1 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4 to support float128 built-in functions that require the ISA 3.0
6 (BU_FLOAT128_3_HW): Likewise.
7 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13 floating point instructions.
14 (rs6000_invalid_builtin): Likewise.
15 (rs6000_builtin_mask_names): Likewise.
16 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
17 (RS6000_BTM_FLOAT128_HW): Likewise.
18 (RS6000_BTM_COMMON): Likewise.
19 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
21 * doc/extend.texi (RS/6000 built-in functions): Document the
22 IEEE 128-bit floating point square root and fused multiply-add
25 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
27 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
28 reg (r2) isn't in the set of registers defined in the prologue.
30 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
31 Alan Hayward <alan.hayward@arm.com>
32 David Sherwood <david.sherwood@arm.com>
34 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
35 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
36 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
37 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
38 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
40 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
41 max_vectorization_factor.
42 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
44 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
45 Alan Hayward <alan.hayward@arm.com>
46 David Sherwood <david.sherwood@arm.com>
48 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
49 (vect_worthwhile_without_simd_p): Declare.
50 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
51 (vectorizable_reduction): Use it.
52 * tree-vect-stmts.c (vectorizable_shift): Likewise.
53 (vectorizable_operation): Likewise.
55 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
56 Alan Hayward <alan.hayward@arm.com>
57 David Sherwood <david.sherwood@arm.com>
59 * tree-vectorizer.h (vect_get_num_copies): New function.
60 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
61 * tree-vect-loop.c (vectorizable_reduction): Likewise.
62 (vectorizable_induction): Likewise.
63 (vectorizable_live_operation): Likewise.
64 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
65 (vectorizable_bswap): Likewise.
66 (vectorizable_call): Likewise.
67 (vectorizable_conversion): Likewise.
68 (vectorizable_assignment): Likewise.
69 (vectorizable_shift): Likewise.
70 (vectorizable_operation): Likewise.
71 (vectorizable_store): Likewise.
72 (vectorizable_load): Likewise.
73 (vectorizable_condition): Likewise.
74 (vectorizable_comparison): Likewise.
75 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
77 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
78 Alan Hayward <alan.hayward@arm.com>
79 David Sherwood <david.sherwood@arm.com>
81 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
84 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
85 Alan Hayward <alan.hayward@arm.com>
86 David Sherwood <david.sherwood@arm.com>
88 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
89 an inline wrapper that provides a location.
90 (gimple_build_vector): Likewise.
91 * gimple-fold.c (gimple_build_vector_from_val): New function.
92 (gimple_build_vector): Likewise.
93 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
94 functions to build the initial value. Always return a gimple value.
95 (get_initial_defs_for_reduction): Likewise. Only compute
97 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
98 vect_init_vector on the results from get_initial_def(s)_for_reduction.
99 (vectorizable_induction): Use gimple_build_vector rather than
102 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
103 Alan Hayward <alan.hayward@arm.com>
104 David Sherwood <david.sherwood@arm.com>
106 * target.h (vec_perm_indices): New typedef.
107 (auto_vec_perm_indices): Likewise.
108 * optabs-query.h: Include target.h
109 (can_vec_perm_p): Take a vec_perm_indices *.
110 * optabs-query.c (can_vec_perm_p): Likewise.
111 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
112 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
113 * tree-vect-generic.c (lower_vec_perm): Likewise.
114 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
115 (vect_grouped_load_supported): Likewise.
116 (vect_shift_permute_load_chain): Likewise.
117 (vect_permute_store_chain): Use auto_vec_perm_indices.
118 (vect_permute_load_chain): Likewise.
119 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
120 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
121 Update uses of can_vec_perm_p.
122 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
123 mode with a number of elements. Take a vec_perm_indices *.
124 (vect_create_epilog_for_reduction): Update accordingly.
125 Use auto_vec_perm_indices.
126 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
127 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
128 (vect_transform_slp_perm_load): Likewise.
129 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
130 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
131 (vect_gen_perm_mask_checked): Likewise.
132 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
133 (vect_gen_perm_mask_checked): Likewise.
134 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
135 (vectorizable_store): Likewise.
136 (vectorizable_load): Likewise.
137 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
138 (vectorizable_bswap): Likewise.
140 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
141 Alan Hayward <alan.hayward@arm.com>
142 David Sherwood <david.sherwood@arm.com>
144 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
145 * tree.c (build_vector): Likewise.
146 (build_vector_from_ctor): Update accordingly.
147 (build_vector_from_val): Likewise.
148 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
149 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
150 * tree-vect-generic.c (add_rshift): Likewise.
151 (expand_vector_divmod): Likewise.
152 (optimize_vector_constructor): Likewise.
153 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
154 (vect_transform_slp_perm_load): Likewise.
155 (vect_schedule_slp_instance): Likewise.
156 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
157 (vectorizable_call): Likewise.
158 (vect_gen_perm_mask_any): Likewise. Add elements in order.
159 * expmed.c (make_tree): Likewise.
160 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
161 a vector passed to build_vector.
162 (fold_convert_const): Likewise.
163 (exact_inverse): Likewise.
164 (fold_ternary_loc): Likewise.
