 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18284 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18285 with CCZmode for TARGET_Z196.
18287 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18289 PR rtl-optimization/80501
18290 * combine.c (make_compound_operation_int): Set subreg_code to SET
18291 even for AND with mask of the sign bit of mode.
18293 PR rtl-optimization/80500
18294 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18295 sum's initial value.
18297 2017-04-25 Julian Brown <julian@codesourcery.com>
18298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18300 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18302 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18304 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18306 2017-04-25 Julian Brown <julian@codesourcery.com>
18307 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18309 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18310 (thunderx2t99_sha): New Reservation.
18312 2017-04-25 Julian Brown <julian@codesourcery.com>
18313 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18315 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18316 type for 1-element load.
18318 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18320 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18322 2017-04-24 Martin Jambor <mjambor@suse.cz>
18324 PR tree-optimization/80293
18325 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18326 char arrays not totally scalarizable if it is false.
18327 (analyze_all_variable_accesses): Pass correct value in the new
18328 parameter. Add a statistics counter.
18330 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18332 PR middle-end/79931
18333 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18335 2017-04-24 Richard Biener <rguenther@suse.de>
18337 PR tree-optimization/80494
18338 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18339 out for complex types.
18341 2017-04-24 Richard Biener <rguenther@suse.de>
18343 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18344 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18345 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18346 (DFS): Adjust and never fail.
18347 (sccvn_dom_walker::fail): Remove.
18348 (sccvn_dom_walker::before_dom_children): Adjust.
18349 (run_scc_vn): Likewise and never fail.
18350 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18351 (pass_fre::execute): Likewise.
18353 2017-04-24 Richard Biener <rguenther@suse.de>
18355 PR tree-optimization/79725
18356 * tree-ssa-sink.c (statement_sink_location): Return whether
18357 failure reason was zero uses. Move that check later.
18358 (sink_code_in_bb): Deal with zero uses by removing the stmt
18361 2017-04-24 Richard Biener <rguenther@suse.de>
18364 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18365 pointer-based references.
18367 2017-04-24 Richard Biener <rguenther@suse.de>
18370 * pass_manager.h (pass_manager::operator new): Remove.
18371 (pass_manager::operator delete): Likewise.
18372 * passes.c (pass_manager::operator new): Remove.
18373 (pass_manager::operator delete): Likewise.
18374 (pass_manager::pass_manager): Zero individual pass members.
18376 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18379 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18380 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18381 Check "XEXP (src, 1)" operand here.
18382 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18383 Check "XEXP (src, 1)" operand here.
18384 (dimode_scalar_chain::make_vector_copies): Detect count register
18385 of a shift instruction. Zero extend count register from QImode
18386 to DImode to satisfy vector shift pattern count operand predicate.
18387 Substitute vector shift count operand with a DImode copy.
18388 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18391 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18393 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18394 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18395 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18396 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18397 (UNSPEC_NOREX_MEM): Remove definition.
18399 2017-04-21 Richard Biener <rguenther@suse.de>
18401 PR tree-optimization/79547
18402 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18403 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18404 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18405 without any constraints.
18407 2017-04-21 Richard Biener <rguenther@suse.de>
18409 PR tree-optimization/78847
18410 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18412 2017-04-21 Richard Biener <rguenther@suse.de>
18414 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18415 (build_distinct_type_copy): Likewise.
18416 (build_variant_type_copy): Likewise.
18417 * tree.c (build_qualified_type): Pass down mem-stat info.
18418 (build_distinct_type_copy): Likewise.
18419 (build_variant_type_copy): Likewise.
18421 2017-04-21 Richard Biener <rguenther@suse.de>
18423 PR tree-optimization/80237
18424 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18426 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18427 for a simplified result.
18429 2016-04-21 Richard Biener <rguenther@suse.de>
18431 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18432 sth as strict as a simple_iv but a chrec without symbols and an
18433 operand defined in the loop we are peeling (and not some subloop).
18434 (propagate_constants_for_unrolling): Propagate all constants.
18436 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18439 * config/i386/i386.c (print_reg): Remove assert for disalowed
18440 regno values, call output_operand_lossage instead.
18442 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18445 * config/i386/constraints.md (Yc): New register constraint.
18446 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18447 Use Yc constraint for alternative 2 of operand 0. Remove
18448 preferred_for_speed attribute.
18450 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18452 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18453 lastprivate clauses in SIMT case.
18455 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18457 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18459 2017-04-20 Richard Biener <rguenther@suse.de>
18461 PR tree-optimization/57796
18462 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18463 as N scalar stores.
18464 (vect_model_load_cost): Cost gathers as N scalar loads.
18466 2017-04-20 Richard Biener <rguenther@suse.de>
18468 * ggc-page.c (ggc_allocated_p): Rename to ...
18469 (safe_lookup_page_table_entry): ... this and return the lookup
18471 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18473 2017-04-20 Richard Biener <rguenther@suse.de>
18475 PR tree-optimization/80453
18476 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18477 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18478 from the conditions.
18479 (vn_phi_eq): Pass them down.
18480 (vn_phi_lookup): Record them.
18481 (vn_phi_insert): Likewise.
18483 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18485 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18486 uninitialized variable warning to avoid buffer overrun.
18488 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18491 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18492 is suppressed for '{ 0 }' in C.
18494 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18496 * BASE-VER: Set to 8.0.0.
18498 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18500 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18501 priority .init_array and .fini_array section with SECTION_NOTYPE
18504 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18506 PR middle-end/80423
18507 * tree.h (build_array_type): Add typeless_storage default argument.
18508 * tree.c (type_cache_hasher::equal): Also compare
18509 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18510 (build_array_type): Add typeless_storage argument, set
18511 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18513 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18514 (build_array_type): Likewise. Add typeless_storage argument.
18516 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18517 Jakub Jelinek <jakub@redhat.com>
18519 PR tree-optimization/80426
18520 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18521 operation on symbolic operands, also compute the overflow for the
18522 invariant part when the operation degenerates into a negation.
18524 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18527 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18528 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18531 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18533 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18536 * config/avr/avr.c (tree.h): Include it.
18537 (cgraph.h): Include it.
18538 (avr_encode_section_info): Don't warn for uninitialized progmem
18539 variable if it's just an alias.
18541 2017-04-19 Richard Biener <rguenther@suse.de>
18544 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18545 when needed by AutoPGO.
18547 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18550 * doc/lto.texi: Remove an extra 'that'.
18552 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18554 PR rtl-optimization/80429
18555 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18556 are only used in debug insns.
18558 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18559 Vladimir Makarov <vmakarov@redhat.com>
18561 * config/sparc/predicates.md (input_operand): Add comment. Return
18562 true for any memory operand when LRA is in progress.
18563 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18565 2017-04-18 Jeff Law <law@redhat.com>
18568 * mips.md ({return,simple_return}_internal): Do not overwrite
18571 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18573 PR tree-optimization/80443
18574 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18575 instead of adding 1, subtract -1 and similarly instead of subtracting
18578 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18580 PR rtl-optimization/80357
18581 * haifa-sched.c (tmp_bitmap): New variable.
18582 (model_recompute): Handle duplicate use records.
18583 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18584 (free_global_sched_pressure_data): Free it.
18586 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18589 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18590 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18591 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18592 instead of SYSTEM_HEADER_DIR.
18594 2017-04-18 Jeff Law <law@redhat.com>
18596 PR middle-end/80422
18597 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18598 predecessors after walking up the insn chain.
18600 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18603 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18604 sizetype type into debug info.
18606 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18609 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18610 unneeded test for TARGET_UPPER_REGS_SF.
18611 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18613 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18616 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18617 instead of gsi_after_labels.
18619 2017-04-18 Jeff Law <law@redhat.com>
18621 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18625 2017-04-13 Jeff Law <law@redhat.com>
18626 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18627 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18629 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18632 * config/avr/avr.c (intl.h): Include it.
18633 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18635 2017-04-18 Martin Liska <mliska@suse.cz>
18637 PR gcov-profile/78783
18638 * gcov-tool.c (gcov_output_files): Validate that destination
18639 file is either removed by the tool or by a user.
18641 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18642 Guy Benyei <guybe@mellanox.com>
18644 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18645 block, and do not negate it, the stored id is already negative.
18647 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18649 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18651 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18654 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18655 masks of options that should be turned off if the VSX vector
18656 options are turned off.
18657 (OTHER_P8_VECTOR_MASKS): Likewise.
18658 (OTHER_VSX_VECTOR_MASKS): Likewise.
18659 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18660 rs6000_disable_incompatible_switches to validate no type switches
18662 (rs6000_incompatible_switch): New function to disallow turning on
18663 other vector options if -mno-vsx, -mno-power8-vector, or
18664 -mno-power9-vector are specified.
18666 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18668 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18670 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18672 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18673 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18674 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18675 (ARG_POINTER_CFA_OFFSET): Likewise.
18677 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18679 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18680 conditions to take advantage of various optimizations.
18682 2017-04-13 Jeff Law <law@redhat.com>
18684 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18685 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18686 (zero_extendsidi2_dext): Likewise.
18688 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18691 * fold-const.c (fold_ternary_loc): Revert
18692 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18695 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18697 PR rtl-optimization/80343
18698 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18699 new scratch pseudo.
18701 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18704 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18705 to ubsan_encode_value.
18707 2017-04-13 Jeff Law <law@redhat.com>
18709 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18710 appearing in DEBUG_INSNs.
18712 2017-04-13 Martin Liska <mliska@suse.cz>
18714 PR gcov-profile/80413
18715 * gcov-io.c (gcov_write_string): Copy to buffer just when
18716 allocated size is greater than zero.
18718 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18721 * dwarf2out.c (decls_for_scope): Ignore declarations of
18722 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18724 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18727 * ipa-visibility.c (non_local_p): Fix typos.
18728 (localize_node): When localizing symbol in same comdat group,
18729 dissolve the group only when we know external symbols are going
18731 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18733 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18735 PR tree-optimization/79390
18736 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18737 order does not result in usable sequence, retry with reversed operand
18743 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18744 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18745 op0 instead of fold_convert_loc (loc, type, arg0).
18747 2017-04-12 Jeff Law <law@redhat.com>
18749 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18750 has a delay slot in the generated code.
18752 * config/cris/cris.md (cris_preferred_reload_class): Return
18753 GENNONACR_REGS rather than GENERAL_REGS.
18755 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18758 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18759 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18760 signedness of the result type.
18762 2017-04-12 Richard Biener <rguenther@suse.de>
18763 Jeff Law <law@redhat.com>
18765 PR tree-optimization/80359
18766 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18767 trim stores to TARGET_MEM_REFs.
18769 2017-04-12 Richard Biener <rguenther@suse.de>
18771 PR tree-optimization/79390
18772 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18773 threading case even more.
18775 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18778 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18779 for quad_address_p for TImode, instead of just not indexed_address.
18781 2017-04-12 Richard Biener <rguenther@suse.de>
18782 Bernd Edlinger <bernd.edlinger@hotmail.de>
18784 PR middle-end/79671
18785 * alias.c (component_uses_parent_alias_set_from): Handle
18786 TYPE_TYPELESS_STORAGE.
18787 (get_alias_set): Likewise.
18788 * tree-core.h (tree_type_common): Add typeless_storage flag.
18789 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18790 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18791 for types containing members with TYPE_TYPELESS_STORAGE.
18792 (place_field): Likewise.
18793 (layout_type): Likewise for ARRAY_TYPE.
18794 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18795 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18796 TYPE_TYPELESS_STORAGE.
18797 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18799 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18802 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18803 first argument to type.
18805 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18809 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18810 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18811 (rs6000_expand_binop_builtin): Likewise.
18812 (rs6000_expand_ternop_builtin): Likewise; also add missing
18813 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18814 vshasigma built-ins.
18815 * doc/extend.texi: Document that vec_xxpermdi's third argument
18816 must be a constant.
18818 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18820 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18821 Use shift_const cost parameter when calculating gain of STV shifts.
18823 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18825 PR rtl-optimization/70478
18826 * lra-constraints.c (process_alt_operands): Check memory for
18827 disfavoring memory insn operand.
18829 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18831 PR middle-end/80100
18832 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18833 left shift in unsigned HOST_WIDE_INT type.
18835 PR rtl-optimization/80385
18836 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18837 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18840 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18841 if they have any depend clauses.
18843 2017-04-11 Martin Liska <mliska@suse.cz>
18846 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18847 * ipa-split.c (split_function): Create a local comdat symbol
18848 if caller is in a comdat group.
18850 2017-04-11 Martin Liska <mliska@suse.cz>
18853 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18856 2017-04-11 Martin Sebor <msebor@redhat.com>
18858 PR middle-end/80364
18859 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18860 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18862 (directive::set_width, directive::set_precision, format_character):
18864 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18867 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18870 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18871 conflict, set target->arch_name instead of target->cpu_name.
18873 2017-04-11 Richard Biener <rguenther@suse.de>
18875 PR tree-optimization/80374
18876 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18877 build_zero_cst, remove fold_convertible_p check again.
18879 2017-04-11 Martin Liska <mliska@suse.cz>
18882 * ubsan.c (instrument_object_size): Do not instrument register
18885 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18888 * config/i386/i386-builtin-types.def
18889 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18890 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18891 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18892 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18893 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18894 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18895 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18896 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18897 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18898 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18899 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18900 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18901 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18902 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18903 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18904 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18905 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18906 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18907 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18908 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18909 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18910 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18911 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18912 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18913 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18914 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18915 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18916 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18917 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18918 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18919 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18920 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18921 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18922 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18923 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18924 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18925 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18926 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18927 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18928 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18929 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18930 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18931 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18932 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18933 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18934 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18936 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18937 flag to second_arg_count, handle 4 argument function type _COUNT
18938 aliases, handle second_arg_count on second argument rather than last.
18940 2017-04-10 Jeff Law <law@redhat.com>
18942 PR tree-optimization/80374
18943 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18944 record anything if we can not convert integer_zero_node to the
18947 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18950 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18951 Enhance special handling given to the TARGET_P9_MINMAX option in
18952 relation to certain other options.
18954 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18956 PR tree-optimization/80153
18957 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18958 remove POINTER_PLUS_EXPR's base part directly, rather than through
18961 2017-04-10 Richard Biener <rguenther@suse.de>
18962 Bin Cheng <bin.cheng@arm.com>
18964 PR tree-optimization/80153
18965 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18966 the first element of pointer type aff_tree. Build result expr in
18968 (add_elt_to_tree): Convert to type unconditionally. Remove other
18969 fold_convert calls.
18970 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18971 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18973 2017-04-10 Richard Biener <rguenther@suse.de>
18975 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18978 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18980 PR rtl-optimization/70478
18981 * lra-constraints.c (curr_small_class_check): New.
18982 (update_and_check_small_class_inputs): New.
18983 (process_alt_operands): Update curr_small_class_check. Disfavor
18984 alternative insn memory operands. Check available regs for small
18987 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18990 * config/mips/mips.opt (-mvirt): Update description.
18991 * doc/invoke.texi (-mvirt): Likewise.
18993 2017-04-10 Richard Biener <rguenther@suse.de>
18995 PR middle-end/80362
18996 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18997 looking for NEGATE_EXPR in -A / -B to A / B folding.
18999 2017-04-10 Martin Liska <mliska@suse.cz>
19001 PR gcov-profile/80224
19002 * gcov.c (print_usage): Fix usage string.
19003 (get_gcov_intermediate_filename): Remove.
19004 (output_gcov_file): Use both for normal and intermediate format.
19005 (generate_results): Do not initialize special file for
19006 intermediate format.
19008 2017-04-10 Richard Biener <rguenther@suse.de>
19010 PR tree-optimization/80304
19011 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19014 2017-04-10 Nathan Sidwell <nathan@acm.org>
19017 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19018 (rs6000_init_builtins): Use it.
19020 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19022 * config/arm/arm.md (<mrc>): Add mode to SET source.
19023 (<mrrc>): Likewise.
19025 2017-04-10 Richard Biener <rguenther@suse.de>
19027 PR middle-end/80344
19028 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19030 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19033 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19034 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19035 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19036 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19037 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19038 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19039 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19040 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19041 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19042 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19043 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19044 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19045 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19046 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19047 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19048 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19049 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19050 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19051 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19052 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19053 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19054 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19055 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19057 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19059 PR rtl-optimization/70478
19060 * lra-constraints.c: Reverse the last patch.
19062 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19064 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19065 Add comment for WCHAR_T.
19067 2017-04-08 Martin Liska <mliska@suse.cz>
19070 2017-04-07 Martin Liska <mliska@suse.cz>
19073 * ipa-split.c (split_function): Add function part to a same comdat
19076 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19079 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19081 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19083 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19084 (vec_pairing): ...to this.
19085 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19086 (rs6000_sched_init): Adjust for name change.
19087 (struct rs6000_sched_context): Likewise.
19088 (rs6000_init_sched_context): Likewise.
19089 (rs6000_set_sched_context): Likewise.
19091 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19097 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19099 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19100 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19101 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19103 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19105 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19107 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19109 PR rtl-optimization/70703
19110 * ira-color.c (update_conflict_hard_regno_costs): Use
19111 int64_t instead of HOST_WIDE_INT.
19113 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19115 PR rtl-optimization/70478
19116 * lra-constraints.c (process_alt_operands): Disfavor alternative
19117 insn memory operands.
19119 2017-04-07 Jeff Law <law@redhat.com>
19121 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19122 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19124 2017-04-07 Martin Liska <mliska@suse.cz>
19127 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19128 Show error message instead of an ICE.
19130 2017-04-07 Martin Liska <mliska@suse.cz>
19133 * ipa-split.c (split_function): Add function part to a same comdat
19136 2017-04-07 Richard Biener <rguenther@suse.de>
19138 PR middle-end/80341
19139 * tree.c (get_unwidened): Also handle ! for_type case for
19141 * convert.c (do_narrow): Split out from ...
19142 (convert_to_integer_1): ... here. Do not pass final truncation
19143 type to get_unwidened for TRUNC_DIV_EXPR.
19145 2017-04-07 Richard Biener <rguenther@suse.de>
19147 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19149 (aff_combination_const): Adjust.
19150 (aff_combination_scale): Likewise.
19151 (aff_combination_add_elt): Likewise.
19152 (aff_combination_add_cst): Likewise.
19153 (aff_combination_convert): Likewise.
19154 (add_elt_to_tree): Likewise. Remove unused argument.
19155 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19157 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19159 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19161 * config/arm/arm.c (arm_default_short_enums): Use
19162 ARM_DEFAULT_SHORT_ENUMS.
19163 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19165 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19168 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19169 members with redundant out-of-class redeclaration.
19171 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19174 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19175 * config/i386/i386.md (*zero_extendsidi2):
19176 Add (?*x,*x) and (?*v,*v) alternatives.
19178 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19181 * config/i386/i386.c (ix86_expand_builtin)
19182 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19183 mode from insn data. Convert operands to insn operand mode.
19184 Copy operands that don't satisfy insn predicate to a register.
19186 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19188 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19191 2017-04-06 Richard Biener <rguenther@suse.de>
19193 PR tree-optimization/80334
19194 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19195 preserve alignment of accesses.
19197 2017-04-06 Richard Biener <rguenther@suse.de>
19199 PR tree-optimization/80262
19200 * tree-sra.c (build_ref_for_offset): Preserve address-space
19202 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19203 Drop useless address-space information on MEM_REF offsets.
19205 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19207 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19209 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19211 PR rtl-optimization/70703
19212 * ira-color.c (update_conflict_hard_regno_costs): Use
19213 HOST_WIDE_INT instead of long.
19215 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19218 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19219 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19221 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19222 for x86_64 target. Handle -m3dnowa option.
19224 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19226 PR rtl-optimization/70703
19227 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19228 (update_conflict_hard_regno_costs): Use long instead of unsigned
19229 arithmetic for cost calculation.
19231 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19232 Bernd Edlinger <bernd.edlinger@hotmail.de>
19235 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19238 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19241 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19242 ptr_mode with Pmode throughout.
19243 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19244 into probe_stack_range and use DImode.
19246 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19249 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19250 call_eh_return is true.
19252 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19254 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19255 Initialize last_match_fntype_index.
19257 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19260 * tree-nvr.c: Include internal-fn.h.
19261 (pass_return_slot::execute): Ignore internal calls without
19264 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19265 Richard Biener <rguenther@suse.de>
19268 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19269 captures used multiple times, except for the last use.
19270 * generic-match-head.c: Include gimplify.h.
19272 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19274 PR tree-optimization/79390
19275 * target.h (struct noce_if_info): Declare.
19276 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19277 * target.def (noce_conversion_profitable_p): New target hook.
19278 * ifcvt.h (struct noce_if_info): New type, moved from ...
19279 * ifcvt.c (struct noce_if_info): ... here.
19280 (noce_conversion_profitable_p): Renamed to ...
19281 (default_noce_conversion_profitable_p): ... this. No longer
19283 (noce_try_store_flag_constants, noce_try_addcc,
19284 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19285 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19286 instead of noce_conversion_profitable_p.
19287 * config/i386/i386.c: Include ifcvt.h.
19288 (ix86_option_override_internal): Don't override
19289 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19290 (ix86_noce_conversion_profitable_p): New function.
19291 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19292 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19293 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19294 * doc/tm.texi: Regenerated.
19296 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19298 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19301 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19304 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19305 instructions for small multiply cores.
19307 2017-04-04 Jeff Law <law@redhat.com>
19309 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19311 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19312 that are not set by the loop over the elements.
19314 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19317 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19318 int mode, convert_modes it to mode as unsigned, otherwise use
19319 lowpart_subreg to mode rather than SImode.
19320 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19321 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19322 Use DImode instead of SImode for the shift count operand.
19323 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19326 2017-04-04 Richard Biener <rguenther@suse.de>
19328 PR middle-end/80281
19329 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19330 arithmetic done for the negate or the plus. Simplify.
19331 (A - (-B) -> A + B): Likewise.
19332 * fold-const.c (split_tree): Make sure to not negate pointers.
19334 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19336 PR rtl-optimization/60818
19337 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19338 a compare of comparisons with the thing compared if this results
19339 in a different machine mode.
19341 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19343 * alias.c (base_alias_check): Fix typo in comment.
19344 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19345 * cgraphunit.c (symbol_table::compile): Likewise.
19346 * collect2.c (maybe_run_lto_and_relink): Likewise.
19347 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19348 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19349 * config/avr/avr.c (avr_map_op_t): Likewise.
19350 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19351 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19352 * config/epiphany/epiphany.md (movcc): Likewise.
19353 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19354 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19356 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19357 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19358 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19359 * config/sh/sh.c (sh_rtx_costs): Likewise.
19360 * fold-const.c (fold_truth_andor): Likewise.
19361 * genautomata.c (collapse_flag): Likewise.
19362 * gengtype.h (struct type::u::s): Likewise.
19363 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19364 * input.c (FORMAT_AMOUNT): Likewise.
19365 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19366 (known_aggs_to_agg_replacement_list): Likewise.
19367 * ipa-inline-analysis.c: Likewise.
19368 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19369 * ipa-polymorphic-call.c
19370 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19371 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19372 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19374 * modulo-sched.c (apply_reg_moves): Likewise.
19375 * omp-expand.c (build_omp_regions_1): Likewise.
19376 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19377 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19378 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19379 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19380 * value-prof.c: Likewise.
19381 * var-tracking.c (val_reset): Likewise.
19383 2017-04-03 Richard Biener <rguenther@suse.de>
19385 PR tree-optimization/80275
19386 * fold-const.c (split_address_to_core_and_offset): Handle
19389 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19391 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19392 descriptors is at least equal to that of functions.
19394 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19396 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19398 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19401 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19402 (mov<IMOD4:mode>): New expander.
19403 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19405 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19407 PR rtl-optimization/79405
19408 * fwprop.c (propagations_left): New variable.
19409 (forward_propagate_into): Decrement it.
19410 (fwprop_init): Initialize it.
19411 (fw_prop): If the variable has reached zero, stop propagating.
19412 (fwprop_addr): Ditto.
19414 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19417 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19418 a FUNCTION_DECL, pass it as decl instead of origin to
19421 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19423 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19426 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19429 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19430 TARGET_VSX_SMALL_INTEGER.
19432 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19434 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19435 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19437 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19439 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19440 extraction from odd-numbered MSA register.
19442 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19444 PR middle-end/80173
19445 * expmed.c (store_bit_field_1): Don't attempt to create
19446 a word subreg out of hard registers wider than word if they
19447 have HARD_REGNO_NREGS of 1 for their mode.
19449 PR middle-end/80163
19450 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19451 conversions to integer types wider than word and pointer.
19454 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19455 (rtx_equal_for_cselib_p): Pass 0 to it.
19456 * cselib.c (cselib_hasher::equal): Likewise.
19457 (rtx_equal_for_cselib_1): Add depth argument. If depth
19458 is 128, don't look up VALUE locs and punt. Increment
19459 depth in recursive calls when walking VALUE locs.
19461 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19463 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19464 (make_gcov_file_name): Use the canonical path name for generating
19466 (read_line): Fix handling of files with ascii null bytes.
19468 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19470 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19471 to initialise a vector register instead
19472 of using a const_int.
19474 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19476 PR translation/80189
19477 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19478 diagnostic messages.
19480 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19483 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19484 (dfp_diex_<mode>): Update mode of operand 1.
19485 * doc/extend.texi (dxex, dxexq): Document change to return type.
19486 (diex, diexq): Document change to argument type.
19488 2017-03-30 Martin Jambor <mjambor@suse.cz>
19491 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19492 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19493 it reflects the signature changes performed at the callee side.
19494 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19495 to cgraph_build_function_type_skip_args.
19496 (build_function_decl_skip_args): Adjust call to the above function.
19498 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19501 * config/i386/sse.md
19502 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19503 register as dest whenever it is a MEM not rtx_equal_p to the
19504 corresponding dup operand, and when forcing into reg move the
19505 reg into the memory afterwards.
19506 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19507 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19508 for the force_reg mode.
19509 (avx512vl_vextractf128<mode>): Use register as dest either
19510 always when a MEM, or when it is a MEM not rtx_equal_p to the
19511 corresponding dup operand, or even not when it is a CONST_VECTOR
19512 depending on the mode and lo vs. hi.
19513 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19515 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19516 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19517 Likewise. Require that operands[2] is even.
19518 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19519 Remove extraneous parens. Require that operands[2] is a multiple
19521 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19522 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19523 disallow memory then.
19525 2017-03-30 Richard Biener <rguenther@suse.de>
19527 PR tree-optimization/77498
19528 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19529 to non-constants over backedges.
19531 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19533 PR rtl-optimization/80233
19534 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19535 as last_combined_insn. Do not test for BARRIER_P separately.
19537 2017-03-29 Andreas Schwab <schwab@suse.de>
19540 * calls.c (prepare_call_address): Convert funexp to Pmode before
19541 copying to temp reg.
19543 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19545 PR tree-optimization/80158
19546 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19547 Handle possible future case of more than one alternate
19549 (replace_rhs_if_not_dup): Likewise.
19550 (replace_one_candidate): Likewise.
19552 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19554 PR rtl-optimization/80193
19555 * ira.c (ira): Do not check allocation for LRA.
19557 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19559 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19560 (nvptx_output_simt_exit): Declare.
19561 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19562 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19563 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19564 (nvptx_declare_function_name): ...here. Emit declaration of local
19565 memory space buffer for omp_simt_enter insn.
19566 (nvptx_output_unisimt_switch): New.
19567 (nvptx_output_softstack_switch): New.
19568 (nvptx_output_simt_enter): New.
19569 (nvptx_output_simt_exit): New.
19570 * config/nvptx/nvptx.h (struct machine_function): New fields
19571 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19572 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19573 (UNSPECV_SIMT_EXIT): Ditto.
19574 (omp_simt_enter_insn): New insn.
19575 (omp_simt_enter): New expansion.
19576 (omp_simt_exit): New insn.
19577 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19579 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19580 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19581 (expand_GOMP_SIMT_EXIT): New.
19582 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19583 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19584 (GOMP_SIMT_EXIT): Ditto.
19585 * target-insns.def (omp_simt_enter): New insn.
19586 (omp_simt_exit): Ditto.
19587 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19589 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19590 (lower_rec_input_clauses): Likewise.
19591 (lower_lastprivate_clauses): Handle SIMT privatization.
19593 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19594 (ompdevlow_adjust_simt_enter): New.
19595 (find_simtpriv_var_op): New.
19596 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19597 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19599 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19600 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19601 (copy_decl_for_dup_finish): Ditto.
19603 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19605 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19608 * config/i386/i386.c (ix86_option_override_internal): Always
19609 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19611 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19613 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19615 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19617 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19618 mark new edge's irreducible flag accordign to it.
19619 (vect_do_peeling): Check loop preheader edge's irreducible flag
19620 and pass it to function slpeel_add_loop_guard.
19622 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19624 PR tree-optimization/80218
19625 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19626 Update block frequencies and counts.
19628 2017-03-28 Richard Biener <rguenther@suse.de>
19630 PR tree-optimization/78644
19631 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19632 of a simplification result we may not use it at all.
19634 2017-03-28 Richard Biener <rguenther@suse.de>
19637 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19638 without arguments, generate default definition of a SSA name.
19640 2017-03-28 Richard Biener <rguenther@suse.de>
19642 PR middle-end/80222
19643 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19644 TYPE_REF_CAN_ALIAS_ALL references.
19645 * fold-const.c (fold_indirect_ref_1): Likewise.
19647 2017-03-28 Martin Liska <mliska@suse.cz>
19650 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19651 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19653 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19654 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19656 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19657 (EXTRA_SPECS): Define.
19658 (SUBTARGET_EXTRA_SPECS): Likewise.
19659 (SUBTARGET_CPP_SPEC): Likewise.
19660 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19661 SUBTARGET_EXTRA_SPECS.
19662 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19664 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19666 * config/arc/simdext.md (vst64_insn): Update pattern.
19667 (vld32wh_insn): Likewise.
19668 (vld32wl_insn): Likewise.
19669 (vld64_insn): Likewise.
19670 (vld32_insn): Likewise.
19672 2017-03-28 Marek Polacek <polacek@redhat.com>
19675 * fold-const.c (fold_comparison): Use protected_set_expr_location
19676 instead of SET_EXPR_LOCATION.
19678 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19680 * tree.c (add_expr): Avoid name lookup warning.
19682 2017-03-27 Jeff Law <law@redhat.com>
19684 PR tree-optimization/80216
19685 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19686 function name. Limit recursion depth.
19687 (record_temporary_equivalences): Corresponding changes.
19689 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19691 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19694 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19697 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19699 * cfgcleanup.c (reg_note_cfa_p): New array.
19700 (insns_have_identical_cfa_notes): New function.
19701 (old_insns_match_p): Don't cross-jump in between /f
19702 and non-/f instructions. If both i1 and i2 are frame related,
19703 verify all CFA notes, their order and content.
19705 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19708 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19709 HImode and SImode with zero extend to DImode to one insn.
19710 (bswap<mode>2_extenddi): Likewise.
19711 (bswapsi2_extenddi): Likewise.
19712 (bswaphi2_extendsi): Likewise.
19713 (bswaphi2): Combine bswap HImode and SImode into one insn.
19714 Separate memory insns from swapping register.
19715 (bswapsi2): Likewise.
19716 (bswap<mode>2): Likewise.
19717 (bswaphi2_internal): Delete, no longer used.
19718 (bswapsi2_internal): Likewise.
19719 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19720 store, and gpr<-gpr swap insns.
19721 (bswap<mode>2_store): Likewise.
19722 (bswaphi2_reg): Register only splitter, combine with the splitter.
19723 (bswaphi2 splitter): Likewise.
19724 (bswapsi2_reg): Likewise.
19725 (bswapsi2 splitter): Likewise.
19726 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19727 the insns into load, store, and register/register insns.
19728 (bswapdi2_ldbrx): Likewise.
19729 (bswapdi2_load): Likewise.
19730 (bswapdi2_store): Likewise.
19731 (bswapdi2_reg): Likewise.
19733 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19735 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19736 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19738 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19741 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19744 special handling for target option conflicts between dform
19745 options (-mpower9-dform, -mpower9-dform-vector,
19746 -mpower9-dform-scalar) and -mno-direct-move.
19748 2017-03-27 Richard Biener <rguenther@suse.de>
19750 PR tree-optimization/80181
19751 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19753 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19755 * config/arc/predicates.md (move_double_src_operand): Replace the
19756 call to move_double_src_operand with a call to address_operand.
19758 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19760 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19761 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19762 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19764 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19766 * config/arc/predicates.md (long_immediate_loadstore_operand):
19767 Consider scaled addresses cases.
19769 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19771 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19772 restored when in interrupt.
19773 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19774 doesn't have delay slot.
19776 2017-03-27 Richard Biener <rguenther@suse.de>
19779 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19780 inlined thunk clones.
19782 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19785 * asan.c (instrument_derefs): Copy over last operand from
19786 original COMPONENT_REF to the new COMPONENT_REF with
19787 DECL_BIT_FIELD_REPRESENTATIVE.
19788 * ubsan.c (instrument_object_size): Likewise.
19790 2017-03-27 Richard Biener <rguenther@suse.de>
19792 PR tree-optimization/80170
19793 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19794 sure DR/SCEV didnt fold in constants we do not see when looking
19795 at the reference base alignment.
19797 2017-03-27 Richard Biener <rguenther@suse.de>
19799 PR middle-end/80171
19800 * gimple-fold.c (fold_ctor_reference): Properly guard against
19801 NULL return value from canonicalize_constructor_val.
19803 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19806 * config/i386/i386.c (ix86_expand_builtin)
19807 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19808 flags reg setting and flags reg using instructions.
19809 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19810 clobbering instructions to zero extend op2.
19812 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19814 * doc/install.texi (Configuration) <--with-aix-soname>:
19815 Update link to AIX ld.
19817 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19819 PR rtl-optimization/80160
19820 PR rtl-optimization/80159
19821 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19822 reg_alternate_class into account.
19824 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19827 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19828 to consider in curr_insn_transform.
19830 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19832 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19833 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19834 and emit_mode_inner.
19836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19838 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19839 argument to the overloaded builtin variants. Use the new flag to
19840 deprecate certain builtin variants.
19841 * config/s390/s390-builtin-types.def: Add new builtin types.
19842 * config/s390/s390-builtins.h: Support new flags field for
19843 overloaded builtins.
19844 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19845 (s390_macro_to_expand): Enable vector float data type.
19846 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19847 builtins by incrementing the __VEC__ version number.
19848 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19850 (s390_resolve_overloaded_builtin): Emit error messages depending
19851 on the builtin flags.
19852 * config/s390/s390.c (s390_expand_builtin): Support additional
19853 flags argument. Change error message to match the messages
19854 emitted in s390-c.c.
19855 * config/s390/s390.md: New UNSPEC_* constants.
19856 (op_type): Add new instruction types.
19857 * config/s390/vecintrin.h: Add new builtins and test data class
19859 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19860 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19861 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19862 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19863 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19864 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19866 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19867 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19868 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19869 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19871 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19872 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19873 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19874 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19876 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19877 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19878 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19879 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19880 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19881 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19882 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19884 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19885 ("vec_scatter_element<V_HW_4:mode>_DI")
19886 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19887 ("vec_fpint<mode>", "vflls")
19888 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19889 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19890 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19891 ("*vec_cmphe<mode>_cc"): ... these.
19893 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19894 mode constant instead of magic value.
19896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19898 * config/s390/s390.c (s390_expand_vec_compare): Support other
19899 vector floating point modes than just V2DF.
19900 (s390_expand_vcond): Likewise.
19901 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19902 (s390_cannot_change_mode_class): Prevent mode changes between TF
19903 and V1TF in vector registers.
19904 * config/s390/s390.md (DF, SF): New mode attributes.
19905 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19906 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19907 SFmode support for VRs.
19908 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19910 (VFT, VF_HW): New mode iterators.
19911 (vw, sdx): New mode attributes.
19912 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19913 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19914 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19915 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19916 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19917 also the new vector floating point modes. Renaming to ...
19919 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19920 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19921 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19922 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19923 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19924 ("vec_unordered<mode>"): ... these.
19926 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19927 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19928 ("*vec_extendv2df"): New insn definitions.
19930 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19932 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19933 ("mulditi3_2", "*muldi3_sign"): New patterns.
19934 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19935 rename the pattern definition.
19937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19939 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19941 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19943 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19945 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19946 instruction if possible.
19947 * config/s390/vector.md (vec_halfnumelts): New mode
19949 ("*vec_vllezlf<mode>"): New pattern.
19951 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19953 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19954 ("popcountv4si2", "popcountv2di2"): Rename to ...
19955 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19956 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19958 ("popcount<mode>2_vxe"): New pattern.
19960 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19962 * common/config/s390/s390-common.c (processor_flags_table): Add
19964 * config.gcc: Add arch12.
19965 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19966 Default to arch12 for unknown CPU model numbers.
19967 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19968 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19969 PROCESSOR_max sanity check.
19970 * config/s390/s390-opts.h (enum processor_type): Add
19972 * config/s390/s390.c (processor_table): Add arch12.
19973 (s390_expand_builtin): Add check for B_VXE flag.
19974 (s390_issue_rate): Add PROCESSOR_ARCH12.
19975 (s390_get_sched_attrmask): Likewise.
19976 (s390_get_unit_mask): Likewise.
19977 (s390_sched_score): Enable z13 scheduling for arch12.
19978 (s390_sched_reorder): Likewise.
19979 (s390_sched_variable_issue): Likewise.
19980 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19982 (s390_tune_attr): Use z13 scheduling also for arch12.
19983 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19984 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19985 (TARGET_VXE_P): New macros.
19986 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19987 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19988 * config/s390/s390.opt: Add arch12 as processor_type.
19990 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19992 * config/s390/s390.md
19993 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19994 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19995 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19997 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19998 Rename expanders to ...
20000 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20001 ("fixuns_truncdddi2_emu"): ... these.
20003 ("fixuns_trunc<mode>si2_emu"): New expander.
20005 ("*fixuns_truncdfdi2_z13"): Rename to ...
20006 ("*fixuns_truncdfdi2_vx"): ... this.
20008 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20010 * config/s390/2964.md: Remove the single element vector compare
20011 instructions which are no longer used.
20012 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20014 (s390_canonicalize_comparison): Remove handling of DFmode
20016 (s390_expand_vec_compare_scalar): Remove function.
20017 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20018 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20020 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20022 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20024 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20026 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20027 will anyway by matched by mov<mode>_64dfp.
20029 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20031 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20032 vlef/vstef. Add missing operand to vleif.
20034 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20036 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20037 pair for all vector types with 64 bit elements.
20038 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20039 * config/s390/vector.md (V_HW_64): ... here.
20040 (V_128_NOSINGLE): New mode iterator.
20041 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20042 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20043 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20044 ("*vec_load_pairv2di"): Change to ...
20045 ("*vec_load_pair<mode>"): ... this one.
20047 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20049 * config/s390/constraints.md: Add comments.
20050 (jKK): Reject element sizes > 8 bytes.
20051 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20053 * config/s390/s390.md: Add the s_operand checks formerly in
20054 s390_split_ok_p to various splitters where they are still
20056 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20057 for 128 bit vectors. Plus two splitters.
20059 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20061 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20064 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20067 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20068 error if the boundary argument is not constant.
20070 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20072 PR rtl-optimization/80112
20073 * loop-doloop.c (doloop_condition_get): Don't check condition
20074 if cmp isn't SET with IF_THEN_ELSE src.
20076 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20078 PR tree-optimization/80158
20079 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20080 replacing a candidate statement, also replace it for the
20081 candidate's alternate interpretation.
20082 (replace_rhs_if_not_dup): Likewise.
20083 (replace_one_candidate): Likewise.
20085 2017-03-24 Richard Biener <rguenther@suse.de>
20087 PR tree-optimization/80167
20088 * graphite-isl-ast-to-gimple.c
20089 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20091 (translate_isl_ast_to_gimple::get_rename): Likewise.
20093 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20095 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20096 handling of certain combinations of target options, including the
20097 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20098 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20100 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20103 * config/arm/arm.md (*load_multiple): Add reload_completed to
20104 matching condition.
20106 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20107 Richard Biener <rguenth@suse.de>
20109 PR tree-optimization/79908
20110 PR tree-optimization/80136
20111 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20112 been cast away, gimplify_and_add suffices.
20114 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20116 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20118 2017-03-23 Richard Biener <rguenther@suse.de>
20120 PR tree-optimization/80032
20121 * gimplify.c (gimple_push_cleanup): Forced unconditional
20122 cleanups still have to go to the conditional_cleanups
20125 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20127 PR tree-optimization/80072
20128 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20130 (next_operand_entry_id): Change type to unsigned int.
20131 (sort_by_operand_rank): Make sure to return the right return value
20132 even if unsigned fields are bigger than INT_MAX.
20133 (struct oecount): Change cnt and id type to unsigned int.
20134 (oecount_hasher::equal): Formatting fix.
20135 (oecount_cmp): Make sure to return the right return value
20136 even if unsigned fields are bigger than INT_MAX.
20137 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20140 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20141 TREE_READONLY on result if writing it more than once.
20144 * doc/invoke.texi (-fsanitize=thread): Document that with
20145 -fnon-call-exceptions atomics are not able to throw
20149 * tsan.c: Include tree-eh.h.
20150 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20151 maybe_clean_or_replace_eh_stmt where needed.
20152 (instrument_memory_accesses): Add cfg_changed argument.
20153 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20154 if it returned true.
20155 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20157 PR rtl-optimization/63191
20158 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20159 wrapper function, moved the whole old content into ...
20160 (ix86_delegitimize_address_1): ... this. New inline function.
20161 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20162 true as last argument instead of ix86_delegitimize_address.
20164 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20166 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20167 cortexa57_branch_cost.
20169 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20171 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20173 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20176 * doc/md.texi (Constraints): Document wA constraint.
20177 * config/rs6000/constraints.md (wA): New.
20178 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20179 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20180 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20181 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20183 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20186 * gimplify.c (is_oacc_declared): New function.
20187 (oacc_default_clause): Use it to set default flags for acc declared
20188 variables inside parallel regions.
20189 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20190 declared variables.
20191 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20192 declare attribute to any decl as necessary.
20194 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20197 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20198 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20199 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20200 (arm_arch_lpae): This.
20201 * config/arm/arm.c (arm_arch7ve): Rename into ...
20202 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20203 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20206 2017-03-22 Martin Liska <mliska@suse.cz>
20209 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20210 error message instead of an ICE.
20212 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20214 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20216 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20218 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20220 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20223 2017-03-21 Martin Sebor <msebor@redhat.com>
20225 * doc/extend.texi: Use "cannot" instead of "can't."
20226 * doc/hostconfig.texi: Same.
20227 * doc/install.texi: Same.
20228 * doc/invoke.texi: Same.
20229 * doc/loop.texi: Same.
20230 * doc/md.texi: Same.
20231 * doc/objc.texi: Same.
20232 * doc/rtl.texi: Same.
20233 * doc/tm.texi: Same.
20234 * doc/tm.texi.in: Same.
20235 * doc/trouble.texi: Same.
20237 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20240 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20241 (collect_checksum_attributes): Set it.
20242 (die_checksum_ordered): Use it.
20244 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20246 PR tree-optimization/79908
20247 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20248 change: For a VA_ARG whose LHS has been cast away, use
20249 force_gimple_operand to construct the side effects.
20251 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20253 PR translation/80001
20254 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20255 more amenable to translation.
20256 (oacc_loop_auto_partitions): Likewise.
20258 2017-03-21 Marek Polacek <polacek@redhat.com>
20259 Martin Sebor <msebor@redhat.com>
20261 PR tree-optimization/80109
20262 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20263 on INTEGRAL_TYPE_P.
20265 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20266 Segher Boessenkool <segher@kernel.crashing.org>
20269 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20270 check reg_used_between_p between insn and one of succ or succ2
20271 depending on if succ is artificial insn not inserted into insn
20274 2017-03-21 Martin Liska <mliska@suse.cz>
20276 PR gcov-profile/80081
20277 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20278 * doc/gcc.texi: Include gcov-dump stuff.
20279 * doc/gcov-dump.texi: New file.
20281 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20283 PR rtl-optimization/79150
20284 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20285 conditional jump, if the jump is the last insn of the loop.
20287 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20288 Richard Biener <rguenth@suse.de>
20290 PR tree-optimization/79908
20291 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20292 been cast away, use force_gimple_operand to construct the side
20295 2017-03-21 Martin Liska <mliska@suse.cz>
20297 PR libfortran/79956
20298 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20301 2017-03-21 Brad Spengler <spender@grsecurity.net>
20304 * plugin.c (htab_hash_plugin): New function.
20305 (add_new_plugin): Use it and adjust.
20306 (parse_plugin_arg_opt): Adjust.
20307 (init_one_plugin): Likewise.
20309 2017-03-21 Richard Biener <rguenther@suse.de>
20311 PR tree-optimization/80032
20312 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20313 if set force the cleanup to happen unconditionally.
20314 (gimplify_target_expr): Push inserted clobbers with force_uncond
20315 to avoid them being removed by control-dependent DCE.
20317 2017-03-21 Richard Biener <rguenther@suse.de>
20319 PR tree-optimization/80122
20320 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20321 va_arg_pack_len when the inlined call stmt requires pack
20323 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20325 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20328 * tsan.c (instrument_builtin_call): If the memory model argument
20329 is not a constant, assume it is valid.
20332 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20335 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20337 PR rtl-optimization/79910
20338 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20339 if its dest is used by an insn before I2 (other than the combined
20340 insns themselves, which are properly handled already).
20342 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20345 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20347 * combine.c (record_used_regs): New static function.
20348 (try_combine): Handle situations where there is an additional
20349 instruction between I2 and I3 which needs to have a LOG_LINK
20353 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20355 * combine.c (try_combine): Delete redundant i1 test. Call
20356 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20358 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20361 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20362 alternatives 13/14.
20364 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20366 PR tree-optimization/80054
20367 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20368 the optimization if a PHI or any of its arguments is not dominated
20369 by the candidate's basis. Use gphi* rather than gimple* as
20371 (replace_profitable_candidates): Clean up a gimple* variable that
20372 should be a gphi* variable.
20374 2017-03-20 Martin Sebor <msebor@redhat.com>
20377 * doc/extend.texi (attribute constructor): Document present limitation.
20379 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20382 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20383 __POWER9_VECTOR__ #ifdef control, change template definition to
20384 use Power9-specific built-in function.
20385 (vec_any_eq): Likewise.
20386 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20387 to control outcomes from this test.
20388 (vector_ae_<mode>p): For VEC_F modes, likewise.
20390 2017-03-20 Ian Lance Taylor <iant@google.com>
20392 * config/i386/i386.c (ix86_function_regparm): Save an extra
20393 register for -fsplit-stack with DECL_STATIC_CHAIN.
20395 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20398 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20399 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20401 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20403 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20405 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20408 2017-03-20 Marek Polacek <polacek@redhat.com>
20411 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20413 2017-03-20 Richard Biener <rguenther@suse.de>
20415 PR tree-optimization/80113
20416 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20417 allocate extra SSA name for PHI def.
20418 (add_close_phis_to_outer_loops): Likewise.
20419 (add_close_phis_to_merge_points): Likewise.
20420 (copy_loop_close_phi_args): Likewise.
20421 (copy_cond_phi_nodes): Likewise.
20423 2017-03-20 Martin Liska <mliska@suse.cz>
20425 PR middle-end/79753
20426 * tree-chkp.c (chkp_build_returned_bound): Do not build
20427 returned bounds for a LHS that's not a BOUNDED_P type.
20429 2017-03-20 Martin Liska <mliska@suse.cz>
20433 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20434 COMPLEX_CST and VECTOR_CST.
20436 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20439 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20440 target operand. A new splitter adds the clobber statement in case
20441 the target operand is dead anyway.
20443 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20445 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20446 to age-old versions of binutils and glibc.
20448 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20450 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20452 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20454 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20456 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20458 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20459 requirement for binutils 2.13.
20461 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20463 * combine.c (try_combine): Delete redundant i1 test. Call
20464 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20466 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20468 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20469 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20471 <riscv64-*-elf>: Re-arrange section
20472 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20473 <riscv32-*-linux>: Likewise.
20474 <riscv64-*-elf>: Likewise
20475 <riscv64-*-linux>: Likewise.
20477 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20480 * aarch64.opt(verbose-cost-dump): Fix typo.
20482 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20485 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20486 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20488 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20490 * reload.c (find_reloads): When reloading a nonoffsettable address,
20491 use RELOAD_OTHER for it and its address reloads.
20493 PR rtl-optimization/79910
20494 * combine.c (record_used_regs): New static function.
20495 (try_combine): Handle situations where there is an additional
20496 instruction between I2 and I3 which needs to have a LOG_LINK
20499 2017-03-17 Jeff Law <law@redhat.com>
20501 PR tree-optimization/71437
20502 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20503 conditional in the hash table first.
20504 (vrp_dom_walker::before_dom_children): Extract condition from
20505 ASSERT_EXPR. Record condition, its inverion and any implied
20506 conditions as well.
20508 2017-03-17 Marek Polacek <polacek@redhat.com>
20509 Markus Trippelsdorf <markus@trippelsdorf.de>
20511 PR tree-optimization/80079
20512 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20515 2017-03-17 Richard Biener <rguenther@suse.de>
20517 PR middle-end/80075
20518 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20519 Properly verify the LHS before the RHS possibly claims to be
20521 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20524 2017-03-17 Martin Jambor <mjambor@suse.cz>
20526 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20527 (List of -O2 options): Likewise.
20528 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20531 2017-03-17 Tom de Vries <tom@codesourcery.com>
20533 * gcov-dump.c (print_usage): Print bug_report_url.
20535 2017-03-17 Richard Biener <rguenther@suse.de>
20537 PR middle-end/80050
20538 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20539 (parser::peek): Likewise.
20541 2017-03-17 Richard Biener <rguenther@suse.de>
20543 PR tree-optimization/80048
20544 * sese.c (free_sese_info): Properly release rename_map and
20545 copied_bb_map elements.
20547 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20549 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20550 Add linked-list forward and backlinks. Insert on
20551 construction, remove on destruction.
20552 (class pass_store_merging): Add m_stores_head field.
20553 (pass_store_merging::terminate_and_process_all_chains):
20554 Iterate over m_stores_head list.
20555 (pass_store_merging::terminate_all_aliasing_chains):
20557 (pass_store_merging::execute): Check for debug stmts first.
20558 Push new chains onto the m_stores_head stack.
20560 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20563 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20564 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20565 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20567 2017-03-16 Jeff Law <law@redhat.com>
20569 PR tree-optimization/71437
20570 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20571 member function. Implementation moved into after_dom_children
20572 member function and into the threader's thread_outgoing_edges
20574 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20575 some code into new thread_outgoing_edges.
20576 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20577 definition. Simplify marker handling (do it here). Assume we always
20578 have the available expression and the const/copies tables.
20579 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20581 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20582 * tree-vrp.c (equiv_stack): No longer file scoped.
20583 (vrp_dom_walker): New class.
20584 (vrp_dom_walker::before_dom_children): New member function.
20585 (vrp_dom_walker::after_dom_children): Likewise.
20586 (identify_jump_threads): Setup domwalker. Use it rather than
20587 walking edges in a random order by hand. Simplify setup/finalization.
20588 (finalize_jump_threads): Remove.
20589 (vrp_finalize): Do not call identify_jump_threads here.
20590 (execute_vrp): Do it here instead and call thread_through_all_blocks
20593 PR tree-optimization/71437
20594 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20596 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20598 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20599 (dom_opt_dom_walker::thread_across_edge): Remove
20600 handle_dominating_asserts argument. All callers changed.
20601 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20602 changes. Remove calls to lhs_of_dominating_assert. Other
20603 uses of handle_dominating_asserts turn into unconditional code
20604 (simplify_control_stmt_condition_1): Likewise.
20605 (simplify_control_stmt_condition): Likewise.
20606 (thread_through_normal_block, thread_across_edge): Likewise.
20607 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20608 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20609 object if it is not an SSA_NAME.
20610 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20611 before calling into the VRP specific simplifiers.
20612 (identify_jump_threads): Remove handle_dominating_asserts
20615 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20618 * tree-diagnostic.c (default_tree_printer): No longer static.
20619 * tree-diagnostic.h (default_tree_printer): New prototype.
20621 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20623 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20624 Change ins into fmov.
20626 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20628 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20629 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20630 Use h_con constraint for operand 1.
20631 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20632 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20634 2017-03-15 Jeff Law <law@redhat.com>
20636 PR tree-optimization/71437
20637 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20638 (record_temporary_equivalences): Use it.
20640 PR tree-optimization/71437
20641 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20642 tree-ssa-scopedtables.
20643 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20644 (record_conditions, record_cond, vuse_eq): Likewise.
20645 (record_edge_info): Adjust to API tweak of record_conditions.
20646 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20647 (record_temporary_equivalences, optimize_stmt): Likewise.
20648 (eliminate_redundant_computations): Likewise.
20649 (record_equivalences_from_stmt): Likewise.
20650 * tree-ssa-scopedtables.c: Include options.h and params.h.
20651 (vuse_eq): New function, moved from tree-ssa-dom.c
20652 (build_and_record_new_cond): Likewise.
20653 (record_conditions): Likewise. Accept vector of conditions rather
20654 than edge_equivalence structure for first argument.
20655 for the first argument.
20656 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20657 from tree-ssa-dom.c.
20658 (avail_exprs_stack::record_cond): Likewise.
20659 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20660 from tree-ssa-dom.c.
20661 (avail_exprs_stack): Add new member functions lookup_avail_expr
20663 (record_conditions): Declare.
20665 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20668 * lra-constraints.c (process_alt_operands): Increase reject for
20669 reloading an input/output operand.
20671 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20674 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20675 insns to convert from signed/unsigned char/short to IEEE 128-bit
20677 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20679 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20682 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20683 subreg of inner mode for values already in registers.
20685 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20687 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20688 iteration reg is used after the loop.
20690 2017-03-14 Martin Sebor <msebor@redhat.com>
20692 PR tree-optimization/79800
20693 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20694 precision in negative-positive range.
20695 (format_floating): Call non-const overload with adjusted precision.
20697 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20700 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20703 2017-03-14 Martin Sebor <msebor@redhat.com>
20705 PR middle-end/80020
20706 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20707 * builtins.def (aligned_alloc): Use it.
20710 * Makefile.in (GTFILES): Add calls.c.
20711 * calls.c: Include "gt-calls.h".
20713 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20715 PR rtl-optimization/79728
20716 * regs.h (struct target_regs): New field
20717 x_contains_allocatable_regs_of_mode.
20718 (contains_allocatable_regs_of_mode): New macro.
20719 * reginfo.c (init_reg_sets_1): Initialize it, and change
20720 contains_reg_of_mode so it includes global regs as well.
20721 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20722 rather than contains_regs_of_mode.
20724 2017-03-14 Martin Liska <mliska@suse.cz>
20726 * doc/invoke.texi: Document options that can't be combined with
20727 -fcheck-pointer-bounds.
20729 2017-03-14 Martin Liska <mliska@suse.cz>
20731 PR middle-end/79831
20732 * doc/invoke.texi (-Wchkp): Document the option.
20734 2017-03-14 Martin Liska <mliska@suse.cz>
20736 * Makefile.in: Install gcov-dump.
20738 2017-03-14 Martin Liska <mliska@suse.cz>
20740 * multiple_target.c (expand_target_clones): Bail out for
20741 an invalid attribute.
20743 2017-03-14 Richard Biener <rguenther@suse.de>
20745 * alias.c (struct alias_set_entry): Pack properly.
20746 * cfgloop.h (struct loop): Likewise.
20747 * cse.c (struct set): Likewise.
20748 * ipa-utils.c (struct searchc_env): Likewise.
20749 * loop-invariant.c (struct invariant): Likewise.
20750 * lra-remat.c (struct cand): Likewise.
20751 * recog.c (struct change_t): Likewise.
20752 * rtl.h (struct address_info): Likewise.
20753 * symbol-summary.h (function_summary): Likewise.
20754 * tree-loop-distribution.c (struct partition): Likewise.
20755 * tree-object-size.c (struct object_size_info): Likewise.
20756 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20757 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20758 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20759 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20760 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20761 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20762 (struct _stmt_vec_info): Likewise.
20764 2017-03-14 Martin Liska <mliska@suse.cz>
20767 * multiple_target.c (create_dispatcher_calls): Check that
20768 a target can create a function dispatcher.
20770 2017-03-14 Martin Liska <mliska@suse.cz>
20773 * multiple_target.c (expand_target_clones): Drop local.local
20774 flag for default implementation.
20776 2017-03-14 Richard Biener <rguenther@suse.de>
20778 PR tree-optimization/80030
20779 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20781 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20783 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20784 gcc_fallthrough() instead of __attribute__((fallthrough));
20786 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20788 * doc/gcc.texi: Remove "up" link to (DIR).
20789 * doc/gccint.texi: Ditto.
20791 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20793 * doc/install.texi (Specific) <avr>: Remove reference to
20796 2017-03-13 Jeff Law <law@redhat.com>
20798 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20799 attribute rather than comments.
20801 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20802 match_scratch operand is highest.
20804 2017-03-13 Martin Liska <mliska@suse.cz>
20806 PR middle-end/78339
20807 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20808 is a CHKP clone, use original declaration.
20810 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20812 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20813 (arc_conditional_register_usage): Use a different allocation order
20814 when optimizing for size.
20815 * common/config/arc/arc-common.c (arc_option_optimization_table):
20816 Section anchors default on when optimizing for size.
20818 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20820 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20822 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20824 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20825 * config/arc/arc.md (cpu_facility): Add cd variant.
20826 (*movqi_insn): Add code density variant.
20827 (*movhi_insn): Likewise.
20828 (*movqi_insn): Likewise.
20829 (*addsi3_mixed): Likewise.
20830 (subsi3_insn): Likewise.
20832 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20834 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20836 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20838 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20839 expressions with MINUS and UNARY ops.
20841 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20844 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20846 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20847 between vec_select and vector argument.
20848 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20849 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20850 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20851 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20852 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20853 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20855 2017-03-13 Richard Biener <rguenther@suse.de>
20858 * params.def (vect-max-peeling-for-alignment): Fix typo.
20860 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20862 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20863 issue that only occurred with binutils below 2.18.
20865 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20867 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20868 refer to binutils 2.11/2.12 minimum.
20870 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20872 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20873 ftp.kernel.org and simplify binutils requirement.
20875 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20877 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20879 (Optimize Options): Ditto. Also remove redundancy.
20881 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20883 PR translation/79848
20884 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20886 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20887 to G_ to avoid double translation.
20889 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20891 PR translation/79923
20892 * auto-profile.c (get_combined_location): Convert leading
20893 character of diagnostics to lower case and remove trailing period.
20894 (read_profile): Likewise for various diagnostics.
20895 * config/arm/arm.c (arm_option_override): Remove trailing period
20896 from various diagnostics.
20897 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20898 (msp430_expand_delay_cycles): Likewise.
20900 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20903 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20904 full command-line argument, rather than just "str".
20905 (aarch64_validate_march): Likewise.
20906 (aarch64_validate_mtune): Likewise.
20908 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20910 PR rtl-optimization/78911
20911 * lra-assigns.c (must_not_spill_p): New function.
20912 (spill_for): Use it.
20914 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20916 PR tree-optimization/79981
20917 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20918 ATOMIC_COMPARE_EXCHANGE ifn result.
20919 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20920 IFN_ATOMIC_COMPARE_EXCHANGE.
20922 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20925 * opts.c (parse_sanitizer_options): Add missing question mark to
20926 "did you mean" message.
20928 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20930 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20932 (VMULEUH_UNS): Likewise.
20933 (VMULOUB_UNS): Likewise.
20934 (VMULOUH_UNS): Likewise.
20935 * config/rs6000/rs6000.c (builtin_function_type): Remove
20936 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20938 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20941 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20942 x with result of extra_parsing_for_operand_code_0.
20943 (function_reader::extra_parsing_for_operand_code_0): Convert
20944 return type from void to rtx, returning x. When reading
20945 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20946 larger size containing struct block_symbol.
20948 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20950 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20951 -mfloat128-hardware without -m64.
20953 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20956 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20957 entries to the case statement that marks unsigned arguments to
20958 overloaded functions.
20960 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20962 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20963 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20965 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20968 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20969 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20971 2017-03-10 Martin Liska <mliska@suse.cz>
20975 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20976 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20979 2017-03-10 Olivier Hainque <hainque@adacore.com>
20981 * tree-switch-conversion (array_value_type): Start by resetting
20982 candidate type to it's main variant.
20984 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20986 PR rtl-optimization/79909
20987 * combine.c (try_combine): Use simplify_replace_rtx on individual
20988 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20989 of the whole CALL_INSN_FUNCTION_USAGE.
20991 PR tree-optimization/79972
20992 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20993 get_range_info on SSA_NAMEs. Formatting fixes.
20995 2017-03-10 Richard Biener <rguenther@suse.de>
20996 Jakub Jelinek <jakub@redhat.com>
20998 PR tree-optimization/77975
20999 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21000 edge to be constant.
21001 (get_val_for): For constant x return it. Formatting fix.
21002 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21003 would use the same bases as the current one.
21005 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21007 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21008 instead of vec_select for V1TImode.
21009 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21011 (VSX_LE_128): Add V1TI to this mode iterator.
21012 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21013 (*vsx_le_perm_store_<mode>): Likewise.
21014 (pre-reload splitter for VSX stores): Likewise.
21015 (post-reload splitter for VSX stores): Likewise.
21016 (*vsx_xxpermdi2_le_<mode>): Likewise.
21017 (*vsx_lxvd2x2_le_<mode>): Likewise.
21018 (*vsx_stxvd2x2_le_<mode>): Likewise.
21020 2017-03-09 Michael Eager <eager@eagercon.com>
21022 Correct failures with --enable-checking=yes,rtl.
21024 * config/microblaze/microblaze.c (microblaze_expand_shift):
21025 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21026 test for const0_rtx.
21027 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21028 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21030 2017-03-09 Richard Biener <rguenther@suse.de>
21032 PR tree-optimization/79977
21033 * graphite-scop-detection.c (scop_detection::merge_sese):
21034 Handle the case of extra exits to blocks dominating the entry.
21036 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21038 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21041 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21043 PR rtl-optimization/79949
21044 * lra-constraints.c (process_alt_operands): Check memory when
21045 trying to predict a cycle. Print about the overall increase.
21047 2017-03-09 Richard Biener <rguenther@suse.de>
21049 PR middle-end/79971
21050 * gimple-expr.c (useless_type_conversion_p): Preserve
21051 TYPE_SATURATING for fixed-point types.
21053 2017-03-09 Richard Biener <rguenther@suse.de>
21056 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21057 alignment of BLKmode params.
21059 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21062 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21063 (VALL_NO_V2Q): Likewise.
21065 * config/aarch64/aarch64-simd.md
21066 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21068 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21069 VALL_NO_V2Q mode iterator.
21070 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21072 2017-03-09 Martin Liska <mliska@suse.cz>
21074 PR tree-optimization/79631
21075 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21076 tree_int_cst_sign_bit just for INTEGER constants.
21078 2017-03-09 Martin Liska <mliska@suse.cz>
21082 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21085 2017-03-09 Marek Polacek <polacek@redhat.com>
21088 * tree.c (inchash::add_expr): Handle TREE_VEC.
21090 2017-03-09 Martin Liska <mliska@suse.cz>
21093 (chkp_narrow_size_and_offset): New function.
21094 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21095 (void chkp_parse_bit_field_ref): New function.
21096 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21097 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21099 2017-03-09 Martin Liska <mliska@suse.cz>
21102 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21103 (chkp_find_bounds_1): Remove gcc_unreachable.
21105 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21108 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21109 BUILT_IN_SYNC*, determine the access type from the size suffix and
21110 always build a MEM_REF with that type. Handle forgotten
21111 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21114 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21115 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21116 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21117 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21118 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21119 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21120 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21121 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21122 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21123 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21124 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21125 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21126 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21127 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21128 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21129 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21130 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21131 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21132 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21133 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21134 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21135 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21136 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21137 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21138 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21139 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21140 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21141 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21142 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21143 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21144 definitions outside of __OPTIMIZE__ guarded section.
21147 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21148 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21149 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21150 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21153 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21155 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21156 ("vfenez<mode>"): Add missing constraints.
21158 2017-03-08 Martin Sebor <msebor@redhat.com>
21161 * config/nds32/nds32.c (nds32_option_override):
21162 Fix misspelled diagnostic.
21164 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21167 * gimplify.c (gimplify_omp_for): Replace index var in outer
21168 taskloop statement with an artificial variable and add
21169 OMP_CLAUSE_PRIVATE clause for it.
21171 2017-03-08 Richard Biener <rguenther@suse.de>
21173 PR tree-optimization/79955
21174 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21175 for accesses that are completely outside of the variable.
21177 2017-03-08 Andrew Haley <aph@redhat.com>
21179 PR tree-optimization/79943
21180 * tree-ssa-loop-split.c (compute_new_first_bound): When
21181 calculating the new upper bound, (END-BEG) should be added, not
21184 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21186 * config/avr/avr.md (setmemhi): Make sure match_dup
21187 operand number comes before match_scratch.
21189 2017-03-08 Richard Biener <rguenther@suse.de>
21191 PR tree-optimization/79920
21192 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21193 with ncopies == 1 to ...
21194 (vect_transform_slp_perm_load): ... here. Properly compute
21195 all element loads by iterating VF times over the group. Do
21196 not handle ncopies (computed in a broken way) in
21197 vect_create_mask_and_perm.
21199 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21202 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21203 is a uniform vector, use uniform_vector_p return value instead of
21204 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21206 2017-03-07 Marek Polacek <polacek@redhat.com>
21208 PR middle-end/79809
21209 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21210 (alloca_call_type): Likewise.
21212 2017-03-07 Martin Liska <mliska@suse.cz>
21214 * gcov.c (process_args): Put comment to correct location.
21216 2017-03-07 Martin Liska <mliska@suse.cz>
21218 PR middle-end/68270
21219 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21220 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21221 (chkp_narrow_bounds_for_field): Likewise.
21222 (chkp_parse_array_and_component_ref): Pass one more argument to
21225 2017-03-07 Richard Biener <rguenther@suse.de>
21227 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21230 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21232 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21233 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21235 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21238 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21239 to end of description.
21240 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21242 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21244 PR rtl-optimization/79901
21245 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21247 (*avx512f_<code><mode>3<mask_name>): ... this.
21248 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21249 iterator instead of VI8_AVX2_AVX512BW.
21251 PR rtl-optimization/79901
21252 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21253 min/max expander, expand it using expand_vec_cond_expr.
21256 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21259 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21262 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21263 to void * for PCH reasons.
21264 * dwarf2out.c (output_loc_operands, output_die): Cast
21265 v.val_vec.array to unsigned char *.
21267 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21270 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21273 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21275 PR rtl-optimization/79571
21276 * lra-constraints.c (process_alt_operands): Calculate static
21277 reject and subtract it from overall when only addresses will be
21280 2017-03-06 Julia Koval <julia.koval@intel.com>
21283 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21284 incoming stack boundary to 128 for 64-bit targets.
21286 2017-03-06 Richard Biener <rguenther@suse.de>
21288 PR tree-optimization/79894
21289 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21290 to NULL after folding it.
21292 2017-03-06 Richard Biener <rguenther@suse.de>
21294 PR tree-optimization/79824
21295 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21296 check disabling peeling for gaps.
21298 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21300 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21301 attributes): Document gettimeofday.
21303 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21305 * config/s390/s390.c (s390_option_override_internal): Set
21306 PARAM_MIN_VECT_LOOP_BOUND
21308 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21310 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21311 * config/s390/s390.md: Likewise.
21313 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21316 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21317 (<avx2_avx512>_perm<mode>): Rename to ...
21318 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21320 (<avx512>_perm<mode>_mask): Rename to ...
21321 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21323 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21324 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21325 instead of VI8F_256_512.
21326 (avx512f_perm<mode>): New define_expand.
21327 (avx512f_perm<mode>_mask): Likewise.
21328 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21329 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21331 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21333 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21334 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21336 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21338 2017-03-06 Martin Liska <mliska@suse.cz>
21341 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21342 when having a SSA NAME w/o VAR_DECL assigned to it.
21344 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21346 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21347 msa_dpsub_<su>_d): Fix MODE for vec_select.
21349 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21351 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21353 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21355 2017-03-06 Richard Biener <rguenther@suse.de>
21357 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21358 * plugin.c (register_plugin_info): Likewise.
21359 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21361 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21363 * config/i386/sse.md (sse_storehps, sse_storelps,
21364 avx_<castmode><avxsizesuffix>_<castmode>,
21365 avx512f_<castmode><avxsizesuffix>_<castmode>,
21366 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21367 in condition that at least one operand is not a MEM.
21369 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21371 PR middle-end/79805
21372 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21373 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21375 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21376 gimple_call_nothrow_p flag based on whether original builtin can throw.
21377 If it can, emit following stmts on the fallthrough edge.
21378 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21379 don't create new bb if inserting just debug stmts on the edge, try to
21380 insert them on the fallthru bb or just reset debug stmts.
21382 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21385 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21386 restore recog_data (including the operand rtxes inside it) around
21387 the call to get_insn_template.
21389 2017-03-03 Martin Sebor <msebor@redhat.com>
21391 PR tree-optimization/79699
21392 * context.c (context::~context): Free MPFR caches to avoid
21393 a memory leak on program exit.
21395 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21397 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21398 Use wide_int::ulow () instead of .elt (0).
21400 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21402 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21403 (*pushxf): Limit oF constraint to 32bit targets and add oC
21404 constraint for 64bit targets.
21405 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21406 (*pushdf): Change rmF constraint to rmC.
21408 2017-03-03 Martin Liska <mliska@suse.cz>
21410 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21411 Remove unused variable.
21413 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21416 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21417 is a memory operand, increase num_memory.
21418 (ix86_expand_args_builtin): Likewise.
21420 2017-03-03 Jan Hubicka <jh@suse.cz>
21423 * ipa-devirt.c (maybe_record_node): Properly handle
21424 __cxa_pure_virtual visibility.
21426 2017-03-03 Martin Liska <mliska@suse.cz>
21428 PR tree-optimization/79803
21429 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21431 (pass_loop_prefetch::execute): Disabled optimization if an
21432 assumption about L1 cache size is not met.
21434 2017-03-03 Martin Liska <mliska@suse.cz>
21436 PR rtl-optimization/79574
21437 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21438 (hash_scan_set): Likewise.
21439 (dump_hash_table): Likewise.
21440 (hoist_code): Likewise.
21442 2017-03-03 Richard Biener <rguenther@suse.de>
21444 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21445 in place of uhigh ().
21446 (fixed_convert_from_real): Likewise.
21448 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21451 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21453 2017-03-03 Richard Biener <rguenther@suse.de>
21455 PR middle-end/79818
21456 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21457 TYPE_OVERFLOW_UNDEFINED check.
21459 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21461 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21463 (vector_ae_<mode>_p): Likewise.
21464 (vector_nez_<mode>_p): Likewise.
21465 (vector_ne_v2di_p): Likewise.
21466 (vector_ae_v2di_p): Likewise.
21467 (vector_ne_<mode>_p): Likewise.
21468 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21470 (vsx_tsqrt<mode>2_fe): Likewise.
21472 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21475 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21477 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21479 PR rtl-optimization/79780
21480 * cprop.c (one_cprop_pass): When second and further conditional trap
21481 in a single basic block is turned into an unconditional trap, turn it
21482 into a deleted note to avoid RTL verification failures.
21484 2017-03-02 Richard Biener <rguenther@suse.de>
21486 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21488 2017-03-02 Richard Biener <rguenther@suse.de>
21490 PR tree-optimization/79345
21492 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21493 param and abort the walk, returning -1 if it is hit.
21494 (walk_aliased_vdefs): Take a limit param and pass it on.
21495 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21496 defaulting to 0 and return a signed int.
21497 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21498 (check_defs): New helper.
21499 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21500 about uninitialized memory.
21501 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21502 bogus uninitialized warning.
21503 (fixed_convert_from_real): Likewise.
21505 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21507 PR tree-optimization/66768
21508 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21509 iv_use if base object can't be determined.
21511 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21513 PR tree-optimization/79345
21514 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21515 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21516 (get_pattern_stats): Initialize it.
21517 * genemit.c (gen_expand): Verify match_scratch numbers come after
21518 match_operand/match_dup numbers.
21519 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21520 match_scratch numbers.
21521 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21523 * config/s390/s390.md (trunctdsd2): Likewise.
21525 2017-03-02 Richard Biener <rguenther@suse.de>
21527 * wide-int.h (wide_int_storage::operator=): Implement in terms
21530 2017-03-02 Richard Biener <rguenther@suse.de>
21532 PR tree-optimization/79777
21533 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21534 the to insert expression to sth existing.
21536 2017-03-01 Martin Sebor <msebor@redhat.com>
21538 PR middle-end/79692
21539 * gimple-ssa-sprintf.c
21540 (directive::known_width_and_precision): New function.
21541 (format_integer): Use it.
21542 (get_mpfr_format_length): Consider the full range of precision
21543 when computing %g output with the # flag. Set the likely byte
21544 count to 3 rather than 1 when precision is indeterminate.
21545 (format_floating): Correct the lower bound of precision.
21547 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21549 * doc/invoke.texi: Document default code model for 64-bit Linux.
21551 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21554 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21555 udiv rather than div since input pattern is unsigned.
21557 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21559 * config/i386/i386.c (print_reg): Warn for values of
21560 unsupported size in integer register.
21562 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21565 * config/rs6000/predicates.md (current_file_function_operand): Do
21566 not allow self calls to be local if the function is replaceable.
21568 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21571 * config/rs6000/altivec.h (vec_ctz and others): Change the
21572 preprocessor macro that controls conditional compilation from
21573 _ARCH_PWR9 to __POWER9_VECTOR__.
21574 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21575 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21576 control (instead of _ARCH_PWR9 control) so that template
21577 definition uses power9-specific function.
21578 (vec_any_eq): Likewise.
21579 (vec_all_ne): Change macro definition to use a power9-specific
21580 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21581 _ARCH_PWR9 control).
21582 (vec_any_eq) Likewise.
21583 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21584 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21585 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21586 support for xvcmpnedp instruction.
21587 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21588 macro expansion so that Power9 implementation of vec_all_ne does
21589 not use the AltiVec predicate framework.
21590 (VCMPNEH_P): Likewise.
21591 (VCMPNEW_P): Likewise.
21592 (VCMPNED_P): Likewise.
21593 (VCMPNEFP_P): Likewise.
21594 (VCMPNEDP_P): Likewise.
21595 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21596 implementation of vec_any_eq to not use AltiVec predicate
21598 (VCMPAEH_P): Likewise.
21599 (VCMPAEW_P): Likewise.
21600 (VCMPAED_P): Likewise.
21601 (VCMPAEFP_P): Likewise.
21602 (VCMPAEDP_P): Likewise.
21603 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21604 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21605 not use the AltiVec predicate framework.
21606 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21607 of vec_any_eq to not use AltiVec predicate framework.
21608 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21609 support for predefined __POWER9_VECTOR__ macro to indicate that
21610 Power9 instruction selection is enabled.
21611 (altivec_overloaded_builtins): Remove extraneous
21612 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21613 function argument types RS6000_BTI_bool_V16QI and
21614 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21615 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21616 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21617 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21618 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21619 Power9 for implementations of vec_cmpne. Change the signature for
21620 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21621 (representing vec_all_ne) to remove the previously described first
21622 argument of type RS6000_BTI_INTSI, as this was an artifact of
21623 reliance on the AltiVec predicate framework, which is no longer
21624 used in the implementation of these functions. Add
21625 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21626 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21627 since, unlike the AltiVec predicate framework implementation, we
21628 do not share function descriptors between vec_alle and vec_anyeq.
21629 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21630 set of modes that receive special treatment even when
21631 TARGET_P9_VECTOR is true. The special treatment emits code that
21632 does not depend on Power9 instructions.
21633 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21634 define_expand to not rely on AltiVec predicate framework.
21635 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21637 (vector_ne_v2di_p): Change this define_expand to not rely on
21638 AltiVec predicate framework.
21639 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21641 (vector_ne_<mode>_p): Change this define_expand to not rely on
21642 AltiVec predicate framework.
21643 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21645 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21646 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21647 define_insn pattern.
21648 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21649 define_insn pattern because the xvcmpne<VSs>. instruction is not
21651 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21652 instruction is not supported.
21654 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21656 * config/nvptx/nvptx.c: Include intl.h.
21658 2017-03-01 Martin Jambor <mjambor@suse.cz>
21661 * ipa-prop.h (ipa_bits): Removed field known.
21662 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21663 to pointers. Adjusted their comments to warn about their sharing.
21664 (ipcp_transformation_summary): Change bits to a vector of pointers.
21665 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21666 (ipa_get_ipa_bits_for_value): Declare.
21667 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21668 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21669 (ipa_bits_hash_table): Likewise.
21670 (ipa_vr_ggc_hash_traits): Likewise.
21671 (ipa_vr_hash_table): Likewise.
21672 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21673 being pointers and vr_known being removed.
21674 (ipa_set_jf_unknown): Likewise.
21675 (ipa_get_ipa_bits_for_value): New function.
21676 (ipa_set_jfunc_bits): Likewise.
21677 (ipa_get_value_range): New overloaded functions.
21678 (ipa_set_jfunc_vr): Likewise.
21679 (ipa_compute_jump_functions_for_edge): Use the above functions to
21680 construct bits and vr parts of jump functions.
21681 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21682 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21684 (ipcp_grow_transformations_if_necessary): Also allocate
21685 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21687 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21688 them. Fix too long lines.
21689 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21690 vr_known being removed.
21691 (ipa_read_jump_function): Use new setter functions to construct bits
21692 and vr parts of jump functions or set them to NULL.
21693 (write_ipcp_transformation_info): Adjust for bits being pointers.
21694 (read_ipcp_transformation_info): Likewise.
21695 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21697 Include gt-ipa-prop.h.
21698 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21700 (ipcp_store_bits_results): Likewise.
21701 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21702 Do not write to existing jump functions but use a temporary instead.
21704 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21707 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21708 attempt to use its first operand as BIT_FIELD_REF base.
21710 2017-03-01 Richard Biener <rguenther@suse.de>
21712 PR middle-end/79721
21713 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21714 interpolating formula in wrapping arithmetic.
21715 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21717 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21719 PR tree-optimization/79734
21720 * tree-vect-generic.c (expand_vector_condition): Optimize
21721 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21722 Handle VEC_COND_EXPR where comparison has different inner width from
21723 type's inner width.
21725 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21727 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21728 markup, and similar issues. Remove @opindex entries for things
21729 that aren't options. Add missing -mmpy-option entries.
21731 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21733 PR tree-optimization/79737
21734 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21735 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21736 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21737 instead of byte_size. Formatting fix.
21738 (shift_bytes_in_array_right): Formatting fix.
21740 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21743 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21744 condition on optimize for the leaf function test.
21746 2017-02-28 Martin Liska <mliska@suse.cz>
21749 * read-rtl-function.c (function_reader::handle_unknown_directive):
21750 Bail out when one uses -flto.
21752 2017-02-28 Martin Liska <mliska@suse.cz>
21754 * common.opt: Replace space with tabular for options of <number>
21756 * config/i386/i386.opt: Show <number> value for
21757 -mlarge-data-threshold.
21758 * opts.c (print_filtered_help): Do not display number in hexadecimal
21761 2017-02-28 Martin Liska <mliska@suse.cz>
21763 * common.opt: Fix --help=option -Q for options which are of
21766 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21768 * config/i386/i386.c (print_reg): Error out for values
21769 of 8-bit size in invalid integer register.
21771 2017-02-28 Martin Sebor <msebor@redhat.com>
21773 PR tree-optimization/79691
21774 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21776 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21779 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21780 gcc_unreachable with output_operand_lossage.
21782 2017-02-28 Richard Biener <rguenther@suse.de>
21784 PR tree-optimization/79740
21785 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21787 (visit_nary_op): Insert the nary into the hashtable if we
21788 pattern-matched sth.
21789 * tree-ssa-pre.c (eliminate_insert): Robustify.
21791 2017-02-28 Richard Biener <rguenther@suse.de>
21793 PR middle-end/79731
21794 * fold-const.c (decode_field_reference): Reject out-of-bound
21797 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21799 * config/i386/i386.c: Include intl.h.
21800 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21801 instead of just cond ? "..." : "...".
21802 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21803 * coverage.c (read_counts_file): Likewise.
21804 * omp-offload.c: Include intl.h.
21805 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21806 of just cond ? "..." : "...".
21807 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21808 of just cond ? "..." : "...".
21810 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21813 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21815 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21816 'tune for' CPU name.
21817 * config/arm/arm-cpu-data.h: Regenerated.
21819 2017-02-28 Richard Biener <rguenther@suse.de>
21821 PR tree-optimization/79732
21822 * tree-inline.c (expand_call_inline): Do not shadow var.
21824 2017-02-28 Richard Biener <rguenther@suse.de>
21826 PR tree-optimization/79723
21827 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21828 address-space properly.
21830 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21832 * doc/optinfo.texi (Optimization groups): Fix option used for
21834 * doc/invoke.texi (-fopt-info): Document "omp".
21835 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21836 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21837 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21838 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21839 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21841 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21843 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21844 OPTGROUP_OMP with "omp".
21846 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21849 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21850 for arithmetic shift of unsigned V2DI.
21852 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21854 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21855 arc/linux.h headers.
21856 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21857 (LINK_SPEC): Likewise.
21858 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21859 (EXTRA_SPECS): Likewise.
21860 (STARTFILE_SPEC): Likewise.
21861 (ENDFILE_SPEC): Likewise.
21862 (LIB_SPEC): Likewise.
21863 (TARGET_SDATA_DEFAULT): Likewise.
21864 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21865 (MULTILIB_DEFAULTS): Likewise.
21866 (DWARF2_UNWIND_INFO): Likewise.
21867 * config/arc/big.h: New file.
21868 * config/arc/elf.h: Likewise.
21869 * config/arc/linux.h: Likewise.
21870 * config/arc/t-uClibc: Remove.
21872 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21874 PR tree-optimization/77536
21875 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21876 (tree_transform_and_unroll_loop): Use above function to compute the
21877 estimated niter of unrolled loop and use it when scaling profile.
21878 Also use count info rather than frequency if it's non-zero.
21879 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21880 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21881 (vect_transform_loop): Call above function.
21883 2017-02-27 Richard Biener <rguenther@suse.de>
21885 PR tree-optimization/45397
21886 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21887 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21888 (visit_nary_op): Add pattern matching for CSEing sign-changed
21889 or truncated operations with wider ones.
21891 2017-02-27 Richard Biener <rguenther@suse.de>
21893 PR tree-optimization/79690
21894 * tree-vect-stmts.c (vectorizable_store): Use vector type
21895 built from the DR with address-space.
21897 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21899 * doc/invoke.texi (Optimize Options): Refine the description
21900 of asan-use-after-return.
21902 2017-02-25 Alan Modra <amodra@gmail.com>
21904 PR rtl-optimization/79584
21905 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21906 base, not ad->base_term, the reg within base. Remove assertion
21907 that ad->base == ad->base_term. Replace gen_int_mode using
21908 bogus mode with const0_rtx.
21910 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21912 PR middle-end/79396
21913 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21914 FMA_EXPR like tcc_binary or tcc_unary.
21916 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21919 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21921 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21922 (output_loc_operands): Handle DW_OP_call_ref and
21923 DW_OP_GNU_variable_value.
21924 (struct variable_value_struct): New type.
21925 (struct variable_value_hasher): Likewise.
21926 (variable_value_hash): New variable.
21927 (string_types): Remove.
21928 (copy_loc_descr): New function.
21929 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21930 (prepend_loc_descr_to_each): New function.
21931 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21932 instead of add_loc_descr_to_each if the first argument is single
21933 location list and the second has multiple.
21934 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21935 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21936 when looking for variable value which doesn't have other location info.
21937 (loc_list_from_tree): Formatting fix.
21938 (gen_array_type_die): Simplify DW_AT_string_length handling.
21939 (adjust_string_types): Remove.
21940 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21941 string_types. Call resolve_variable_values.
21942 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21943 (resolve_addr_in_expr): Likewise. Add A argument.
21944 (copy_deref_exprloc): Remove deref argument. Adjust for the
21945 original expression being DW_OP_GNU_variable_value with optionally
21946 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21947 optionally after it.
21948 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21949 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21950 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21951 (variable_value_hasher::hash, variable_value_hasher::equal): New
21953 (resolve_variable_value_in_expr, resolve_variable_value,
21954 resolve_variable_values, note_variable_value_in_expr,
21955 note_variable_value): New functions.
21956 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21959 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21962 * opts.h (handle_generated_option): Add GENERATED_P argument.
21963 * opts-common.c (handle_option): Adjust function comment.
21964 (handle_generated_option): Add GENERATED_P argument, pass it to
21966 (control_warning_option): Pass false to handle_generated_option
21968 * opts.c (maybe_default_option): Pass true to handle_generated_option
21970 * optc-gen.awk: Likewise.
21972 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21974 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21975 a REG, look at the REG it is a SUBREG of.
21976 (splitter for cmpeqsi_t): Ditto.
21978 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21980 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21981 the special USEs with the pattern of the insn, not the insn itself.
21983 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21986 * doc/invoke.texi: Document -mload-store-pairs.
21988 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21989 Sandra Loosemore <sandra@codesourcery.com>
21991 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21992 argument isn't a CONST_INT.
21993 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21994 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21995 (nios2_validate_compare): Bypass alternate compare logic if *op2
21996 is not a CONST_INT.
21997 (ldstwm_operation_p): Return false if first_base is not a REG or
21998 if first_offset is not a CONST_INT.
22000 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22002 * config/cris/cris.md: Use correct operand in a define_peephole2.
22004 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22006 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22008 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22010 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22011 this_insn if it is an INSN or JUMP_INSN.
22012 (force_offsettable): Look at base, not at addr.
22013 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22014 on things that aren't necessarily CONST_INTs.
22016 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22018 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22019 -mfpmath=sse is the default also for x86-32 targets with SSE2
22020 instruction set when @option{-ffast-math} is enabled
22022 2017-02-24 Jeff Law <law@redhat.com>
22024 PR rtl-optimizatoin/79286
22025 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22028 2017-02-24 Richard Biener <rguenther@suse.de>
22030 PR tree-optimization/79389
22031 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22034 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22036 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22037 function comment to reflect reality.
22038 (loop_exits_before_overflow): Fix typo in function description.
22040 2017-02-24 Richard Biener <rguenther@suse.de>
22042 PR tree-optimization/79389
22043 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22044 properly that a threading opportunity exists. Detect conditional
22045 copy/constant propagation opportunities.
22047 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22049 * config/visium/visium.md (type): Add trap.
22050 (b): New mode attribute.
22051 (*btst): Rename into...
22052 (*btst<mode>): ...this and adjust.
22053 (*cbranchsi4_btst_insn): Rename into...
22054 (*cbranch<mode>4_btst_insn): ...this and adjust.
22055 (trap): New define_insn.
22057 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22059 PR tree-optimization/79389
22060 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22061 (noce_reversed_cond_code): New function.
22062 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22063 reversed_comparison_code. Formatting fix.
22064 (noce_try_store_flag): Test rev_cond != NULL in addition to
22065 reversed_comparison_code.
22066 (noce_try_store_flag_constants): Likewise.
22067 (noce_try_store_flag_mask): Likewise.
22068 (noce_try_addcc): Use rev_cond if non-NULL instead of
22069 reversed_comparison_code.
22070 (noce_try_cmove_arith): Likewise. Formatting fixes.
22071 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22072 (noce_find_if_block): Initialize rev_cond.
22073 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22074 instead of false as last argument never attempt to reverse it
22077 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22079 PR tree-optimization/79663
22080 * tree-predcom.c (combine_chains): Process refs in reverse order
22081 only for ZERO length chains, and add explaining comment.
22083 2017-02-23 Jeff Law <law@redhat.com>
22085 PR tree-optimization/79578
22086 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22087 in call to operand_equal_p.
22089 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22092 * config/i386/cpuid.h: Fix another undefined behavior.
22094 2017-02-23 Richard Biener <rguenther@suse.de>
22096 PR tree-optimization/79683
22097 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22098 vector types for data-refs.
22100 2017-02-23 Martin Liska <mliska@suse.cz>
22102 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22104 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22106 PR middle-end/79665
22107 * internal-fn.c (get_range_pos_neg): Moved to ...
22108 * tree.c (get_range_pos_neg): ... here. No longer static.
22109 * tree.h (get_range_pos_neg): New prototype.
22110 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22111 are known to be in between 0 and signed maximum inclusive, try to
22112 expand both unsigned and signed divmod and use the cheaper one from
22115 2017-02-22 Jeff Law <law@redhat.com>
22117 PR tree-optimization/79578
22118 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22119 to compare base operands.
22121 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22124 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22125 gpc_reg_operand instead of fpr_reg_operand.
22127 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22129 * config/mips/mips.c (mips_return_in_memory): Force FP
22130 vector types to be returned in memory for o32 ABI.
22132 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22134 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22135 instead of DW_TAG_member for static data member declarations and don't
22136 set no_linkage_name for static inline data members.
22137 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22140 2017-02-22 Martin Liska <mliska@suse.cz>
22142 * doc/invoke.texi: Replace inequality signs with square brackets
22145 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22147 PR tree-optimization/68644
22148 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22150 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22153 * lra-constraints.c (simplify_operand_subreg): Handle
22154 WORD_REGISTER_OPERATIONS targets.
22156 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22159 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22160 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22161 elimination by swapping fld*.
22163 2017-02-22 Richard Biener <rguenther@suse.de>
22165 PR tree-optimization/79673
22166 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22167 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22168 irrelevant address-space qualifiers and avoiding a
22169 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22171 2017-02-22 Richard Biener <rguenther@suse.de>
22173 PR tree-optimization/79666
22174 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22175 to not symbolically negate if that may introduce undefined
22178 2017-02-22 Martin Liska <mliska@suse.cz>
22181 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22182 * data-streamer-out.c (streamer_write_gcov_count_stream):
22184 * value-prof.c (stream_out_histogram_value): Make assert more
22185 precise based on type of counter.
22187 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22190 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22191 Use nonimmediate_operand instead of memory_operand for operand 1.
22192 (float-extend standard_x87sse_constant_load splitter): Ditto.
22194 2017-02-21 Jeff Law <law@redhat.com>
22196 PR tree-optimization/79621
22197 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22198 blocks with edges to themselves.
22200 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22203 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22204 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22205 Use gimple_call_builtin_p.
22208 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22209 on temporarily removed DEBUG_INSNs.
22211 PR tree-optimization/79649
22212 * tree-loop-distribution.c (classify_partition): Give up on
22213 non-generic address space loads/stores.
22215 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22217 * doc/loop.texi (Loop manipulation): Remove nonexistent
22218 tree_ssa_loop_version from the documentation.
22219 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22221 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22224 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22225 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22226 * config/rs6000/rs6000.c: Include except.h.
22227 (rs6000_expand_split_stack_prologue): Call
22228 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22230 2017-02-21 Martin Jambor <mjambor@suse.cz>
22233 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22234 have been analyzed.
22236 2017-02-21 Martin Jambor <mjambor@suse.cz>
22238 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22239 for backward compatibility only.
22240 * doc/invoke.texi (Option Summary): Remove all references to
22241 -fipa-cp-alignment.
22243 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22247 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22249 * lra-constraints.c (curr_insn_transform): Handle
22250 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22252 2017-02-21 Martin Liska <mliska@suse.cz>
22254 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22256 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22258 PR translation/79638
22259 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22261 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22264 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22265 (arm_function_ok_for_sibcall): Return false for an indirect call by
22266 descriptor if all the argument registers are used.
22267 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22268 alignment of the function.
22270 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22272 PR tree-optimization/61441
22273 * simplify-rtx.c (simplify_const_unary_operation): For
22274 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22275 the sNaN unmodified.
22277 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22279 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22280 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22281 instead of SYSTEM_HEADER_DIR.
22283 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22284 Martin Liška <mliska@suse.cz>
22286 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22287 Fix typos and grammar, use active voice, and clarify.
22289 2017-02-20 Marek Polacek <polacek@redhat.com>
22291 PR middle-end/79537
22292 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22295 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22297 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22300 * config/i386/i386.c (ix86_expand_builtin): Handle
22301 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22302 ix86_builtins_isa[fcode].isa as a requirement of those
22303 flags and any other flag in the bitmask.
22304 (ix86_init_mmx_sse_builtins): Use 0 instead of
22305 ~OPTION_MASK_ISA_64BIT as mask.
22306 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22307 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22308 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22309 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22311 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22314 * lra-constraints.c (split_reg): Check requested split mode
22315 is supported by the register.
22317 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22319 * lra-constraints.c (simplify_operand_subreg): Remove early
22322 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22325 * lra-constraints.c (curr_insn_transform): Tighten condition
22326 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22328 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22331 * lra-constraints.c (curr_insn_transform): Handle
22332 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22334 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22337 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22339 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22341 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22344 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22347 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22349 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22352 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22354 * final.c (last_columnnum, override_columnnum): New variables.
22355 (final_start_function): Set last_columnnum, pass it to begin_prologue
22356 hook and pass 0 to dwarf2out_begin_prologue.
22357 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22358 to source_line debug hook.
22359 (notice_source_line): Compute last_columnnum and for debug_column_info
22360 return true on column changes.
22361 * debug.h (struct gcc_debug_hooks): Add column argument to
22362 source_line and begin_prologue hooks.
22363 (debug_nothing_int_charstar_int_bool): Remove prototype.
22364 (debug_nothing_int_int_charstar,
22365 debug_nothing_int_int_charstar_int_bool): New prototypes.
22366 (dwarf2out_begin_prologue): Add column argument.
22367 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22368 begin_prologue hooks.
22369 (debug_nothing_int_charstar_int_bool): Remove.
22370 (debug_nothing_int_int_charstar,
22371 debug_nothing_int_int_charstar_int_bool): New functions.
22372 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22373 through to dwarf2out_source_line.
22374 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22375 (dwarf2out_source_line): Add column argument, emit it if requested.
22376 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22378 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22379 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22380 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22381 through to dwarf2out_begin_prologue.
22382 (vmsdbgout_source_line): Add column argument, pass it through to
22383 dwarf2out_source_line.
22384 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22385 dbxout_source_line caller.
22386 (dbxout_source_line): Add column argument.
22388 * common.opt (gno-column-info, gcolumn-info): New options.
22389 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22390 (check_die): Also test for multiple DW_AT_decl_column attributes.
22391 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22392 DW_AT_decl_column if requested.
22393 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22395 (gen_variable_die): Likewise.
22396 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22397 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22400 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22401 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22402 (ix86_handle_option): Handle OPT_m3dnowa.
22403 * doc/invoke.texi (-m3dnowa): Document.
22404 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22405 -m3dnowa instead of -m3dnow -march=athlon.
22408 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22409 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22411 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22414 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22415 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22416 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22417 generator for vsx_xxpermdi_<mode>_be.
22418 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22419 force big-endian semantics.
22420 (vsx_xxpermdi_<mode>_be): New define_expand with same
22421 implementation as previous version of vsx_xxpermdi_<mode>.
22423 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22425 PR tree-optimization/79327
22426 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22427 variable, its initialization and use.
22429 2017-02-17 Julia Koval <julia.koval@intel.com>
22431 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22432 (OPTION_MASK_ISA_PKU_UNSET): New.
22433 (ix86_handle_option): Handle -mrdpid.
22434 * config/i386/cpuid.h (bit_RDPID): New.
22435 * config/i386/driver-i386.c (host_detect_local_cpu):
22436 Detect RDPID feature.
22437 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22438 * config/i386/i386-c.c (ix86_target_macros_internal):
22440 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22441 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22442 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22443 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22444 * config/i386/i386.md (define_insn "rdpid"): New.
22445 * config/i386/i386.opt Add -mrdpid.
22446 * config/i386/immintrin.h (_rdpid_u32): New.
22448 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22450 PR rtl-optimization/79541
22451 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22452 instead of transforming it into USE.
22454 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22456 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22457 If HONOR_SNANS (SFmode) force the input to a register.
22458 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22459 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22460 an frsp or similar insn.
22462 2017-02-17 Martin Liska <mliska@suse.cz>
22464 PR rtl-optimization/79577
22465 * params.def (selsched-max-sched-times): Increase minimum to 1.
22467 2017-02-17 Martin Liska <mliska@suse.cz>
22469 PR rtl-optimization/79574
22470 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22472 2017-02-17 Martin Liska <mliska@suse.cz>
22474 PR tree-optimization/79529
22475 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22476 ssa_defined_default_def_p to handle cases which are implicitly
22478 * tree-ssa.c (ssa_defined_default_def_p): New function.
22479 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22480 which are implicitly defined.
22481 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22483 2017-02-17 Richard Biener <rguenther@suse.de>
22485 PR middle-end/79576
22486 * params.def (max-ssa-name-query-depth): Limit to 10.
22488 2017-02-17 Richard Biener <rguenther@suse.de>
22490 PR tree-optimization/79552
22491 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22494 2017-02-17 Richard Biener <rguenther@suse.de>
22497 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22499 2017-02-17 Marek Polacek <polacek@redhat.com>
22501 PR middle-end/79536
22502 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22503 (fold_negate_expr): New wrapper.
22505 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22507 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22508 Correct terminology and de-emphasize pre-standard behavior.
22510 2017-02-16 Alan Modra <amodra@gmail.com>
22512 PR rtl-optimization/79286
22513 * ira.c (def_dominates_uses): New function.
22514 (update_equiv_regs): Don't create an equivalence for insns that
22515 may trap where the register def does not dominate the use.
22517 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22519 PR rtl-optimization/78127
22520 * lra.c (lra): Call lra_eliminate before finish the loop after
22523 2017-02-16 Richard Biener <rguenther@suse.de>
22525 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22527 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22528 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22529 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22530 (isl_val_int_from_wi): New function.
22531 (extract_affine_gmp): Rename to ...
22532 (extract_affine_wi): ... this, take a widest_int.
22533 (extract_affine_int): Just wrap extract_affine_wi.
22534 (add_param_constraints): Use isl_val_int_from_wi.
22535 (add_loop_constraints): Likewise, and extract_affine_wi.
22537 2017-02-15 Jeff Law <law@redhat.com>
22539 PR middle-end/79521
22540 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22541 ira_init_register_move_cost_if_necessary.
22543 2017-02-15 Martin Sebor <msebor@redhat.com>
22545 PR middle-end/32003
22546 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22547 removed in a prior commit.
22549 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22551 PR tree-optimization/79347
22552 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22553 counters during peeling.
22555 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22557 * Makefile.in (site.exp): Remove "set ISLVER".
22559 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22562 * real.c (real_from_integer): Call real_convert even for decimal.
22564 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22567 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22569 2017-02-14 Andrew Pinski <apinski@cavium.com>
22571 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22572 cores and change the partno/implementer to be correct.
22573 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22574 the 'B" as the implementer.
22575 * config/aarch64/aarch64-tune.md: Regenerate.
22577 2017-02-14 Carl Love <cel@us.ibm.com>
22579 * config/rs6000/rs6000.c: Add case statement entry to make the
22580 xvcvuxdsp built-in argument unsigned.
22581 * config/rs6000/vsx.md: Fix the source and return operand types so they
22582 match the instruction definitions from the ISA document. Fix typo
22583 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22586 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22589 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22590 member early_clobber_alts.
22591 * lra-lives.c (reg_early_clobber_p): New.
22592 (process_bb_lives): Use it.
22593 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22594 (debug_operand_data): Initialize early_clobber_alts.
22595 (setup_operand_alternative): Set up early_clobber_alts.
22596 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22597 alternatives to new_insn_reg.
22598 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22600 (lra_update_insn_regno_info): Pass the new arg.
22602 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22604 PR middle-end/79505
22605 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22606 (new_oacc_loop_raw): Don't clear already cleared fields.
22609 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22610 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22611 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22613 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22616 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22618 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22621 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22622 the extra instruction to the right place to store 128-bit constant
22625 2017-02-14 Martin Sebor <msebor@redhat.com>
22627 PR middle-end/79448
22628 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22629 warning for strings of unknown length.
22631 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22633 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22635 2017-02-14 Jeff Law <law@redhat.com>
22638 * ira-costs.c (scan_one_insn): Initialize register move costs
22639 for pseudos seen in USE/CLOBBER insns.
22641 PR tree-optimization/79095
22642 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22643 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22644 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22645 if the operands are known to be not equal, then the resulting range
22647 (intersect_ranges): If the new range is ~[0,0] and the old range is
22648 wide, then prefer ~[0,0].
22649 * tree-vrp.c (overflow_comparison_p_1): New function.
22650 (overflow_comparison_p): New function.
22651 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22652 if NAME is used in an overflow test.
22653 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22654 overflow check that can be expressed as an equality test, then adjust
22655 ops to be that equality test.
22657 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22659 * config/s390/s390-builtin-types.def: Remove flags argument.
22660 * config/s390/s390.c (s390_init_builtins): Likewise.
22662 2017-02-14 Martin Liska <mliska@suse.cz>
22664 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22665 vector. Fix trailing white spaces.
22667 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22669 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22672 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22674 PR rtl-optimization/68664
22675 * config/arm/arm.c (arm_sched_can_speculate_insn):
22676 New function. Declare prototype.
22677 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22679 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22681 PR rtl-optimization/68664
22682 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22684 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22686 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22688 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22689 max skip bytes for function, loop and jump.
22691 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22693 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22694 ABS_EXPR for gimple dump.
22696 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22699 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22701 PR tree-optimization/79408
22702 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22703 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22704 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22705 also if rhs1 is INTEGER_CST.
22707 2017-02-14 Richard Biener <rguenther@suse.de>
22709 PR middle-end/79432
22710 * tree-into-ssa.c (insert_phi_nodes): When the function can
22711 have abnormal edges rewrite SSA names with broken use-def
22712 dominance out of SSA and register them for PHI insertion.
22714 2017-02-13 Martin Sebor <msebor@redhat.com>
22716 PR middle-end/79496
22717 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22718 clearing info.nowrite flag when snprintf size argument is a range.
22720 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22722 * cprop.c (cprop_jump): Add missing space in string literal.
22723 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22724 (get_constraint_for_component_ref): Likewise.
22725 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22726 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22727 * lra-constraints.c (process_alt_operands): Likewise.
22728 * ipa-inline.c (inline_small_functions): Likewise.
22729 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22730 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22731 * trans-mem.c (diagnose_tm_1_op): Likewise.
22732 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22733 (grid_parallel_clauses_gridifiable): Likewise.
22735 * config/nvptx/mkoffload.c (process): Add space in between
22738 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22739 "MOD4_SSE_REGS" and "ALL_REGS".
22741 * spellcheck.c (test_data): Add , in between "foo" and "food".
22743 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22746 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22747 boundary crossing check and subsequent code generation agree.
22749 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22751 * config/aarch64/aarch64.c (has_memory_op): Delete.
22752 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22755 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22757 PR rtl-optimization/79388
22758 PR rtl-optimization/79450
22759 * combine.c (distribute_notes): When removing TEM_INSN for which
22760 corresponding dest has last value recorded, invalidate that last
22763 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22765 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22766 of explicit '@'. Add missing assembly comment marker on branch costs
22769 2017-02-13 Nathan Sidwell <nathan@acm.org>
22771 * gengtype-lex.l (<in_struct>): Add '/'.
22773 2017-02-13 Martin Liska <mliska@suse.cz>
22776 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22778 2017-02-13 Richard Biener <rguenther@suse.de>
22780 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22782 * configure: Re-generate.
22783 * config.in: Likewise.
22784 * graphite-dependences.c: Simplify as if
22785 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22786 * graphite-isl-ast-to-gimple.c: Likewise.
22787 * graphite-optimize-isl.c: Likewise.
22788 * graphite-poly.c: Likewise.
22789 * graphite-sese-to-poly.c: Likewise.
22790 * graphite.h: Likewise.
22791 * toplev.c: Include isl/version.h and use isl_version () for
22792 printing the ISL version.
22793 * doc/install.texi: Update ISL requirement.
22795 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22797 * doc/standards.texi (Standards): Update reference to
22800 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22802 * doc/extend.texi (Named Address Spaces): sourceware.org now
22804 * doc/install.texi (Binaries): Ditto.
22807 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22809 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22810 standard terminology "stringize"/"stringizing" throughout.
22811 * doc/cppinternals.texi: Likewise.
22813 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22815 * doc/extend.texi: Fix some spelling mistakes and typos.
22816 * doc/invoke.texi: Likewise.
22818 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22821 * params.def (inline-min-speedup) Change from 10 to 8.
22823 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22825 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22828 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22831 * ipa-inline-analysis.c (get_minimal_bb): New function.
22832 (record_modified): Use it.
22833 (remap_edge_change_prob): Handle also ancestor functions.
22835 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22837 * doc/contrib.texi (Contributors): Remove broken link into
22838 the Mauve CVS repository.
22840 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22842 PR middle-end/79454
22843 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22844 result computation whenever lhs doesn't have vector mode, not
22845 just when it has BLKmode.
22847 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22849 * doc/makefile.texi (profiledbootstrap): Refer to the
22850 installation instructions only in textual form.
22852 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22855 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22857 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22859 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22860 (Specific): Update mingw-w64 reference.
22862 (Specific): Remove broken link to Renesas RX processor.
22864 2017-02-10 Richard Biener <rguenther@suse.de>
22866 * toplev.c (process_options): Do not mention obsolete graphite
22867 options when printing sorry message about missing graphite support.
22868 Mention -floop-nest-optimize.
22870 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22872 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22873 (vtst_p16): Likewise.
22874 (vtstq_p8): Likewise.
22875 (vtstq_p16): Likewise.
22877 (vtstq_p64): Likewise.
22878 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22879 (vset_lane_p64): New.
22880 (vsetq_lane_p64): New.
22882 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22884 PR tree-optimization/79411
22885 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22886 stmt operands are SSA_NAMEs used in abnormal phis.
22887 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22890 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22893 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22896 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22898 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22900 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22902 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22906 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22907 not arbitrary TREE_CONSTANT.
22910 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22911 "omp declare target link" attribute unless is_global_var.
22912 * omp-offload.c (find_link_var_op): Likewise.
22914 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22915 Chung-Lin Tang <cltang@codesourcery.com>
22917 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22919 (gimplify_adjust_omp_clauses): Don't delete TILE.
22920 (gimplify_omp_for): Deal with TILE.
22921 * internal-fn.c (expand_GOACC_TILE): New function.
22922 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22924 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22925 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22927 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22928 avoid DIV for outermost collapse var.
22929 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22930 Remove out of date comments, fix whitespace.
22931 * omp-general.c (omp_extract_for_data): Deal with tiling.
22932 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22933 adjust OLF_DIM_BASE value.
22934 (struct omp_for_data): Add tiling field.
22935 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22936 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22937 for auto loops. Remove default auto determining, moved to
22938 oacc_loop_fixed_partitions.
22939 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22940 stmts, add e_mask field.
22941 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22942 (oacc_thread_numbers): Use oacc_dim_call.
22943 (oacc_xform_tile): New.
22944 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22945 (finish_oacc_loop): Adjust for ifns vector.
22946 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22947 add case for GOACC_TILE fns.
22948 (oacc_loop_xform_loop): Delete.
22949 (oacc_loop_process): Iterate over call list directly, and add
22950 handling for GOACC_TILE fns.
22951 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22953 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22954 vector partitioning to outer loops. Assign 2 partitions to loops
22955 when available. Add TILE handling.
22956 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22957 (execite_oacc_device_lower): Process GOACC_TILE fns,
22958 ignore unknown specs.
22959 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22960 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22961 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22963 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22965 * configure.ac (ACX_BUGURL): Update.
22966 * configure: Regenerate.
22968 2017-02-09 Richard Biener <rguenther@suse.de>
22970 PR tree-optimization/69823
22971 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22972 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22974 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22976 * config/arc/arc-c.def: Add __NPS400__ definition.
22977 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22978 (TARGET_NPS400): Define.
22980 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22982 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22984 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22985 pointer, arch_info.
22986 (arc_cpu_types): Fill the arch_info field with a pointer into the
22987 arc_arch_types table.
22988 (arc_selected_cpu): Declare.
22989 * config/arc/arc.c (arc_selected_cpu): Make global.
22990 (arc_selected_arch): Delete.
22991 (arc_base_cpu): Delete.
22992 (arc_override_options): Remove references to deleted variables,
22993 update access to arch information.
22994 (ARC_OPT): Update access to arch information.
22995 (ARC_OPTX): Likewise.
22996 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22997 (TARGET_ARC600): Update access to arch information.
22998 (TARGET_ARC601): Likewise.
22999 (TARGET_ARC700): Likewise.
23000 (TARGET_EM): Likewise.
23001 (TARGET_HS): Likewise.
23002 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23005 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23008 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23009 condition/operands for integer GE/LE/GEU/LEU operations.
23011 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23013 PR translation/79397
23014 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23017 2017-02-08 Martin Jambor <mjambor@suse.cz>
23020 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23021 whether allocation happened.
23022 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23023 nothing was allocated.
23025 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23027 PR tree-optimization/79408
23028 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23029 constant, but SSA_NAME with a known integer range, use the minimum
23030 of that range instead of op1 to determine if modulo can be replaced
23031 with its first operand.
23033 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23035 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23037 2017-02-08 Richard Biener <rguenther@suse.de>
23039 PR tree-optimization/71824
23040 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23041 Check all loops contained in the merged region.
23043 2017-02-07 Andrew Pinski <apinski@cavium.com>
23045 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23047 2017-02-07 Andrew Pinski <apinski@cavium.com>
23049 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23050 (thunderxt88): Likewise.
23051 (thunderxt81): Disable LSE and change v8.1 to v8.
23052 (thunderxt83): Likewise.
23054 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23055 Richard Biener <rguenther@suse.de>
23057 PR middle-end/79399
23058 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23059 type from int to size_t.
23060 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23062 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23064 PR rtl-optimization/79386
23065 * cprop.c (bypass_conditional_jumps): Initialize
23066 bypass_last_basic_block already before splitting bbs after
23067 unconditional traps...
23068 (bypass_conditional_jumps): ... rather than here.
23071 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23072 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23073 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23074 fix -masm=intel patterns.
23076 2017-02-07 Richard Biener <rguenther@suse.de>
23078 PR tree-optimization/79256
23079 PR middle-end/79278
23080 * builtins.c (get_object_alignment_2): Use min_align_of_type
23081 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23082 and ADJUST_FIELD_ALIGN.
23084 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23086 * doc/tm.texi: Regenerate.
23087 * stor-layout.c (layout_decl): Adjust.
23088 (update_alignment_for_field): Likewise.
23089 (place_field): Likewise.
23090 (min_align_of_type): Likewise.
23091 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23092 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23093 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23094 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23095 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23096 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23097 * config/i386/i386.c (x86_field_alignment): Likewise.
23098 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23099 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23100 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23101 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23102 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23103 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23107 2017-01-30 Richard Biener <rguenther@suse.de>
23109 PR tree-optimization/79256
23110 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23111 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23114 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23116 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23117 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23118 builtins to SImode and emit a zero-extend, if necessary.
23120 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23122 * docs/invoke.texi (RISC-V Options): Alphabetize.
23124 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23126 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23129 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23131 * config/riscv/riscv.c: New file.
23132 * common/config/riscv/riscv-common.c: Likewise.
23133 * config.gcc: Likewise.
23134 * config/riscv/constraints.md: Likewise.
23135 * config/riscv/elf.h: Likewise.
23136 * config/riscv/generic.md: Likewise.
23137 * config/riscv/linux.h: Likewise.
23138 * config/riscv/multilib-generator: Likewise.
23139 * config/riscv/peephole.md: Likewise.
23140 * config/riscv/pic.md: Likewise.
23141 * config/riscv/predicates.md: Likewise.
23142 * config/riscv/riscv-builtins.c: Likewise.
23143 * config/riscv/riscv-c.c: Likewise.
23144 * config/riscv/riscv-ftypes.def: Likewise.
23145 * config/riscv/riscv-modes.def: Likewise.
23146 * config/riscv/riscv-opts.h: Likewise.
23147 * config/riscv/riscv-protos.h: Likewise.
23148 * config/riscv/riscv.h: Likewise.
23149 * config/riscv/riscv.md: Likewise.
23150 * config/riscv/riscv.opt: Likewise.
23151 * config/riscv/sync.md: Likewise.
23152 * config/riscv/t-elf-multilib: Likewise.
23153 * config/riscv/t-linux: Likewise.
23154 * config/riscv/t-linux-multilib: Likewise.
23155 * config/riscv/t-riscv: Likewise.
23156 * configure.ac: Likewise.
23157 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23158 Waterman as RISC-V maintainers.
23159 * doc/install.texi: Add RISC-V entries.
23160 * doc/invoke.texi: Add RISC-V options section.
23161 * doc/md.texi: Add RISC-V constraints section.
23162 * configure: Regenerated.
23164 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23167 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23168 false values to be constant vectors with all 0 or all 1 bits set.
23169 (vcondu<mode><mode>): Likewise.
23170 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23172 (fpmask_comparison_operator): Update comment.
23173 (vecint_comparison_operator): New predicate.
23174 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23175 vector conditionals when the true and false values are constant
23176 vectors with all 0 bits or all 1 bits set.
23178 2017-02-06 Martin Sebor <msebor@redhat.com>
23180 PR tree-optimization/79376
23181 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23183 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23185 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23186 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23187 to simplify split condition.
23189 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23191 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23192 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23195 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23197 PR rtl-optimization/68664
23198 * target.def (can_speculate_insn): New hook.
23199 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23200 * doc/tm.texi: Regenerate.
23201 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23202 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23203 (rs6000_sched_can_speculate_insn): New function.
23205 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23207 PR tree-optimization/79284
23208 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23209 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23210 vectorizable_mask_load_store, vectorizable_operation,
23211 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23212 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23213 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23214 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23215 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23216 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23217 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23218 is_gimple_assign (stmt). Replace another such test with
23219 is_gimple_assign (stmt).
23221 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23224 * config/avr/avr.c (rtl-iter.h): Include it.
23225 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23226 (avr_legitimate_combined_insn): ...and implementation.
23228 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23230 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23231 * config/s390/s390.c (s390_const_operand_ok)
23232 (s390_canonicalize_comparison, s390_extract_part)
23233 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23234 (s390_contiguous_bitmask_p, s390_rtx_costs)
23235 (legitimize_pic_address): Likewise.
23236 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23237 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23238 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23239 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23240 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23242 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23244 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23245 REGNO($0) == REGNO($1).
23247 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23249 * config/s390/linux.h(SIZE_TYPE): Add comment.
23251 2017-02-06 Julian Brown <julian@codesourcery.com>
23252 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23253 Virendra Pathak <virendra.pathak@broadcom.com>
23255 * config/aarch64/aarch64-cores.def: Change the scheduler
23257 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23258 * config/aarch64/thunderx2t99.md: New file.
23260 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23262 * doc/standards.texi (Go Language): Update link to language
23265 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23267 * tree-eh.c (lower_resx): Sanitize profile.
23268 (cleanup_empty_eh_move_lp): Likewise.
23270 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23273 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23275 * cfgloopmanip.h (loop_version): Update prototype.
23276 * modulo-sched.c (sms_schedule): Update call of loop_version.
23277 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23278 * tree-parloops.c (gen_parallel_loop): Likewise.
23279 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23280 * tree-ssa-loop-split.c (split_loop): Likewise.
23281 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23282 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23284 2017-02-05 Martin Liska <mliska@suse.cz>
23287 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23289 (print_operand_address): Initialize a struct to zero.
23291 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23293 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23294 garbage collector only in textual form.
23296 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23298 * doc/extend.texi (x86 specific memory model extensions for
23299 transactional memory): Simplify a phrase.
23301 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23304 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23305 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23306 (atomic_storedi_1): Likewise.
23308 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23310 PR tree-optimization/79338
23311 * tree-parloops.c (gather_scalar_reductions): Don't call
23312 vect_analyze_loop_form for loop->inner before destroying loop's
23315 2017-02-03 Martin Sebor <msebor@redhat.com>
23317 PR tree-optimization/79327
23318 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23319 when precision has resulted in leading zeros.
23320 (format_integer): Adjust the likely counter to assume an unknown
23321 argument that may be zero is non-zero.
23323 2017-02-03 Jason Merrill <jason@redhat.com>
23326 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23327 avoid copying non-taken branch.
23329 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23331 PR tree-optimization/79340
23332 * tree-vect-loop.c (vectorizable_reduction): Release
23333 vec_defs elements after safe_splicing them into other vectors.
23336 PR tree-optimization/79327
23337 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23338 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23340 (format_integer): Use wide_int_to_tree instead of build_int_cst
23341 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23342 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23343 of shortest and longest sequence.
23345 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23347 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23348 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23350 2017-02-03 Walter Lee <walt@tilera.com>
23353 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23354 after initial stackframe link reg save.
23355 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23357 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23360 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23361 wu for stxssp alternative.
23363 2017-02-03 Martin Sebor <msebor@redhat.com>
23365 PR tree-optimization/79352
23366 * gimple-fold.c (get_range_strlen): Add argument.
23367 (get_range_strlen): Change return type to bool.
23368 (get_maxval_strlen): Pass in a dummy argument.
23369 * gimple-fold.h (get_range_strlen): Change return type to bool.
23370 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23371 * tree.h (array_at_struct_end_p): Add argument.
23372 * tree.c (array_at_struct_end_p): Handle it.
23374 2017-02-03 Martin Liska <mliska@suse.cz>
23377 * multiple_target.c (create_dispatcher_calls): Redirect edge
23378 from a caller of a dispatcher.
23379 (expand_target_clones): Make the clones local.
23380 (ipa_target_clone): Do both target clones and resolvers.
23381 (ipa_dispatcher_calls): Remove the pass.
23382 (pass_dispatcher_calls::gate): Likewise.
23383 (make_pass_dispatcher_calls): Likewise.
23384 * passes.def (pass_target_clone): Put as very first IPA early
23387 2017-02-03 Martin Liska <mliska@suse.cz>
23389 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23390 in case of a function with ifunc attribute.
23392 2017-02-03 Martin Liska <mliska@suse.cz>
23394 * cgraph.c (cgraph_node::dump): Dump function version info.
23395 * symtab.c (symtab_node::dump_base): Add missing new line.
23397 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23399 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23400 (ifcombine_ifandif): Use it.
23402 2017-02-03 Martin Liska <mliska@suse.cz>
23404 * doc/invoke.texi: Document default value for
23405 use-after-scope-direct-emission-threshold.
23407 2017-02-03 Martin Liska <mliska@suse.cz>
23409 PR tree-optimization/79339
23410 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23411 (format_floating): Likewise.
23413 2017-02-03 Martin Liska <mliska@suse.cz>
23416 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23418 (ipa_node_params_t::remove): Likewise.
23419 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23420 initialization from removed ipa_node_params_t::insert.
23421 (ipa_node_params::~ipa_node_params): Move from removed
23422 ipa_node_params_t::release.
23423 * symbol-summary.h (symbol_summary::m_released): New member.
23424 Do not release a summary twice. Do not allow to call finalizer
23425 for types of a summary that live in GGC memory.
23427 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23429 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23432 2017-02-02 Martin Sebor <msebor@redhat.com>
23434 PR middle-end/79275
23435 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23436 (format_string): Tighten up the range of output for non-constant
23437 strings and correct the expected range for wide non-constant strings.
23439 2017-02-02 Martin Sebor <msebor@redhat.com>
23441 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23443 PR middle-end/32003
23444 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23446 (-fdump-tree-@var): Add to index and document how to come up
23447 with pass-specific option and dump file names.
23448 (-fdump-passes): Clarify where to look for output.
23450 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23452 PR middle-end/77445
23453 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23454 statistics of the analyzed path; allow threading for speed when
23455 any of BBs along the path are optimized for speed.
23457 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23459 PR middle-end/78468
23460 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23461 settings of the virtual registers.
23464 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23466 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23467 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23470 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23472 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23473 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23475 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23477 * config/s390/s390.md: Add missing comments with the expanded
23479 * config/s390/vector.md: Likewise.
23480 * config/s390/vx-builtins.md: Likewise.
23482 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23485 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23486 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23487 conditions on a single line.
23489 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23491 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23492 __S390_VX__ to __VX__.
23494 2017-02-01 Andrew Pinski <apinski@cavium.com>
23496 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23497 stmt_info to record_stmt_cost.
23498 (vect_get_known_peeling_cost): Pass stmt_info if known to
23500 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23501 cpu_vector_cost field into
23502 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23503 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23504 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23505 splitting of scalar_stmt_cost and vec_stmt_cost.
23506 (thunderx_vector_cost): Likewise.
23507 (cortexa57_vector_cost): LIkewise.
23508 (exynosm1_vector_cost): Likewise.
23509 (xgene1_vector_cost): Likewise.
23510 (thunderx2t99_vector_cost): Improve after the splitting of the two
23512 (aarch64_builtin_vectorization_cost): Update for the splitting of
23513 scalar_stmt_cost and vec_stmt_cost.
23515 2017-02-01 Torvald Riegel <triegel@redhat.com>
23516 Richard Henderson <rth@redhat.com>
23518 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23519 conditional on existance of a fast atomic load.
23520 * optabs-query.c (can_atomic_load_p): New function.
23521 * optabs-query.h (can_atomic_load_p): Declare it.
23522 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23523 no fast atomic load is available for the particular size of access.
23524 (expand_atomic_compare_and_swap): Likewise.
23525 (expand_atomic_load): Likewise.
23526 (expand_atomic_store): Likewise.
23527 (expand_atomic_fetch_op): Likewise.
23528 * testsuite/lib/target-supports.exp
23529 (check_effective_target_sync_int_128): Remove x86 because it provides
23530 no fast atomic load.
23531 (check_effective_target_sync_int_128_runtime): Likewise.
23533 2017-02-01 Richard Biener <rguenther@suse.de>
23535 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23536 (graphite_transform_loops): Provide opt-info for optimized nests.
23537 * tree-parloop.c (parallelize_loops): Provide opt-info for
23538 parallelized loops.
23540 2017-02-01 Richard Biener <rguenther@suse.de>
23542 PR middle-end/79315
23543 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23544 was not set before.
23546 2017-02-01 Richard Biener <rguenther@suse.de>
23548 PR tree-optimization/71824
23549 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23550 Verify the loops are valid in the merged SESE region.
23551 (scop_detection::can_represent_loop_1): Check analyzing the
23552 evolution of the number of iterations in the region succeeds.
23554 2017-01-31 Ian Lance Taylor <iant@golang.org>
23556 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23557 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23559 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23561 PR preprocessor/79210
23562 * input.c (get_substring_ranges_for_loc): Replace line_width
23563 assertion with error-handling.
23565 2017-01-31 Richard Biener <rguenther@suse.de>
23567 PR tree-optimization/77318
23568 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23569 (create_pw_aff_from_tree): Take loop parameter.
23570 (add_condition_to_pbb): Pass loop of the condition to
23571 create_pw_aff_from_tree.
23573 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23575 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23576 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23578 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23582 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23584 (convert_int_to_float128): Likewise.
23585 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23586 (convert_int_to_float128): Likewise.
23587 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23588 (UNSPEC_IEEE128_CONVERT): Likewise.
23589 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23590 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23591 Use local variables for IBM extended format.
23592 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23593 (fix_trunc<mode>si2_fprs): Likewise.
23594 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23595 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23596 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23597 to know that we can now have integers of all sizes in vector
23599 (fix<uns>_<mode>di2_hw): Likewise.
23600 (float<uns>_<mode>si2_hw): Likewise.
23601 (fix_<mode>si2_hw): Likewise.
23602 (fixuns_<mode>si2_hw): Likewise.
23603 (float<uns>_<mode>di2_hw): Likewise.
23604 (float_<mode>di2_hw): Likewise.
23605 (float_<mode>si2_hw): Likewise.
23606 (floatuns_<mode>di2_hw): Likewise.
23607 (floatuns_<mode>si2_hw): Likewise.
23608 (xscvqp<su>wz_<mode>): Delete, no longer used.
23609 (xscvqp<su>dz_<mode>): Likewise.
23610 (xscv<su>dqp_<mode>): Likewise.
23611 (ieee128_mfvsrd_64bit): Likewise.
23612 (ieee128_mfvsrd_32bit): Likewise.
23613 (ieee128_mfvsrwz): Likewise.
23614 (ieee128_mtvsrw): Likewise.
23615 (ieee128_mtvsrd_64bit): Likewise.
23616 (ieee128_mtvsrd_32bit): Likewise.
23618 2017-01-31 Martin Liska <mliska@suse.cz>
23621 * ipa-prop.c (ipa_free_all_node_params): Call release method
23622 instead of ~sumbol_summary to not to trigger double times
23625 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23627 PR tree-optimization/71691
23628 * bitmap.h (class auto_bitmap): New.
23629 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23630 is_maybe_undefined instead of ssa_undefined_value_p.
23632 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23634 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23635 __S390_ARCH_LEVEL__ to __ARCH__.
23637 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23639 PR tree-optimization/79267
23640 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23641 if should_remove_lhs_p is true.
23643 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23646 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23647 (add_alignment_attribute): New.
23648 (base_type_die): Add alignment attribute.
23649 (subrange_type_die): Likewise.
23650 (modified_type_die): Likewise.
23651 (gen_array_type_die): Likewise.
23652 (gen_descr_array_type_die: Likewise.
23653 (gen_enumeration_type_die): Likewise.
23654 (gen_subprogram_die): Likewise.
23655 (gen_variable_die): Likewise.
23656 (gen_field_die): Likewise.
23657 (gen_ptr_to_mbr_type_die): Likewise.
23658 (gen_struct_or_union_type_die): Likewise.
23659 (gen_subroutine_type_die): Likewise.
23660 (gen_typedef_die): Likewise.
23661 (base_type_cmp): Compare alignment attribute.
23663 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23666 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23667 (setb_unsigned) New pattern for setb with CCUNS.
23668 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23669 subfc./subfe sequence to avoid overflow problems. Generate a
23670 shorter sequence with cmpld/setb for power9.
23671 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23672 for generating subfc. instruction.
23673 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23674 now uses this instruction.
23676 2017-01-30 Ian Lance Taylor <iant@google.com>
23679 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23680 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23682 2017-01-30 Martin Sebor <msebor@redhat.com>
23684 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23685 Move constant to the right of a relational operator.
23686 (get_mpfr_format_length, format_character, format_string): Ditto.
23687 (should_warn_p, maybe_warn): Same.
23689 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23691 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23694 * asan.c (get_translation_unit_decl): Remove function.
23695 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23697 2017-01-30 Martin Liska <mliska@suse.cz>
23699 PR gcov-profile/79259
23700 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23701 -fprofile-generate.
23703 2017-01-30 Martin Liska <mliska@suse.cz>
23706 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23707 Initialize variables with NULL value.
23709 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23712 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23714 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23716 2017-01-30 Richard Biener <rguenther@suse.de>
23718 PR tree-optimization/79276
23719 * tree-vrp.c (process_assert_insertions): Properly adjust common
23720 when removing a duplicate.
23722 * gcc.dg/torture/pr79276.c: New testcase.
23724 2017-01-30 Richard Biener <rguenther@suse.de>
23726 PR tree-optimization/79256
23727 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23728 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23730 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23732 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23735 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23736 ("*r<noxa>sbg_<mode>_sll_bitmask")
23737 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23738 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23739 Use contiguous_bitmask_nowrap_operand.
23741 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23744 * config/rs6000/altivec.h (vec_xl): Revise #define.
23745 (vec_xst): Likewise.
23747 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23749 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23751 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23753 PR rtl-optimization/79194
23754 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23755 traps before call to bypass_conditional_jumps.
23757 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23759 PR tree-optimization/71374
23760 * lra-constraints.c (check_conflict_input_operands): New.
23761 (match_reload): Use it.
23763 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23766 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23767 account to calculate conflict_set.
23769 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23771 PR rtl-optimization/78559
23772 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23773 other_insn in combine.
23775 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23777 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23778 uint16_type_node for BT_UINT16.
23780 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23782 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23783 "RTL Tests" to menu.
23784 (GIMPLE Tests): New node.
23785 (RTL Tests): New node.
23787 2017-01-27 Richard Biener <rguenther@suse.de>
23789 PR tree-optimization/79245
23790 * tree-loop-distribution.c (distribute_loop): Apply cost
23791 modeling also to detected patterns.
23793 2017-01-27 Richard Biener <rguenther@suse.de>
23795 PR tree-optimization/71433
23796 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23797 (compare_assert_loc): New function.
23798 (process_assert_insertions): Sort and optimize assert locations
23799 to remove duplicates and push down identical assertions on
23800 edges to their destination block.
23802 2017-01-27 Richard Biener <rguenther@suse.de>
23804 PR tree-optimization/79244
23805 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23806 out SSA names even if abnormal.
23808 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23810 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23811 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23812 instead of MPFR_RNDN.
23814 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23817 * arm.c (arm_option_override): Don't call build_target_option_node
23818 until after doing all option overrides.
23819 (arm_valid_target_attribute_tree): Likewise.
23821 2017-01-27 Martin Liska <mliska@suse.cz>
23823 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23824 for {cd}tors and C++ {cd}tors.
23826 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23828 * config/s390/s390.md ("*setmem_long_and")
23829 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23831 2017-01-26 Martin Sebor <msebor@redhat.com>
23833 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23836 2017-01-26 Martin Sebor <msebor@redhat.com>
23838 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23839 HAVE_DFmode before using XFmode or DFmode.
23840 (parse_directive): Avoid using the z length modifier to avoid
23841 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23843 PR middle-end/78703
23844 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23845 to accept adjustment as an array.
23846 (get_int_range): New function.
23847 (struct directive): Make width and prec arrays.
23848 (directive::set_width, directive::set_precision): Call get_int_range.
23849 (format_integer, format_floating): Handle width and precision ranges.
23850 (format_string, parse_directive): Same.
23852 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23855 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23856 comdat_type_p set, just clone them, but keep the children in the
23860 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23861 which have direct callers with -fvar-tracking-assignments enabled
23863 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23864 inside of type units.
23866 2017-01-26 Martin Sebor <msebor@redhat.com>
23868 PR middle-end/78703
23869 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23871 (struct format_result): Replace number_chars, number_chars_min,
23872 and number_chars_max with a single member of struct result_range.
23874 (format_result::operator+=): Adjust.
23875 (struct fmtresult): Remove bounded. Handle likely and unlikely
23877 (fmtresult::adjust_for_width_or_precision): New function.
23878 (fmtresult:type_max_digits): New function.
23879 (bytes_remaining): Handle likely and unlikely counters.
23880 (min_bytes_remaining): Remove.
23881 (format_percent): Simplify.
23882 (format_integer, format_floating): Set likely and unlikely counters.
23883 (get_string_length, format_character, format_string): Same.
23884 (format_plain, should_warn_p): New function.
23885 (maybe_warn): Call should_warn_p. Update diagnostic messages
23886 and handle those for all directives, including plain strings.
23887 (format_directive): Handle likely and unlikely counters.
23888 Remove unnecessary quoting from diagnostics. Add an informational
23890 (add_bytes): Remove.
23891 (pass_sprintf_length::compute_format_length): Simplify.
23892 (try_substitute_return_value): Handle likely and unlikely counters.
23894 2017-01-26 Carl Love <cel@us.ibm.com>
23896 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23897 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23899 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23902 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23903 endianess for subregs into account.
23904 * lra-constraints.c (lra_constraints): Do risky transformations
23905 always on the first iteration.
23906 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23907 last_call_used_reg_set.
23908 (process_bb_lives): Define and use last_call_used_reg_set.
23909 * lra.c (lra): Always continue after lra_constraints on the first
23912 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23914 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23916 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23918 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23920 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23921 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23922 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23923 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23924 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23925 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23926 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23927 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23928 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23930 2017-01-26 Marek Polacek <polacek@redhat.com>
23933 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23934 for the third operand.
23936 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23938 PR middle-end/79236
23939 * omp-low.c (struct omp_context): Add simt_stmt field.
23940 (scan_omp_for): Return omp_context *.
23941 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23942 context to the _simt_ SIMD stmt.
23943 (lower_omp_for): For combined SIMD with sibling _simt_
23944 SIMD, make sure to use the same decls in _looptemp_
23945 clauses as in the sibling.
23947 2017-01-26 David Sherwood <david.sherwood@arm.com>
23949 PR middle-end/79212
23950 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23953 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23956 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23957 emit fld b; fld a; if possible.
23959 * brig-builtins.def: Update copyright years.
23960 * config/arm/arm_acle_builtins.def: Update copyright years.
23962 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23965 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23966 constraint instead of o for the stxsd instruction.
23968 2017-01-25 Carl Love <cel@us.ibm.com>
23970 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23971 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23973 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23975 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23977 2017-01-25 Richard Biener <rguenther@suse.de>
23979 PR tree-optimization/69264
23980 * target.def (vector_alignment_reachable): Improve documentation.
23981 * doc/tm.texi: Regenerate.
23982 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23984 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23985 earlier changes with respect to TYPE_USER_ALIGN.
23986 (vector_alignment_reachable_p): Likewise. Improve dumping.
23988 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23991 * config/arm/arm.md (xordi3): Force constant operand into a register
23994 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23996 * doc/invoke.texi (-fstore-merging): Correct default optimization
23997 levels at which it is enabled.
23998 (-O): Move -fstore-merging from list to...
24001 2017-01-25 Richard Biener <rguenther@suse.de>
24004 * omp-expand.c: Include debug.h.
24005 (expand_omp_taskreg): Make sure to generate early debug before
24006 outlining anything from a function.
24007 (expand_omp_target): Likewise.
24008 (grid_expand_target_grid_body): Likewise.
24010 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24013 * asan.c (get_translation_unit_decl): New function.
24014 (asan_add_global): Extract modules file name from globals
24015 TRANSLATION_UNIT_DECL name.
24017 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24020 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24021 for long calls with APCS frame and VFP.
24023 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24025 * cfg.c (original_copy_tables_initialized_p): New function.
24026 * cfg.h (original_copy_tables_initialized_p): New decl.
24027 * cfgrtl.c (relink_block_chain): Guard the call to
24028 free_original_copy_tables with a call to
24029 original_copy_tables_initialized_p.
24030 * cgraph.h (symtab_node::native_rtl_p): New decl.
24031 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24032 (symtab_node::needed_p): Don't assert for early assembly output
24033 for __RTL functions.
24034 (cgraph_node::finalize_function): Set "force_output" for __RTL
24036 (cgraph_node::analyze): Bail out early for __RTL functions.
24037 (analyze_functions): Update assertion to support __RTL functions.
24038 (cgraph_node::expand): Bail out early for __RTL functions.
24039 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24041 * function.h (struct function): Update comment for field
24043 * gimple-expr.c: Include "tree-pass.h".
24044 (gimple_has_body_p): Return false for __RTL functions.
24045 * Makefile.in (OBJS): Add run-rtl-passes.o.
24046 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24048 (gcc::pass_manager::get_clean_slate): New accessor.
24049 * passes.c: Include "insn-addr.h".
24050 (should_skip_pass_p): Add logging. Update logic for running
24051 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24052 property-provider override so it is only done for gimple passes.
24054 (skip_pass): New function.
24055 (execute_one_pass): Call skip_pass when skipping passes.
24056 * read-md.c (md_reader::read_char): Support filtering
24057 the input to a subset of line numbers.
24058 (md_reader::md_reader): Initialize fields
24059 m_first_line and m_last_line.
24060 (md_reader::read_file_fragment): New function.
24061 * read-md.h (md_reader::read_file_fragment): New decl.
24062 (md_reader::m_first_line): New field.
24063 (md_reader::m_last_line): New field.
24064 * read-rtl-function.c (function_reader::create_function): Only
24065 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24066 curr_properties. Set DECL_INITIAL to a dummy block.
24067 (read_rtl_function_body_from_file_range): New function.
24068 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24070 * run-rtl-passes.c: New file.
24071 * run-rtl-passes.h: New file.
24073 2017-01-24 Jeff Law <law@redhat.com>
24075 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24078 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24080 PR tree-optimization/79159
24081 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24082 (record_nonwrapping_iv): Improve boundary using above function if no
24083 value range information.
24085 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24086 Martin Jambor <mjambor@suse.cz>
24088 * brig-builtins.def: New file.
24089 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24090 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24091 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24092 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24093 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24094 * builtin-types.def (BT_INT8): New.
24095 (BT_INT16): Likewise.
24096 (BT_UINT8): Likewise.
24097 (BT_UINT16): Likewise.
24098 (BT_FN_ULONG): Likewise.
24099 (BT_FN_UINT_INT): Likewise.
24100 (BT_FN_UINT_ULONG): Likewise.
24101 (BT_FN_UINT_LONG): Likewise.
24102 (BT_FN_UINT_PTR): Likewise.
24103 (BT_FN_ULONG_PTR): Likewise.
24104 (BT_FN_INT8_FLOAT): Likewise.
24105 (BT_FN_INT16_FLOAT): Likewise.
24106 (BT_FN_UINT32_FLOAT): Likewise.
24107 (BT_FN_UINT16_FLOAT): Likewise.
24108 (BT_FN_UINT8_FLOAT): Likewise.
24109 (BT_FN_UINT64_FLOAT): Likewise.
24110 (BT_FN_UINT16_UINT32): Likewise.
24111 (BT_FN_UINT32_UINT16): Likewise.
24112 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24113 (BT_FN_INT_PTR_INT): Likewise.
24114 (BT_FN_UINT_PTR_UINT): Likewise.
24115 (BT_FN_LONG_PTR_LONG): Likewise.
24116 (BT_FN_ULONG_PTR_ULONG): Likewise.
24117 (BT_FN_VOID_UINT64_UINT64): Likewise.
24118 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24119 (BT_FN_INT8_INT8_INT8): Likewise.
24120 (BT_FN_INT16_INT16_INT16): Likewise.
24121 (BT_FN_INT_INT_INT): Likewise.
24122 (BT_FN_UINT_FLOAT_UINT): Likewise.
24123 (BT_FN_FLOAT_UINT_UINT): Likewise.
24124 (BT_FN_ULONG_UINT_UINT): Likewise.
24125 (BT_FN_ULONG_UINT_PTR): Likewise.
24126 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24127 (BT_FN_UINT_UINT_UINT): Likewise.
24128 (BT_FN_VOID_UINT_PTR): Likewise.
24129 (BT_FN_UINT_UINT_PTR: Likewise.
24130 (BT_FN_UINT32_UINT64_PTR): Likewise.
24131 (BT_FN_INT_INT_UINT_UINT): Likewise.
24132 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24133 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24134 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24135 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24136 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24137 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24138 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24139 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24140 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24141 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24142 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24143 * doc/frontends.texi: List BRIG FE.
24144 * doc/install.texi (Testing): Add BRIG tesring requirements.
24145 * doc/invoke.texi (Overall Options): Mention BRIG.
24146 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24148 2017-01-24 Richard Biener <rguenther@suse.de>
24150 PR translation/79208
24151 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24153 2017-01-24 Martin Jambor <mjambor@suse.cz>
24156 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24157 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24158 and known_contexts.
24160 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24162 PR middle-end/79123
24163 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24164 casts from signed to unsigned really don't have a range.
24166 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24168 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24169 GMP_RNDx for compatiblity.
24171 2017-01-24 Martin Liska <mliska@suse.cz>
24174 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24175 that would prevent us to call alloca with -1 as argument.
24177 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24179 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24180 Avoid -Wformat-security warning.
24182 2017-01-23 Andrew Pinski <apinski@cavium.com>
24184 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24187 2017-01-23 Martin Sebor <msebor@redhat.com>
24189 PR middle-end/78703
24190 * gimple-ssa-sprintf.c (warn_level): New global.
24191 (format_integer): Use it here and throughout the rest of the file.
24192 Use the same switch to compute sign as base.
24193 (maybe_warn): New function.
24194 (format_directive): Factor out warnings into maybe_warn.
24195 Add debugging output. Use warn_level.
24196 (add_bytes): Use warn_level.
24197 (pass_sprintf_length::compute_format_length): Add debugging output.
24198 (try_substitute_return_value): Same.
24199 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24201 PR middle-end/78703
24202 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24203 (struct fmtresult, format_integer, format_floating): Adjust.
24204 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24205 (get_string_length, format_string,format_directive): Same.
24206 (pass_sprintf_length::compute_format_length): Same.
24207 (try_substitute_return_value): Simplify slightly.
24209 PR middle-end/78703
24210 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24211 (fmtresult::operator+=): Outlined.
24212 (struct fmtresult): Add ctors.
24213 (struct conversion_spec): Rename...
24214 (struct directive): ...to this. Add and remove data members.
24215 (directive::set_width, directive::set_precision): New functions.
24216 (format_percent): Use fmtresult ctor.
24217 (get_width_and_precision): Remove.
24218 (format_integer): Make naming changes. Avoid computing width and
24220 (format_floating): Same. Adjust indentation.
24221 (format_character, format_none): New functions.
24222 (format_string): Moved character handling to format_character.
24223 (format_directive): Remove arguments, change return type.
24224 (parse_directive): New function.
24225 (pass_sprintf_length::compute_format_length): Move directive
24226 parsing to parse_directive.
24228 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24230 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24231 (assign_assembler_name_if_needed): ... this.
24232 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24233 (assign_assembler_name_if_needed): ... this.
24234 (free_lang_data_in_cgraph): Adjust callers.
24235 * cgraphunit.c (cgraph_node::analyze): Likewise.
24236 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24239 2017-01-23 Richard Biener <rguenther@suse.de>
24241 PR tree-optimization/79088
24242 PR tree-optimization/79188
24243 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24244 resetting loop bounds after last path deletion. Reset loop
24245 bounds of the target loop, make code match the comments.
24246 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24247 Make sure loops need no fixups.
24249 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24251 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24252 exponent support with double type for first argument.
24253 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24254 type returned by __builtin_vec_extract_sig,
24255 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24256 functions from "vector int" to "vector unsigned int" or from
24257 "vector long long int" to "vector unsigned long long int".
24258 Changed type returned by __builtin_vec_extract_exp,
24259 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24260 functions from "vector int" to "vector unsigned int" or from
24261 "vector long long int" to "vector unsigned long long int".
24262 Changed return type of __builtin_vec_test_data_class,
24263 __builtin_vec_test_data_class_sp, and
24264 __builtin_vec_test_data_class_dp from "vector int" to
24265 "vector bool int" or from "vector long long int" to "vector bool
24266 long long int" and changed second argument type from "unsigned
24267 int" to "int". Added new overloaded function forms "vector float
24268 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24269 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24270 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24271 double, vector unsigned long long int)" and "vector double
24272 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24273 long int)". Changed return type of
24274 __builtin_scalar_test_data_class and
24275 __builtin_scalar_test_data_class_sp and
24276 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24277 int" and changed second argument from "unsigned int" to "int".
24278 Changed type returned by __builtin_scalar_test_neg,
24279 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24280 from "int" to "bool int". Added new overloaded function form
24281 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24282 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24283 exponent double-precision with floating point first argument.
24284 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24285 documentation of scalar_test_data_class, scalar_test_neg,
24286 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24287 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24288 vec_test_data_class built-in functions to reflect refinements in
24289 their type signatures.
24291 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24293 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24295 (aarch64_elf_asm_destructor): Likewise.
24297 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24299 PR rtl-optimization/78634
24300 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24301 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24302 * ifcvt.c (noce_try_cmove): Add missing cost check.
24304 PR rtl-optimization/71724
24305 * combine.c (if_then_else_cond): Look for situations where it is
24306 beneficial to undo the work of one of the recursive calls.
24308 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24310 PR tree-optimization/70754
24311 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24312 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24313 combined stmt before it if not NULL.
24314 (combine_chains): Process refs reversely and compute dominance point
24317 2017-01-23 Martin Liska <mliska@suse.cz>
24319 PR tree-optimization/79196
24320 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24321 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24323 (strlen_optimize_stmt): Call the renamed function.
24325 2017-01-23 Michael Matz <matz@suse.de>
24327 PR tree-optimization/78384
24328 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24330 2017-01-23 Richard Biener <rguenther@suse.de>
24332 PR tree-optimization/79186
24333 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24334 both incoming edges before moving an assert.
24336 2017-01-23 Martin Jambor <mjambor@suse.cz>
24338 * ipa-prop.c (load_from_param_1): Removed.
24339 (load_from_unmodified_param): Bits from load_from_param_1 put back
24341 (load_from_param): Removed.
24342 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24343 with stmt. Reverted back to use of load_from_unmodified_param.
24345 2017-01-23 Martin Jambor <mjambor@suse.cz>
24348 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24349 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24350 field a pointer to garbage collected vector, mark lattices and
24351 ipcp_orig_node with GTY((skip)).
24352 (ipa_get_param_count): Adjust to descriptors being a pointer.
24353 (ipa_get_param): Likewise.
24354 (ipa_get_type): Likewise.
24355 (ipa_get_param_move_cost): Likewise.
24356 (ipa_set_param_used): Likewise.
24357 (ipa_get_controlled_uses): Likewise.
24358 (ipa_set_controlled_uses): Likewise.
24359 (ipa_is_param_used): Likewise.
24360 (ipa_node_params_t): Move into garbage collector. New methods insert
24362 (ipa_node_params_sum): Annotate wth GTY(()).
24363 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24365 (ipa_load_from_parm_agg): Adjust declaration.
24366 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24367 * ipa-profile.c (ipa_profile): Likewise.
24368 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24369 (ipa_populate_param_decls): Make descriptors parameter garbage
24371 (ipa_dump_param): Adjust to descriptors being a pointer.
24372 (ipa_alloc_node_params): Likewise.
24373 (ipa_initialize_node_params): Likewise.
24374 (load_from_param_1): Make descriptors parameter garbage collected.
24375 (load_from_unmodified_param): Likewise.
24376 (load_from_param): Likewise.
24377 (ipa_load_from_parm_agg): Likewise.
24378 (ipa_node_params::~ipa_node_params): Removed.
24379 (ipa_free_all_node_params): Remove call to delete operator.
24380 (ipa_node_params_t::insert): New.
24381 (ipa_node_params_t::remove): Likewise.
24382 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24383 copy known_csts and known_contexts vectors.
24384 (ipa_read_node_info): Adjust to descriptors being a pointer.
24385 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24387 (ipcp_transform_function): Make descriptors variable garbage
24390 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24392 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24393 * config/i386/avx512dqintrin.h: Ditto.
24394 * config/i386/avx512fintrin.h: Ditto.
24395 * config/i386/i386.c: Handle new builtins.
24396 * config/i386/i386-builtin.def: Add new builtins.
24397 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24398 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24400 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24401 Martin Liska <mliska@suse.cz>
24403 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24404 * asan.c (asan_expand_poison_ifn): Support stores and use
24405 appropriate ASAN report function.
24406 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24407 * internal-fn.def (ASAN_POISON_USE): Declare.
24408 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24409 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24410 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24411 ASAN_POISON calls w/o LHS.
24412 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24413 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24414 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24415 * gimplify.c (asan_poison_variables): Add attribute
24416 use_after_scope_memory to variables that really needs to live
24418 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24419 having the attribute.
24421 2017-01-23 Martin Liska <mliska@suse.cz>
24423 * asan.c (create_asan_shadow_var): New function.
24424 (asan_expand_poison_ifn): Likewise.
24425 * asan.h (asan_expand_poison_ifn): New declaration.
24426 * internal-fn.c (expand_ASAN_POISON): Likewise.
24427 * internal-fn.def (ASAN_POISON): New builtin.
24428 * sanopt.c (pass_sanopt::execute): Expand
24429 asan_expand_poison_ifn.
24430 * tree-inline.c (copy_decl_for_dup_finish): Make function
24432 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24433 * tree-ssa.c (is_asan_mark_p): New function.
24434 (execute_update_addresses_taken): Rewrite local variables
24435 (identified just by use-after-scope as addressable) into SSA.
24437 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24439 * doc/install.texi (Specific): opensource.apple.com uses https
24440 now. Remove trailing slash.
24442 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24444 * README.Portability: Remove note on an Irix compatibility issue.
24446 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24448 * gcov.c (INCLUDE_ALGORITHM): Define.
24449 (INCLUDE_VECTOR): Define.
24450 No longer include <vector> and <algorithm> directly.
24452 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24454 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24456 * doc/invoke.texi (Code Gen Options): Ditto.
24458 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24461 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24463 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24465 rtl-optimization/79125
24466 * cprop.c (local_cprop_pass): Handle cases where we make an
24467 unconditional trap.
24469 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24473 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24474 read from, for big endian.
24476 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24478 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24479 register pauth builtins for LP64 only.
24481 2017-01-20 Marek Polacek <polacek@redhat.com>
24484 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24487 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24489 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24491 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24492 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24493 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24495 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24498 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24499 in big-endian mode when they are not a single duplicated value.
24501 2017-01-20 Richard Biener <rguenther@suse.de>
24503 * BASE-VER: Bump to 7.0.1.
24505 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24507 * omp-low.c (omplow_simd_context): New struct. Use it...
24508 (lower_rec_simd_input_clauses): ...here and...
24509 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24510 references to idx, lane, max_vf, is_simt.
24512 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24514 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24517 2017-01-20 Martin Jambor <mjambor@suse.cz>
24519 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24520 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24522 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24523 (GTFILES): Rename hsa.c to hsa-common.c.
24524 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24525 * hsa-dump.c: Likewise.
24526 * hsa-gen.c: Likewise.
24527 * hsa-regalloc.c: Likewise.
24528 * ipa-hsa.c: Likewise.
24529 * omp-expand.c: Likewise.
24530 * omp-low.c: Likewise.
24531 * toplev.c: Likewise.
24533 2017-01-20 Marek Polacek <polacek@redhat.com>
24536 * doc/invoke.texi: Document -Wduplicated-branches.
24537 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24538 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24539 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24540 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24541 return 0 only when not OEP_LEXICOGRAPHIC.
24542 (fold_build_cleanup_point_expr): Use the expression
24543 location when building CLEANUP_POINT_EXPR.
24544 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24545 * tree.c (add_expr): Handle error_mark_node.
24547 2017-01-20 Martin Liska <mliska@suse.cz>
24550 * tree-profile.c (init_ic_make_global_vars): Do not call
24552 (gimple_init_gcov_profiler): Likewise.
24554 2017-01-20 Martin Liska <mliska@suse.cz>
24557 * cgraph.h (maybe_create_reference): Remove argument and
24559 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24561 * ipa-cp.c (create_specialized_node): Likewise.
24562 * symtab.c (symtab_node::maybe_create_reference): Handle
24563 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24565 2017-01-20 Martin Liska <mliska@suse.cz>
24567 * read-rtl-function.c (function_reader::create_function): Use
24568 build_decl instread of build_decl_stat.
24570 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24572 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24573 * config/i386/avx512dqintrin.h: Ditto.
24574 * config/i386/avx512fintrin.h: Ditto.
24575 * config/i386/i386-builtin-types.def: Add new types.
24576 * config/i386/i386.c: Handle new types.
24577 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24578 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24579 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24580 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24581 (__builtin_ia32_kshiftridi): New.
24582 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24584 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24588 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24589 define to rs6000_init_stack_protect_guard.
24590 (rs6000_init_stack_protect_guard): New function.
24592 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24593 Yunqiang Su <yunqiang.su@imgtec.com>
24595 * config.gcc (supported_defaults): Add madd4.
24596 (with_madd4): Add validation.
24597 (all_defaults): Add madd4.
24598 * config/mips/mips.opt (mmadd4): New option.
24599 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24601 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24603 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24604 (ISA_HAS_FUSED_MADD4): Likewise.
24605 * doc/invoke.texi (-mmadd4): Document the new option.
24606 * doc/install.texi (--with-madd4): Document the new option.
24608 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24610 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24611 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24612 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24613 (aarch64_init_pauth_hint_builtins): New.
24614 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24615 (aarch64_expand_builtin): Expand new builtins.
24617 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24619 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24620 * combine-stack-adj.c (no_unhandled_cfa): Handle
24621 REG_CFA_TOGGLE_RA_MANGLE.
24622 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24623 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24624 info for return address signing.
24625 (aarch64_expand_epilogue): Likewise.
24627 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24629 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24630 * config/aarch64/aarch64-protos.h
24631 (aarch64_return_address_signing_enabled): New declaration.
24632 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24634 (aarch64_expand_prologue): Sign return address before it's pushed onto
24636 (aarch64_expand_epilogue): Authenticate return address fetched from
24638 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24639 (aarch64_attributes): New function attributes for "sign-return-address".
24640 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24641 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24642 ("*do_return"): Generate combined instructions according to key index.
24643 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24644 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24646 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24647 * config/aarch64/aarch64.opt (msign-return-address=): New.
24648 * doc/extend.texi (AArch64 Function Attributes): Documents
24649 "sign-return-address=".
24650 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24652 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24654 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24655 overall option summary.
24657 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24659 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24660 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24661 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24662 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24664 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24666 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24667 -mpower9-minmax by default for -mcpu=power9.
24668 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24669 128-bit floating point.
24671 2017-01-20 Alan Modra <amodra@gmail.com>
24673 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24674 optimizing for size.
24676 2017-01-20 Alan Modra <amodra@gmail.com>
24679 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24680 for strcmp and strncmp from corresponding builtin decl.
24682 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24684 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24685 instead of i386/rtems-64.h.
24686 * config/i386/rtems-64.h: Remove.
24688 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24692 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24694 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24696 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24698 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24699 Change int to HOST_WIDE_INT.
24700 * config/aarch64/aarch64-protos.h
24701 (aarch64_simd_gen_const_vector_dup): Likewise.
24702 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24704 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24706 * langhooks-def.h (lhd_type_for_size): New decl.
24707 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24708 * langhooks.c (lhd_type_for_size): New function, taken from
24711 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24713 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24714 define_bypass for CR latency.
24715 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24716 (power9-alu2): Add define_bypass for CR latency.
24718 (power9-mul): Update insn latency.
24719 (power9-mul-compare): Update insn latency, bypass latency and remove
24722 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24724 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24726 * config/aarch64/aarch64.md
24727 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24728 aarch64_nopcrelative_literal_loads.
24729 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24731 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24733 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24734 TARGET_LOONGSON_3A.
24735 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24737 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24740 * config.gcc (supported_defaults): Add lxc1-sxc1.
24741 (with_lxc1_sxc1): Add validation.
24742 (all_defaults): Add lxc1-sxc1.
24743 * config/mips/mips.opt (mlxc1-sxc1): New option.
24744 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24746 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24747 __mips_no_lxc1_sxc1.
24748 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24749 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24750 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24752 2017-01-19 Richard Biener <rguenther@suse.de>
24754 PR tree-optimization/72488
24755 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24756 sure to restore SSA info.
24757 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24759 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24761 PR rtl-optimization/79121
24762 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24763 of the inner type when shifting an extended value.
24765 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24768 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24769 interposable aliases.
24771 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24774 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24775 Use the evmergelohi instruction.
24776 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24777 (mov_sitf_e500_subreg8_2_be): Likewise.
24778 (mov_sitf_e500_subreg12_2_le): Likewise.
24779 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24780 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24781 (mov_sitf_e500_subreg8_2_le): Likewise.
24782 (mov_sitf_e500_subreg12_2_be): Likewise.
24784 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24786 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24787 attribute from vecsimple to vecperm.
24788 (altivec_vbpermq2): Likewise.
24790 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24793 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24795 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24796 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24797 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24798 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24799 case where N arg is SIZE_MAX.
24800 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24801 (cmpstrsi): Add pattern.
24803 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24805 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24806 __builtin_vec_revb builtins.
24807 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24808 built-in functions to support generation of the ISA 3.0 XXBR<x>
24809 vector byte reverse instructions.
24810 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24811 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24812 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24813 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24814 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24815 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24816 (P9V_BUILTIN_VEC_REVB): Likewise.
24817 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24818 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24819 (p9_xxbrq_v16qi): Likewise.
24820 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24821 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24822 (p9_xxbrh_v8hi): Likewise.
24823 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24824 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24825 vec_revb built-in functions.
24827 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24829 PR rtl-optimization/78952
24830 * config/i386/i386.md (any_extract): New code iterator.
24831 (*insvqi_2): Use any_extract for source operand.
24832 (*insvqi_3): Use any_shiftrt for source operand.
24834 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24836 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24838 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24840 2017-01-18 Matthias Klose <doko@ubuntu.com>
24842 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24844 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24846 * config/rs6000/altivec.h (vec_bperm): Change #define.
24847 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24848 (altivec_vbpermq2): New define_insn.
24849 (altivec_vbpermd): Likewise.
24850 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24851 function interface.
24852 (VBPERMD): Likewise.
24853 (VBPERM): New polymorphic function interface.
24854 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24855 Add entries for P9V_BUILTIN_VEC_VBPERM.
24856 * doc/extend.texi: Add interfaces for vec_bperm.
24858 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24860 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24861 first letter of error messages.
24862 (s390_resolve_overloaded_builtin): Likewise.
24863 * config/s390/s390.c (s390_expand_builtin): Likewise.
24864 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24865 (s390_valid_target_attribute_inner_p): Likewise.
24866 * config/s390/s390.md ("tabort"): Likewise.
24868 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24870 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24871 (ISA_AVOID_DIV_HILO): New macro.
24872 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24873 (ISA_HAS_DDIV): Likewise.
24875 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24877 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24879 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24881 * doc/invoke.texi (fabi-version): Spelling fix.
24883 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24886 * doc/invoke.texi (fabi-version): Mention mangling fix for
24889 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24892 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24894 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24897 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24898 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24900 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24902 (stack_protect_test): Ditto.
24903 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24904 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24906 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24907 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24908 -mstack-protector-guard-offset=.
24909 (RS/6000 and PowerPC Options): Ditto.
24911 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24913 * config/i386/i386.h (MASK_CLASS_P): New define.
24914 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24915 there are no registers from different register sets also when
24916 mask registers are used. Update function comment.
24917 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24918 to (*k/*r) and (*k/*km) alternatives.
24920 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24922 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24923 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24924 (EH_RETURN_HANDLER_RTX): New define.
24925 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24926 Force frame pointer in EH return functions.
24927 (aarch64_expand_epilogue): Add barrier for eh_return.
24928 (aarch64_final_eh_return_addr): Remove.
24929 (aarch64_eh_return_handler_rtx): New function.
24930 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24932 (aarch64_eh_return_handler_rtx): New prototype.
24934 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24936 * config/rs6000/altivec.h (vec_rlmi): New #define.
24937 (vec_vrlnm): Likewise.
24938 (vec_rlnm): Likewise.
24939 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24940 (UNSPEC_VRLNM): Likewise.
24941 (VIlong): New mode iterator.
24942 (altivec_vrl<VI_char>mi): New define_insn.
24943 (altivec_vrl<VI_char>nm): Likewise.
24944 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24946 (VRLDNM): Likewise.
24947 (RLNM): New polymorphic function entry.
24948 (VRLWMI): New monomorphic function entry.
24949 (VRLDMI): Likewise.
24950 (RLMI): New polymorphic function entry.
24951 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24952 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24953 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24956 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24959 * dwarf2out.c (field_byte_offset): Restore the
24960 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24961 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24962 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24965 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24968 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24970 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24973 * dwarf2out.c (add_data_member_location_attribute): For constant
24974 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24975 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24976 DW_AT_byte_size attributes.
24978 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24980 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24981 after forcing to constant memory when the code model is medium.
24983 2017-01-17 Julia Koval <julia.koval@intel.com>
24986 * config/i386/avx512fintrin.h
24987 (_mm512_i32gather_ps): Change __addr type to void const*.
24988 (_mm512_mask_i32gather_ps): Ditto.
24989 (_mm512_i32gather_pd): Ditto.
24990 (_mm512_mask_i32gather_pd): Ditto.
24991 (_mm512_i64gather_ps): Ditto.
24992 (_mm512_mask_i64gather_ps): Ditto.
24993 (_mm512_i64gather_pd): Ditto.
24994 (_mm512_mask_i64gather_pd): Ditto.
24995 (_mm512_i32gather_epi32): Ditto.
24996 (_mm512_mask_i32gather_epi32): Ditto.
24997 (_mm512_i32gather_epi64): Ditto.
24998 (_mm512_mask_i32gather_epi64): Ditto.
24999 (_mm512_i64gather_epi32): Ditto.
25000 (_mm512_mask_i64gather_epi32): Ditto.
25001 (_mm512_i64gather_epi64): Ditto.
25002 (_mm512_mask_i64gather_epi64): Ditto.
25003 (_mm512_i32scatter_ps): Change __addr type to void*.
25004 (_mm512_mask_i32scatter_ps): Ditto.
25005 (_mm512_i32scatter_pd): Ditto.
25006 (_mm512_mask_i32scatter_pd): Ditto.
25007 (_mm512_i64scatter_ps): Ditto.
25008 (_mm512_mask_i64scatter_ps): Ditto.
25009 (_mm512_i64scatter_pd): Ditto.
25010 (_mm512_mask_i64scatter_pd): Ditto.
25011 (_mm512_i32scatter_epi32): Ditto.
25012 (_mm512_mask_i32scatter_epi32): Ditto.
25013 (_mm512_i32scatter_epi64): Ditto.
25014 (_mm512_mask_i32scatter_epi64): Ditto.
25015 (_mm512_i64scatter_epi32): Ditto.
25016 (_mm512_mask_i64scatter_epi32): Ditto.
25017 (_mm512_i64scatter_epi64): Ditto.
25018 (_mm512_mask_i64scatter_epi64): Ditto.
25019 * config/i386/avx512pfintrin.h
25020 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25021 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25022 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25023 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25024 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25025 (_mm512_prefetch_i32scatter_ps): Ditto.
25026 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25027 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25028 (_mm512_prefetch_i64scatter_pd): Ditto.
25029 (_mm512_prefetch_i64scatter_ps): Ditto.
25030 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25031 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25032 * config/i386/avx512vlintrin.h
25033 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25034 (_mm_mmask_i32gather_ps): Ditto.
25035 (_mm256_mmask_i32gather_pd): Ditto.
25036 (_mm_mmask_i32gather_pd): Ditto.
25037 (_mm256_mmask_i64gather_ps): Ditto.
25038 (_mm_mmask_i64gather_ps): Ditto.
25039 (_mm256_mmask_i64gather_pd): Ditto.
25040 (_mm_mmask_i64gather_pd): Ditto.
25041 (_mm256_mmask_i32gather_epi32): Ditto.
25042 (_mm_mmask_i32gather_epi32): Ditto.
25043 (_mm256_mmask_i32gather_epi64): Ditto.
25044 (_mm_mmask_i32gather_epi64): Ditto.
25045 (_mm256_mmask_i64gather_epi32): Ditto.
25046 (_mm_mmask_i64gather_epi32): Ditto.
25047 (_mm256_mmask_i64gather_epi64): Ditto.
25048 (_mm_mmask_i64gather_epi64): Ditto.
25049 (_mm256_i32scatter_ps): Change __addr type to void*.
25050 (_mm256_mask_i32scatter_ps): Ditto.
25051 (_mm_i32scatter_ps): Ditto.
25052 (_mm_mask_i32scatter_ps): Ditto.
25053 (_mm256_i32scatter_pd): Ditto.
25054 (_mm256_mask_i32scatter_pd): Ditto.
25055 (_mm_i32scatter_pd): Ditto.
25056 (_mm_mask_i32scatter_pd): Ditto.
25057 (_mm256_i64scatter_ps): Ditto.
25058 (_mm256_mask_i64scatter_ps): Ditto.
25059 (_mm_i64scatter_ps): Ditto.
25060 (_mm_mask_i64scatter_ps): Ditto.
25061 (_mm256_i64scatter_pd): Ditto.
25062 (_mm256_mask_i64scatter_pd): Ditto.
25063 (_mm_i64scatter_pd): Ditto.
25064 (_mm_mask_i64scatter_pd): Ditto.
25065 (_mm256_i32scatter_epi32): Ditto.
25066 (_mm256_mask_i32scatter_epi32): Ditto.
25067 (_mm_i32scatter_epi32): Ditto.
25068 (_mm_mask_i32scatter_epi32): Ditto.
25069 (_mm256_i32scatter_epi64): Ditto.
25070 (_mm256_mask_i32scatter_epi64): Ditto.
25071 (_mm_i32scatter_epi64): Ditto.
25072 (_mm_mask_i32scatter_epi64): Ditto.
25073 (_mm256_i64scatter_epi32): Ditto.
25074 (_mm256_mask_i64scatter_epi32): Ditto.
25075 (_mm_i64scatter_epi32): Ditto.
25076 (_mm_mask_i64scatter_epi32): Ditto.
25077 (_mm256_i64scatter_epi64): Ditto.
25078 (_mm256_mask_i64scatter_epi64): Ditto.
25079 (_mm_i64scatter_epi64): Ditto.
25080 (_mm_mask_i64scatter_epi64): Ditto.
25081 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25082 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25083 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25084 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25085 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25086 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25087 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25088 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25089 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25090 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25091 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25092 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25093 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25094 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25095 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25096 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25097 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25098 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25099 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25100 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25101 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25102 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25103 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25104 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25105 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25106 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25107 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25108 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25109 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25110 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25111 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25112 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25113 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25114 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25115 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25116 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25117 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25118 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25119 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25120 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25121 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25122 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25123 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25124 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25125 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25126 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25127 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25128 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25129 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25130 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25131 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25132 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25133 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25134 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25135 definitions accordingly.
25137 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25138 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25141 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25144 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25147 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25148 hard regs for inner regno.
25150 2017-01-17 Martin Liska <mliska@suse.cz>
25153 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25154 assumption and add comment.
25156 2017-01-17 Nathan Sidwell <nathan@acm.org>
25158 * ipa-visibility.c (localize_node): New function, broken out of ...
25159 (function_and_variable_visibility): ... here. Call it.
25161 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25163 PR middle-end/77445
25164 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25165 correctly set frequency of oudgoing edge.
25166 (duplicate_thread_path): Fix profile updating.
25168 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25171 * configure.ac: Add GCC_BASE_VER.
25172 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25173 version from BASE-VER file.
25174 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25175 (gcc.o): Depend on $(BASEVER).
25176 * common.opt (dumpfullversion): New option.
25177 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25178 * doc/invoke.texi: Document -dumpfullversion.
25179 * doc/install.texi: Document --with-gcc-major-version-only.
25180 * configure: Regenerated.
25182 2017-01-17 Richard Biener <rguenther@suse.de>
25184 PR tree-optimization/71433
25185 * tree-vrp.c (register_new_assert_for): Merge same asserts
25186 on all incoming edges.
25187 (process_assert_insertions_for): Handle insertions at the
25190 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25192 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25193 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25195 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25198 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25201 2017-01-17 Alan Modra <amodra@gmail.com>
25204 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25205 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25206 symbolic stack limit when pic.
25208 2017-01-16 Martin Sebor <msebor@redhat.com>
25210 PR tree-optimization/78608
25211 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25213 2017-01-16 Jeff Law <law@redhat.com>
25216 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25217 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25218 for several include directories that may be relative to sysroot.
25219 * config/i386/x-mingw32 (gplus_includedir): Define.
25220 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25221 (native_system_includedir): Likewise.
25222 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25223 override if TARGET_SYSTEM_ROOT is defined.
25224 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25226 PR tree-optimization/79090
25227 PR tree-optimization/33562
25228 PR tree-optimization/61912
25229 PR tree-optimization/77485
25230 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25231 and computed trims into the dump file.
25233 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25235 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25237 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25240 * gimplify.c (gimplify_init_constructor): If want_value and
25241 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25245 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25246 sequence. Formatting fixes.
25247 (doloop_optimize): Formatting fixes.
25250 * gcc.c (debug_level_greater_than_spec_func): New function.
25251 (static_spec_functions): Add debug-level-gt spec function.
25252 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25254 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25255 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25256 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25257 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25258 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25259 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25261 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25263 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25264 QImode fixups to general and mask registers only.
25266 2017-01-16 Carl Love <cel@us.ibm.com>
25268 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25269 for built-in functions
25270 vector signed char vec_nabs (vector signed char)
25271 vector signed short vec_nabs (vector signed short)
25272 vector signed int vec_nabs (vector signed int)
25273 vector signed long long vec_nabs (vector signed long long)
25274 vector float vec_nabs (vector float)
25275 vector double vec_nabs (vector double)
25276 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25278 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25279 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25280 * doc/extend.texi: Update the documentation file for the new built-in
25283 2017-01-16 Martin Sebor <msebor@redhat.com>
25285 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25288 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25290 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25291 UNSPEC_VSX__XXSPLTD to require special splat handling.
25293 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25296 * system.h: Poison strndup.
25298 2017-01-16 Alan Modra <amodra@gmail.com>
25301 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25304 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25306 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25308 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25310 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25311 call recog here. Assert that INSN_CODE (insn) is non-negative.
25313 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25316 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25318 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25319 in the currently scheduled RTL fragment.
25321 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25323 PR rtl-optimization/78751
25324 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25327 2017-01-14 Jeff Law <law@redhat.com>
25329 PR tree-optimization/79090
25330 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25331 variable length stores.
25332 (compute_trims): Delete dead assignment to *trim_tail.
25333 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25336 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25338 PR rtl-optimization/78626
25339 PR rtl-optimization/78727
25340 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25341 of a block, and split such blocks after everything else is finished.
25343 2017-01-14 Alan Modra <amodra@gmail.com>
25346 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25347 target legitimate_combined_insn.
25348 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25349 (rs6000_legitimate_combined_insn): New function.
25350 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25352 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25353 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25354 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25356 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25358 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25360 2017-01-13 Jeff Law <law@redhat.com>
25362 PR tree-optimization/33562
25363 PR tree-optimization/61912
25364 PR tree-optimization/77485
25365 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25367 (delete_dead_assignment): Likewise.
25368 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25369 statement to delete_dead_call and delete_dead_assignment.
25371 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25374 * substring-locations.c (format_warning_va): Strengthen case 1 so
25375 that both endpoints of the substring must be within the format
25376 range for just the substring to be printed.
25378 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25380 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25381 * config/i386/i386.c (ix86_target_string): Add missing options
25382 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25383 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25384 flags_other and ix86_target_other to flags2_other. Display unknown
25386 (ix86_valid_target_attribute_inner_p): Add missing options and
25387 reorder options by implied ISAs, as in ix86_target_string.
25389 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25391 * hash-table.h (hash_table::too_empty_p): New function.
25392 (hash_table::expand): Use it.
25393 (hash_table::traverse): Likewise.
25394 (hash_table::empty_slot): Use sizeof (value_type) instead of
25395 sizeof (PTR) to convert bytes to elements. Shrink the table
25396 if the current size is excessive for the current number of
25399 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25401 * ira-costs.c (record_reg_classes): Break from the inner loop
25402 early once alt_fail is known to be true. Update outer loop
25403 handling accordingly.
25405 2017-01-13 Jeff Law <law@redhat.com>
25407 * tree-ssa-dse.c (decrement_count): New function.
25408 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25409 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25410 when we know the partially dead statement is a mem* function.
25412 PR tree-optimization/61912
25413 PR tree-optimization/77485
25414 * tree-ssa-dse.c: Include expr.h.
25415 (maybe_trim_constructor_store): New function.
25416 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25418 PR tree-optimization/33562
25419 PR tree-optimization/61912
25420 PR tree-optimization/77485
25421 * doc/invoke.texi: Document new dse-max-object-size param.
25422 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25423 * tree-ssa-dse.c: Include params.h.
25424 (dse_store_status): New enum.
25425 (initialize_ao_ref_for_dse): New, partially extracted from
25427 (valid_ao_ref_for_dse, normalize_ref): New.
25428 (setup_live_bytes_from_ref, compute_trims): Likewise.
25429 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25430 (maybe_trim_partially_dead_store): Likewise.
25431 (maybe_trim_complex_store): Likewise.
25432 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25433 Track what bytes live from the original store. Return tri-state
25434 for dead, partially dead or live.
25435 (dse_dom_walker): Add constructor, destructor and new private members.
25436 (delete_dead_call, delete_dead_assignment): New extracted from
25438 (dse_optimize_stmt): Make a member of dse_dom_walker.
25439 Use initialize_ao_ref_for_dse.
25441 PR tree-optimization/33562
25442 PR tree-optimization/61912
25443 PR tree-optimization/77485
25444 * sbitmap.h (bitmap_count_bits): Prototype.
25445 (bitmap_clear_range, bitmap_set_range): Likewise.
25446 * sbitmap.c (bitmap_clear_range): New function.
25447 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25449 2017-01-13 Martin Liska <mliska@suse.cz>
25452 * function.c (set_cfun): Add new argument force.
25453 * function.h (set_cfun): Likewise.
25454 * ipa-inline-transform.c (inline_call): Use the function when
25455 strict alising from is dropped for function we inline to.
25457 2017-01-13 Richard Biener <rguenther@suse.de>
25459 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25460 for dumping GIMPLE INTEGER_CSTs.
25462 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25464 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25465 to 201112L since C++17.
25467 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25470 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25471 if -fsanitize=kernel-address is present.
25473 2017-01-13 Richard Biener <rguenther@suse.de>
25475 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25476 as _Literal ( type ) number in case usual suffixes do not
25477 preserve all information.
25479 2017-01-13 Richard Biener <rguenther@suse.de>
25481 PR tree-optimization/77283
25482 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25483 and ssa-iterators.h.
25484 (is_feasible_trace): Implement a cost model based on joiner
25487 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25490 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25491 char or short to __float128/_Float128 directly.
25493 2017-01-12 Martin Sebor <msebor@redhat.com>
25495 to -Wformat-overflow.
25496 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25497 (min_bytes_remaining): Same.
25498 (get_string_length): Same.
25499 (format_string): Same.
25500 (format_directive): Same.
25502 (pass_sprintf_length::handle_gimple_call): Same.
25504 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25506 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25507 info.nowrite calls with no lhs that can't throw. Return bool
25508 whether gsi_remove has been called or not.
25509 (pass_sprintf_length::handle_gimple_call): Return bool whether
25510 try_substitute_return_value called gsi_remove. Formatting fix.
25511 (pass_sprintf_length::execute): Don't use gsi_remove if
25512 handle_gimple_call returned true.
25515 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25516 be removed due to side-effects, don't remove following barrier nor
25517 turn the successor edge into fallthru edge.
25519 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25522 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25523 element-reversing loads and stores as not swappable.
25525 2017-01-12 Nathan Sidwell <nathan@acm.org>
25526 Nicolai Stange <nicstange@gmail.com>
25528 * combine.c (try_combine): Don't ignore result of overlap checking
25529 loop. Combine overlap & asm check into single loop.
25531 2017-01-12 Richard Biener <rguenther@suse.de>
25533 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25534 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25536 2017-01-12 Richard Biener <rguenther@suse.de>
25538 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25539 and TS_TARGET_OPTION directly derive from TS_BASE.
25540 * tree-core.h (tree_optimization_option): Derive from tree_base.
25541 (tree_target_option): Likewise.
25543 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25545 * config/i386/i386.c (memory_address_length): Increase len
25546 only when rip_relative_addr_p returns false.
25548 2017-01-11 Julia Koval <julia.koval@intel.com>
25550 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25551 (OPTION_MASK_ISA_SGX_SET): New.
25552 (ix86_handle_option): Handle OPT_msgx.
25553 * config.gcc: Added sgxintrin.h.
25554 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25555 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25556 * config/i386/i386.c (ix86_target_string): Add -msgx.
25558 (ix86_option_override_internal): Handle new options.
25559 (ix86_valid_target_attribute_inner_p): Add sgx.
25560 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25561 * config/i386/i386.opt: Add msgx.
25562 * config/i386/sgxintrin.h: New file.
25563 * config/i386/x86intrin.h: Add sgxintrin.h.
25565 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25568 * fold-const.c (maybe_nonzero_address): Return 1 for function
25570 (tree_single_nonzero_warnv_p): Don't handle function local objects
25574 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25577 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25580 * opts.c: Include "spellcheck.h"
25581 (struct string_fragment): New struct.
25582 (struct edit_distance_traits<const string_fragment &>): New
25584 (get_closest_sanitizer_option): New function.
25585 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25587 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25589 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25591 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25592 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25593 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25594 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25595 for initial die_offset if dwarf_split_debug_info.
25596 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25597 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25599 (output_skeleton_debug_sections): Formatting fix. Use
25600 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25601 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25603 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25605 * config/arm/cortex-a53.md: Add bypasses for
25606 cortex_a53_r2f_cvt.
25607 (cortex_a53_r2f): Only use for transfers.
25608 (cortex_a53_f2r): Likewise.
25609 (cortex_a53_r2f_cvt): Add reservation for conversions.
25610 (cortex_a53_f2r_cvt): Likewise.
25612 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25614 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25615 to all inlined functions, change static to extern.
25617 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25620 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25622 (arm_assemble_integer): Likewise.
25624 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25626 * config.gcc: Use new awk script to check CPU, FPU and architecture
25627 parameters for --with-... options.
25628 * config/arm/parsecpu.awk: New file
25629 * config/arm/arm-cpus.in: New file.
25630 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25632 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25634 * config/arm/t-arm: Update dependency rules.
25635 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25636 of processing .def files.
25637 * config/arm/genopt.sh: Deleted.
25638 * config/arm/gentune.sh: Deleted.
25639 * config/arm/arm-cores.def: Deleted.
25640 * config/arm/arm-arches.def: Deleted.
25641 * config/arm/arm-fpus.def: Deleted.
25642 * config/arm/arm-tune.md: Regenerated.
25643 * config/arm/arm-tables.opt: Regenerated.
25644 * config/arm/arm-cpu.h: New generated file.
25645 * config/arm/arm-cpu-data.h: New generated file.
25646 * config/arm/arm-cpu-cdata.h: New generated file.
25648 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25651 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25653 (input_varpool_node): Unpack dynamically_initialized bit.
25655 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25657 PR rtl-optimization/79032
25658 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25659 the alignment of the adjusted memory reference against that of MODE,
25660 instead of the alignment of the original memory reference.
25662 2017-01-11 Martin Jambor <mjambor@suse.cz>
25664 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25666 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25667 decorated functions.
25669 2017-01-11 Richard Biener <rguenther@suse.de>
25671 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25672 set range/nonnull info for PHI results. Do not set it on
25673 stmts marked for removal.
25675 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25677 * expr.c (store_field): In the bitfield case, fetch the return value
25678 from the registers before applying a single big-endian adjustment.
25679 Always do a final load for a BLKmode value not larger than a word.
25681 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25684 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25685 that we correctly handle column numbers greater than
25686 LINE_MAP_MAX_COLUMN_NUMBER.
25688 2017-01-10 Martin Sebor <msebor@redhat.com>
25690 PR middle-end/78245
25691 * gimple-ssa-sprintf.c (get_destination_size): Call
25692 {init,fini}object_sizes.
25693 * tree-object-size.c (addr_object_size): Adjust.
25694 (pass_through_call): Adjust.
25695 (pass_object_sizes::execute): Adjust.
25696 * tree-object-size.h (fini_object_sizes): Declare.
25698 2017-01-10 Martin Sebor <msebor@redhat.com>
25700 PR tree-optimization/78775
25701 * builtins.c (get_size_range): Move...
25702 * calls.c: ...to here.
25703 (alloc_max_size): Accept zero argument.
25704 (operand_signed_p): Remove.
25705 (maybe_warn_alloc_args_overflow): Call get_size_range.
25706 * calls.h (get_size_range): Declare.
25708 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25710 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25711 from TI's devices.csv file as of September 2016.
25712 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25714 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25716 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25717 * doc/invoke.texi: Likewise.
25718 * doc/md.texi: Likewise.
25719 * doc/objc.texi: Likewise.
25721 2017-01-10 Joshua Conner <joshconner@google.com>
25723 * config/arm/fuchsia-elf.h: New file.
25724 * config/fuchsia.h: New file.
25725 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25726 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25728 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25730 2016-01-10 Richard Biener <rguenther@suse.de>
25732 PR tree-optimization/79034
25733 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25734 Propagate out degenerate PHIs in the joiner.
25736 2017-01-10 Martin Liska <mliska@suse.cz>
25738 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25739 (sort_congruence_classes_by_decl_uid): Likewise.
25740 (sort_congruence_class_groups_by_decl_uid): Likewise.
25741 (sem_item_optimizer::merge_classes): Sort class, groups in these
25742 classes and members in the groups by DECL_UID of declarations.
25743 This would make merge operations stable.
25745 2017-01-10 Martin Liska <mliska@suse.cz>
25747 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25748 usage of m_classes_vec.
25749 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25750 (sem_item_optimizer::get_group_by_hash): Likewise.
25751 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25752 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25753 (sem_item_optimizer::verify_classes): Likewise.
25754 (sem_item_optimizer::process_cong_reduction): Likewise.
25755 (sem_item_optimizer::dump_cong_classes): Likewise.
25756 (sem_item_optimizer::merge_classes): Likewise.
25757 * ipa-icf.h (congruence_class_hash): Rename from
25758 congruence_class_group_hash. Remove declaration of m_classes_vec.
25760 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25762 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25763 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25764 * config.gcc: Add avx512vpopcntdqintrin.h.
25765 * config/i386/avx512vpopcntdqintrin.h: New.
25766 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25767 * config/i386/i386-builtin-types.def: Add new types.
25768 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25769 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25770 __builtin_ia32_vpopcountq_v8di_mask): New.
25771 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25772 __AVX512VPOPCNTDQ__.
25773 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25774 (PTA_AVX512VPOPCNTDQ): Define.
25775 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25776 TARGET_AVX512VPOPCNTDQ_P): Define.
25777 * config/i386/i386.opt: Add mavx512vpopcntdq.
25778 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25779 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25781 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25783 PR middle-end/77484
25784 * predict.def (PRED_CALL): Set to 67.
25786 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25788 * expr.c (store_field): In the bitfield case, if the value comes from
25789 a function call and is of an aggregate type returned in registers, do
25790 not modify the field mode; extract the value in all cases if the mode
25791 is BLKmode and the size is not larger than a word.
25793 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25796 * config/i386/cpuid.h: Fix undefined behavior.
25798 2017-01-04 Jeff Law <law@redhat.com>
25800 PR tree-optimization/79007
25801 PR tree-optimization/67955
25802 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25803 conservative for pt.null when flag_non_call_exceptions is on.
25805 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25807 PR translation/79019
25808 PR translation/79020
25809 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25810 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25811 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25813 * config/avr/avr.opt (maccumulate-args): Likewise.
25814 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25815 * common.opt (freport-bug): Likewise.
25816 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25817 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25818 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25819 translatable string.
25820 * config/i386/i386.c (function_value_32): Likewise.
25821 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25822 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25824 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25825 * common/config/msp430/msp430-common.c (msp430_handle_option):
25827 * symtab.c (symtab_node::verify_base): Likewise.
25828 * opts.c (set_debug_level): Likewise.
25829 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25830 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25831 missing whitespace to translatable strings.
25832 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25833 * config/sh/superh.h: Likewise.
25834 * config/i386/xopintrin.h: Likewise.
25835 * config/i386/znver1.md: Likewise.
25836 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25837 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25838 * double-int.h (struct double_int): Likewise.
25839 * double-int.c (div_and_round_double): Likewise.
25840 * wide-int.cc: Likewise.
25841 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25842 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25843 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25844 (crossjumps_occurred): ... this.
25845 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25848 PR tree-optimization/78899
25849 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25850 returning bool return struct loop *, NULL for failure and the new
25852 (versionable_outer_loop_p): Don't version outer loop if it has
25853 dont_vectorized bit set.
25854 (tree_if_conversion): When versioning outer loop, ensure
25855 tree_if_conversion is performed also on the inner loop of the
25856 non-vectorizable outer loop copy.
25857 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25858 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25859 prevent vectorization of it.
25860 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25861 the outer loop vectorization of the non-scalar version is attempted
25862 before vectorization of the inner loop in scalar version. If
25863 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25864 vectorization of its inner loop.
25865 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25866 has 2 inner loops, rename also on edges from bb whose single pred
25867 is outer_loop->header. Fix typo in function comment.
25869 2017-01-09 Martin Sebor <msebor@redhat.com>
25872 * asan.c (asan_emit_stack_protection): Increase local buffer size
25873 to avoid snprintf truncation warning.
25875 2017-01-09 Andrew Pinski <apinski@cavium.com>
25877 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25878 to reference thunderx2t99 for the tuning structure
25879 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25881 (thunderx2t99_extra_costs): This.
25882 * config/aarch64/aarch64-tune.md: Regenerate.
25883 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25884 (vulcan_addrcost_table): This.
25885 (vulcan_regmove_cost): Rename to ...
25886 (thunderx2t99_regmove_cost): This.
25887 (vulcan_vector_cost): Rename to ...
25888 (thunderx2t99_vector_cost): this.
25889 (vulcan_branch_cost): Rename to ...
25890 (thunderx2t99_branch_cost): This.
25891 (vulcan_tunings): Rename to ...
25892 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25893 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25895 2017-01-09 Martin Jambor <mjambor@suse.cz>
25899 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25900 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25901 (propagate_vr_accross_jump_function): Use the above function for all
25902 value range computations for pass-through jump functions and type
25903 converasion from explicit value range values.
25904 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25905 parameters from TYPE_ARG_TYPES.
25906 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25907 (ipa_write_node_info): Stream type of the actual argument.
25908 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25910 2017-01-09 Martin Liska <mliska@suse.cz>
25913 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25914 (lookup_compiler): Do not show error message with have_E.
25916 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25918 PR tree-optimization/78938
25919 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25920 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25921 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25922 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25925 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25927 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25930 2017-01-09 Richard Biener <rguenther@suse.de>
25932 PR tree-optimization/78997
25933 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25934 name condition properly.
25936 2017-01-09 Richard Biener <rguenther@suse.de>
25939 * dwarf2out.c (is_cxx): New overload with context.
25940 (is_naming_typedef_decl): Use it.
25942 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25944 * invoke.texi (Option Summary): Correct spacing in option lists
25945 and add line breaks to fix over-long lines.
25947 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25949 PR middle-end/17660
25951 * extend.texi (Common Variable Attributes): Add xref to GCC
25952 Internals manual to explain mode attribute keywords.
25954 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25957 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25958 and Preprocessor Options.
25959 (Options for Linking): Document -pthread here....
25960 (RS/6000 and PowerPC Options): ...not here.
25961 (Solaris 2 Options): ...or here.
25962 * doc/cppopts.texi: Document -pthread.
25964 2017-01-08 Martin Sebor <msebor@redhat.com>
25966 PR middle-end/77708
25967 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25968 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25969 New member functions.
25970 (format_directive): Used them.
25972 (pass_sprintf_length::handle_gimple_call): Same.
25973 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25974 to avoid truncation for any argument.
25975 (extract_affine_mul): Same.
25976 * tree.c (get_file_function_name): Same.
25978 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25980 PR middle-end/77484
25981 * predict.def (PRED_INDIR_CALL): Set to 86.
25983 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25985 PR preprocessor/54124
25986 * doc/cppopts.texi: Reformat -d subtable to list the full name
25987 of the options. Add cross-reference to the docs for the general
25988 compiler -d options.
25989 * doc/invoke.texi (Developer Options): Add cross-reference to the
25990 preprocessor-specific -d option documentation.
25992 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25994 PR preprocessor/13498
25995 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25996 redudant material, and reflect new command-line options.
25997 (System Headers): Likewise.
25999 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26001 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26002 -isystem, and -idirafter. Copy-edit.
26003 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26004 default for -ftrack-macro-expansion. Delete obsolete and
26005 badly-formatted implementation details about -fdebug-cpp output.
26006 * doc/cppwarnopts.texi: Copy-edit.
26008 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26011 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26012 that the transition from a max line width >= 1<<10 to narrower
26013 lines works correctly.
26015 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26017 * doc/options.texi (PerFunction): New.
26018 * opt-functions.awk (switch_flags): Map both Optimization and
26019 PerFunction to CL_OPTIMIZATION.
26020 * opth-gen.awk: Test for PerFunction flag along with
26022 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26023 it only when the latter is present. Skip those that don't in
26024 the hash function generator.
26025 * common.opt (fvar-tracking): Mark as PerFunction instead of
26027 (fvar-tracking-assignments): Likewise.
26028 (fvar-tracking-assignments-toggle): Likewise.
26029 (fvar-tracking-uninit): Likewise.
26031 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26033 PR translation/79018
26034 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26037 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26040 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26041 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26042 TARGET_LONG_JUMP_TABLE_OFFSETS.
26043 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26044 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26045 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26046 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26047 * config/m68k/m68k.md (tablejump expander): Likewise.
26048 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26049 TARGET_LONG_JUMP_TABLE_OFFSETS.
26050 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26051 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26053 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26054 David Holsgrove <david.holsgrove@xilinx.com>
26056 * common/config/microblaze/microblaze-common.c
26057 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26058 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26060 * config/microblaze/microblaze.c (microblaze_must_save_register)
26061 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26063 (microblaze_eh_return): New function.
26064 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26065 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26066 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26067 * config/microblaze/microblaze.md (eh_return): New pattern.
26069 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26071 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26072 GCC_DIAGNOSTIC_STRINGIFY): Define.
26074 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26076 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26078 * config/arm/arm.md (<mcrr>): New.
26080 * config/arm/arm.c (arm_arch5te): New.
26081 (arm_option_override): Set arm_arch5te.
26082 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26084 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26085 (arm_mcrr_qualifiers): ... this. New.
26086 (MRRC_QUALIFIERS): Define to...
26087 (arm_mrrc_qualifiers): ... this. New.
26088 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26090 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26091 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26092 (MRRCI, mrrc, MRRC): New.
26093 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26094 VUNSPEC_MRRC2): New.
26096 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26098 * config/arm/arm.md (<mcr>): New.
26100 * config/arm/arm.c (arm_coproc_builtin_available): Add
26101 support for mcr, mrc, mcr2 and mrc2.
26102 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26103 (arm_mcr_qualifiers): ... this. New.
26104 (MRC_QUALIFIERS): Define to ...
26105 (arm_mrc_qualifiers): ... this. New.
26106 (MCR_QUALIFIERS): Define to ...
26107 (arm_mcr_qualifiers): ... this. New.
26108 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26110 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26111 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26112 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26113 VUNSPEC_MRC2): New.
26115 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26117 * config/arm/arm.md (*ldc): New.
26121 * config/arm/arm.c (arm_coproc_builtin_available): Add
26122 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26123 (arm_coproc_ldc_stc_legitimate_address): New.
26124 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26125 'qualifier_const_pointer'.
26126 (LDC_QUALIFIERS): Define to...
26127 (arm_ldc_qualifiers): ... this. New.
26128 (STC_QUALIFIERS): Define to...
26129 (arm_stc_qualifiers): ... this. New.
26130 * config/arm/arm-protos.h
26131 (arm_coproc_ldc_stc_legitimate_address): New.
26132 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26133 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26134 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26135 stc2, stcl, stc2l): New.
26136 * config/arm/constraints.md (Uz): New.
26137 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26138 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26139 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26140 VUNSPEC_STC2L): New.
26142 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26144 * config/arm/arm.md (<cdp>): New.
26145 * config/arm/arm.c (neon_const_bounds): Rename this ...
26146 (arm_const_bounds): ... this.
26147 (arm_coproc_builtin_available): New.
26148 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26149 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26150 (CDP_QUALIFIERS): Define to...
26151 (arm_cdp_qualifiers): ... this. New.
26153 (arm_expand_builtin_args): Add case for 6 arguments.
26154 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26155 (arm_const_bounds): ... this.
26156 (arm_coproc_builtin_available): New.
26157 * config/arm/arm_acle.h (__arm_cdp): New.
26159 * config/arm/arm_acle_builtins.def (cdp): New.
26161 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26162 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26163 'arm_const_bounds'.
26164 * config/arm/types.md (coproc): New.
26165 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26166 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26167 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26168 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26170 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26172 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26173 (UBINOP_QUALIFIERS): New.
26175 (acle_builtin_data): New. Change comment.
26176 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26177 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26178 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26179 arm_acle_builtins.def.
26180 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26181 (arm_init_acle_builtins): New.
26182 (CRC32_BUILTIN): Remove.
26183 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26184 crc32cb, crc32ch and crc32cw.
26185 (arm_init_crc32_builtins): Remove.
26186 (arm_init_builtins): Use arm_init_acle_builtins rather
26187 than arm_init_crc32_builtins.
26188 (arm_expand_acle_builtin): New.
26189 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26190 * config/arm/arm_acle_builtins.def: New.
26192 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26194 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26195 (arm_builtin_datum): ... this.
26196 (arm_init_neon_builtin): Rename to ...
26197 (arm_init_builtin): ... this. Add a new parameters PREFIX
26198 and USE_SIG_IN_NAME.
26199 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26200 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26201 'arm_builtin_datum'.
26202 (arm_init_vfp_builtins): Likewise.
26203 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26204 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26205 (arm_expand_neon_args): Rename to ...
26206 (arm_expand_builtin_args): ... this. Rename builtin_arg
26207 enum values and differentiate between ARG_BUILTIN_MEMORY
26208 and ARG_BUILTIN_NEON_MEMORY.
26209 (arm_expand_neon_builtin_1): Rename to ...
26210 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26211 values, arm_expand_builtin_args and add bool parameter NEON.
26212 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26213 (arm_expand_vfp_builtin): Likewise.
26214 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26216 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26218 PR middle-end/77484
26219 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26220 * predict.c (tree_estimate_probability_bb): Reverse direction of
26221 polymorphic call predictor.
26223 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26225 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26226 (determine_pass_name_match): ...this new function and...
26227 (should_skip_pass_p): ...this new function.
26229 2017-01-06 Nathan Sidwell <nathan@acm.org>
26231 * ipa-visibility.c (function_and_variable_visibility): Reformat
26232 comments and long lines. Remove extrneous if.
26233 * symtab.c (symtab_node::make_decl_local): Fix code format.
26234 (symtab_node::set_section_for_node): Fix comment typo.
26236 2017-01-06 Martin Liska <mliska@suse.cz>
26239 * lra-constraints.c: Rename invariant to lra_invariant.
26240 * predict.c (set_even_probabilities): Initialize e to NULL.
26242 2017-01-05 Martin Sebor <msebor@redhat.com>
26244 PR tree-optimization/78910
26245 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26246 (format_integer): Correct off-by-one error in the handling
26247 of precision with negative numbers in signed conversions..
26249 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26251 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26253 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26255 PR tree-optimization/71016
26256 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26257 factor_out_conditional_conversion. Formatting fix.
26258 (factor_out_conditional_conversion): Add cond_stmt argument.
26259 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26260 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26263 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26265 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26266 read-rtl-function.o, and selftest-rtl.o.
26267 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26268 (selftest::aarch64_test_loading_full_dump): New function.
26269 (selftest::aarch64_run_selftests): New function.
26270 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26271 selftest::aarch64_run_selftests.
26272 * config/i386/i386.c
26273 (selftest::ix86_test_loading_dump_fragment_1): New function.
26274 (selftest::ix86_test_loading_call_insn): New function.
26275 (selftest::ix86_test_loading_full_dump): New function.
26276 (selftest::ix86_test_loading_unspec): New function.
26277 (selftest::ix86_run_selftests): Call the new functions.
26278 * emit-rtl.c (maybe_set_max_label_num): New function.
26279 * emit-rtl.h (maybe_set_max_label_num): New decl.
26280 * function.c (instantiate_decls): Guard call to
26281 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26282 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26284 * gensupport.c (gen_reader::gen_reader): Pass "false"
26285 for new "compact" param of rtx_reader.
26286 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26287 rather than an empty string for NULL strings.
26288 * read-md.c: Potentially include config.h rather than bconfig.h.
26289 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26290 (have_error): New global, copied from errors.c.
26291 (md_reader::read_name): Rename to...
26292 (md_reader::read_name_1): ...this, adding "out_loc" param,
26293 and converting "missing name or number" to returning false, rather
26295 (md_reader::read_name): Reimplement in terms of read_name_1.
26296 (md_reader::read_name_or_nil): New function.
26297 (md_reader::read_string): Handle "(nil)" by returning NULL.
26298 (md_reader::md_reader): Add new param "compact".
26299 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26300 (md_reader::read_file): New method.
26301 * read-md.h (md_reader::md_reader): Add new param "compact".
26302 (md_reader::read_file): New method.
26303 (md_reader::is_compact): New accessor.
26304 (md_reader::read_name): Convert return type from void to file_location.
26305 (md_reader::read_name_or_nil): New decl.
26306 (md_reader::read_name_1): New decl.
26307 (md_reader::m_compact): New field.
26308 (noop_reader::noop_reader): Pass "false" for new "compact" param
26310 (rtx_reader::rtx_reader): Add new "compact" param.
26311 (rtx_reader::read_rtx_operand): Make virtual and convert return
26312 type from void to rtx.
26313 (rtx_reader::read_until): New decl.
26314 (rtx_reader::handle_any_trailing_information): New virtual function.
26315 (rtx_reader::postprocess): New virtual function.
26316 (rtx_reader::finalize_string): New virtual function.
26317 (rtx_reader::m_in_call_function_usage): New field.
26318 (rtx_reader::m_reuse_rtx_by_id): New field.
26319 * read-rtl-function.c: New file.
26320 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26321 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26322 (selftest::verify_three_block_rtl_cfg): New decl.
26323 * read-rtl-function.h: New file.
26324 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26325 For host, include function.h, memmodel.h, and emit-rtl.h.
26326 (one_time_initialization): New function.
26327 (struct compact_insn_name): New struct.
26328 (compact_insn_names): New array.
26329 (find_code): Handle insn codes in compact dumps.
26330 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26331 (bind_subst_iter_and_attr): Likewise.
26332 (add_condition_to_string): Likewise.
26333 (add_condition_to_rtx): Likewise.
26334 (apply_attribute_uses): Likewise.
26335 (add_current_iterators): Likewise.
26336 (apply_iterators): Likewise.
26337 (initialize_iterators): Guard usage of apply_subst_iterator with
26338 #ifdef GENERATOR_FILE.
26339 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26340 (md_reader::read_mapping): Likewise.
26341 (add_define_attr_for_define_subst): Likewise.
26342 (add_define_subst_attr): Likewise.
26343 (read_subst_mapping): Likewise.
26344 (check_code_iterator): Likewise.
26345 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26347 (one_time_initialization): New function.
26348 (rtx_reader::read_until): New method.
26349 (read_flags): New function.
26350 (parse_reg_note_name): New function.
26351 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26352 Handle reuse_rtx ids.
26353 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26354 Add parsing support for RTL dumps, mirroring the special-cases in
26355 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26356 values, and calling handle_any_trailing_information.
26357 (rtx_reader::read_rtx_operand): Convert return type from void
26358 to rtx, returning return_rtx. Handle case 'e'. Call
26359 finalize_string on XSTR and XTMPL fields.
26360 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26361 "(nil)" values were omitted. Call the postprocess vfunc on the
26363 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26364 class ctor. Initialize m_in_call_function_usage. Call
26365 one_time_initialization.
26366 * rtl-tests.c (selftest::test_uncond_jump): Call
26367 set_new_first_and_last_insn.
26368 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26369 * selftest-rtl.c: New file.
26370 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26371 (selftest::get_insn_by_uid): New decl.
26372 * selftest-run-tests.c (selftest::run_tests): Call
26373 read_rtl_function_c_tests.
26374 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26375 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26378 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26380 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26381 operands in a special way. Assert that pos+len <= mode precision.
26383 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26385 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26386 3 argument Alias with unlimited for the negative form.
26387 (fno-vect-cost-model): Removed.
26389 2017-01-05 Martin Liska <mliska@suse.cz>
26391 * hsa-gen.c (gen_hsa_divmod): New function.
26392 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26394 2017-01-05 Martin Liska <mliska@suse.cz>
26397 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26400 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26402 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26403 small constant length operands.
26405 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26407 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26408 between loop iterations.
26410 2017-01-05 Martin Liska <mliska@suse.cz>
26413 * gimplify.c (gimplify_decl_expr): Compare to
26414 asan_poisoned_variables instread of checking flags.
26415 (gimplify_target_expr): Likewise.
26416 (gimplify_expr): Likewise.
26417 (gimplify_function_tree): Conditionally initialize
26418 asan_poisoned_variables.
26420 2017-01-04 Jeff Law <law@redhat.com>
26422 PR tree-optimizatin/78812
26423 * rtl.h (contains_mem_rtx_p): Prototype.
26424 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26425 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26426 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26427 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26428 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26430 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26432 * input.c (assert_char_at_range): Default-initialize actual_range.
26434 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26436 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26437 to match the caller.
26439 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26441 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26442 insns after final jump in test to emit dummy move.
26444 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26446 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26447 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26449 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26451 * multiple_target.c (create_dispatcher_calls): Init e_next.
26452 * tree-ssa-loop-split.c (split_loop): Init border.
26453 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26456 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26461 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26462 (altivec_register_operand): Do not return true if the operand
26463 contains a SUBREG mixing SImode and SFmode.
26464 (vsx_register_operand): Likewise.
26465 (vsx_reg_sfsubreg_ok): New predicate.
26466 (vfloat_operand): Do not return true if the operand contains a
26467 SUBREG mixing SImode and SFmode.
26468 (vint_operand): Likewise.
26469 (vlogical_operand): Likewise.
26470 (gpc_reg_operand): Likewise.
26471 (int_reg_operand): Likewise.
26472 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26473 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26474 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26476 (rs6000_emit_move_si_sf_subreg): New helper function.
26477 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26478 fixup SUBREGs involving SImode and SFmode.
26479 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26480 numbers for the new peephole2 optimization.
26481 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26482 the GLIBC math library that do AND/IOR/XOR operations on single
26483 precision floating point.
26484 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26485 target macros to say whether we need to avoid SUBREGs mixing
26487 (TARGET_ALLOW_SF_SUBREG): Likewise.
26488 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26489 (UNSPEC_SI_FROM_SF): Likewise.
26490 (iorxor): Change spacing.
26491 (and_ior_xor): New iterator for AND, IOR, and XOR.
26492 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26493 (movdi_from_sf_zero_ext): Likewise.
26494 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26495 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26496 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26497 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26498 (fms<mode>4): Likewise.
26499 (fnma<mode>4): Likewise.
26500 (fnms<mode>4): Likewise.
26501 (nfma<mode>4): Likewise.
26502 (nfms<mode>4): Likewise.
26504 2017-01-04 Marek Polacek <polacek@redhat.com>
26507 * doc/invoke.texi: Document -Wpointer-compare.
26509 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26511 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26514 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26515 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26518 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26521 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26522 documentation of the powerpc_popcntb_ok attribute.
26523 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26524 code to issue warning messages if a requested CPU configuration is
26525 not supported by the binary (assembler and loader) toolchain.
26526 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26527 made to define a built-in function that has been disabled.
26528 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26529 made to define a built-in function that has been disabled.
26530 (altivec_init_builtins): Add comment explaining why definition
26531 of the DST built-in functions is not preceded by an assertion
26532 check. Add assertions to prevent ICE if attempts are made to
26533 define an altivec predicate or an abs* built-in function that has
26535 (htm_init_builtins): Add comment explaining why definition of the
26536 htm built-in functions is not preceded by an assertion check.
26538 2017-01-04 Jeff Law <law@redhat.com>
26540 PR tree-optimizatin/67955
26541 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26542 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26543 the points-to solution does not include pt_null. Use DECL_PT_UID
26546 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26548 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26549 Use gen_int_mode instead of gen_lopwart for const_int operands.
26551 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26553 PR tree-optimization/71563
26554 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26555 out of range value - has low bits known to be zero.
26557 2017-01-04 Alan Modra <amodra@gmail.com>
26559 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26560 * configure: Regenerate.
26561 * config.in: Regenerate.
26563 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26566 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26567 a substring of the message, but strcmp with the whole message. Ifdef
26568 ENABLE_NLS, translate the message first using dgettext.
26570 2017-01-03 Jeff Law <law@redhat.com>
26572 PR tree-optimizatin/78856
26573 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26574 (mark_threaded_blocks): Remove code to truncate thread paths that
26575 cross multiple loop headers. Instead invalidate the cached loop
26576 iteration information and handle case of a thread path walking
26577 into an irreducible region.
26579 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26582 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26583 assertions. Add support for doing the signbit if the IEEE 128-bit
26584 floating point value is in a GPR.
26585 * config/rs6000/rs6000.md (Fsignbit): Delete.
26586 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26587 Update the length attribute if the value is in a GPR.
26588 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26589 the sign or zero extension instruction, since the value is always 0/1.
26590 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26593 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26594 extracting SImode to a GPR register so that we can generate a
26595 store, limit the vector to be in a traditional Altivec register
26596 for the vextuwrx instruction.
26598 2017-01-03 Ian Lance Taylor <iant@google.com>
26600 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26602 2017-01-03 Martin Sebor <msebor@redhat.com>
26604 PR tree-optimization/78696
26605 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26606 precision. Use MPFR for %f for greater fidelity. Correct handling
26608 (pass_sprintf_length::compute_format_length): Set width and precision
26609 specified by asrerisk to void_node for vararg functions.
26610 (try_substitute_return_value): Adjust dump output.
26612 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26614 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26616 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26618 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26619 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26620 -mlra/-mno-lra was passed to the compiler.
26622 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26624 PR rtl-optimization/65618
26625 * emit-rtl.c (try_split): Move initialization of "before" and
26626 "after" to just before the call to emit_insn_after_setloc.
26628 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26630 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26632 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26634 * dwarf2out.c (gen_enumeration_type_die): When
26635 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26637 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26639 PR tree-optimization/78965
26640 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26641 Change first argument from const call_info & to call_info &. For %n
26642 set info.nowrite to false.
26644 PR middle-end/78901
26645 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26646 possibly throwing calls.
26648 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26649 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26650 and fns handling, rather than in a separate case SSA_NAME.
26652 2017-01-02 Jeff Law <law@redhat.com>
26654 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26655 fixes for first_period and second_period variables.
26657 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26660 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26661 (*insvqi_1): New insn pattern.
26662 (*insvqi_1_mem_rex64): Ditto.
26663 (*insvqi_2): Ditto.
26664 (*insvqi_3): Rename from *insvqi.
26666 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26668 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26670 * doc/cfg.texi (Edges): Remove reference to Java.
26671 (Maintaining the CFG): Ditto.
26673 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26675 PR middle-end/77674
26676 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26677 transparent aliases.
26679 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26681 PR middle-end/77484
26682 * predict.def (PRED_CALL): Update hitrate.
26683 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26684 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26685 into direct/indirect/polymorphic variants.
26687 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26689 Update copyright years.
26691 * gcc.c (process_command): Update copyright notice dates.
26692 * gcov-dump.c (print_version): Ditto.
26693 * gcov.c (print_version): Ditto.
26694 * gcov-tool.c (print_version): Ditto.
26695 * gengtype.c (create_file): Ditto.
26696 * doc/cpp.texi: Bump @copying's copyright year.
26697 * doc/cppinternals.texi: Ditto.
26698 * doc/gcc.texi: Ditto.
26699 * doc/gccint.texi: Ditto.
26700 * doc/gcov.texi: Ditto.
26701 * doc/install.texi: Ditto.
26702 * doc/invoke.texi: Ditto.
26704 Copyright (C) 2017 Free Software Foundation, Inc.
26706 Copying and distribution of this file, with or without modification,
26707 are permitted in any medium without royalty provided the copyright
26708 notice and this notice are preserved.