165 (fold_relational_const): Likewise.
166 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
167 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
168 (const_unop): Likewise. Store the reduction accumulator in a
169 variable rather than an array.
170 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
171 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
172 the new vector, rather than constructing it after the input arrays.
173 (native_interpret_vector): Use auto_vec<tree> when building
174 a vector passed to build_vector. Add elements in order.
175 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
176 auto_vec<tree> when building a vector passed to build_vector.
177 (vect_create_epilog_for_reduction): Likewise.
178 (vectorizable_induction): Likewise.
179 (get_initial_def_for_reduction): Likewise. Fix indentation of
181 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
183 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
184 passed to build_vector.
186 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
187 Alan Hayward <alan.hayward@arm.com>
188 David Sherwood <david.sherwood@arm.com>
190 * tree-core.h (tree_base::u): Add an "nelts" field.
191 (tree_vector): Use VECTOR_CST_NELTS as the length.
192 * tree.c (tree_size): Likewise.
193 (make_vector): Initialize VECTOR_CST_NELTS.
194 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
195 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
196 TYPE_VECTOR_SUBPARTS.
197 * expr.c (const_vector_mask_from_tree): Consistently use "units"
198 as the number of units, setting it from VECTOR_CST_NELTS.
199 (const_vector_from_tree): Likewise.
200 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
201 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
202 (fold_negate_expr_1): Likewise.
203 (fold_convert_const): Likewise.
204 (const_binop): Likewise. Differentiate the number of output and
206 (const_unop): Likewise.
207 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
208 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
209 in cases that did the opposite.
211 2017-09-14 Richard Biener <rguenther@suse.de>
213 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
216 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
218 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
220 2017-09-14 Jakub Jelinek <jakub@redhat.com>
223 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
224 if and where to split a bb, except for splitting before debug insn
225 sequences followed by non-label real insn. Delete debug insns
226 in between basic blocks.
228 * combine.c (make_compound_operation_int): Formatting fixes.
230 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
231 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
232 * config/netbsd.h (LINK_EH_SPEC): Likewise.
233 * config/sol2.h (LINK_EH_SPEC): Likewise.
234 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
235 * config/s390/linux.h (LINK_SPEC): Likewise.
236 * config/freebsd.h (LINK_EH_SPEC): Likewise.
237 * config/openbsd.h (LINK_EH_SPEC): Likewise.
238 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
239 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
240 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
241 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
242 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
243 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
245 2017-09-13 Jakub Jelinek <jakub@redhat.com>
247 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
249 (ENDFILE_LINUX_SPEC): Likewise.
250 (LINK_EH_SPEC): Likewise.
251 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
252 (LINK_OS_LINUX_SPEC32): Likewise.
253 (LINK_OS_LINUX_SPEC64): Likewise.
254 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
255 (LINK_OS_LINUX_SPEC): Likewise.
257 2017-09-13 Martin Liska <mliska@suse.cz>
260 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
261 CASE_HIGH is NULL_TREE.
263 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hayward@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * target.def (secondary_memory_needed): New hook.
268 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
269 instead of SECONDARY_MEMORY_NEEDED.
270 (secondary_memory_needed_mode): Likewise.
271 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
272 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
273 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
274 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
275 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
276 * doc/tm.texi: Regenerate.
277 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
278 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
279 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
280 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
281 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
282 * config/i386/i386.c (inline_secondary_memory_needed): Put the
283 mode argument first and change the reg_class arguments to reg_class_t.
284 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
285 Make static. Update the call to inline_secondary_memory_needed.
286 (ix86_register_move_cost): Update the call to
287 inline_secondary_memory_needed.
288 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
289 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
291 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
292 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
294 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
295 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
296 * config/mips/mips.c (mips_secondary_memory_needed): Make static
297 and match hook interface. Add comment from mips.h.
298 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
299 * config/mmix/mmix.md (truncdfsf2): Refer to
300 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
302 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
303 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
304 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
305 (pa_secondary_memory_needed): New function.
306 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
307 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
308 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
309 (pdp11_secondary_memory_needed): Make static and match hook interface.
310 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
311 * config/powerpcspe/powerpcspe-protos.h
312 (rs6000_secondary_memory_needed_ptr): Delete.
313 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
315 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
316 (rs6000_option_override_internal): Assign to
317 targetm.secondary_memory_needed rather than
318 rs6000_secondary_memory_needed_ptr.
319 (rs6000_secondary_memory_needed): Match hook interface.
320 (rs6000_debug_secondary_memory_needed): Likewise.
321 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
322 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
323 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
324 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
325 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
326 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
328 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
329 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
330 (rs6000_option_override_internal): Assign to
331 targetm.secondary_memory_needed rather than
332 rs6000_secondary_memory_needed_ptr.
333 (rs6000_secondary_memory_needed): Match hook interface.
334 (rs6000_debug_secondary_memory_needed): Likewise.
335 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
336 * config/s390/s390.c (s390_secondary_memory_needed): New function.
337 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
338 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
339 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
340 (sparc_secondary_memory_needed): New function.
341 * lra-constraints.c (check_and_process_move): Refer to
342 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
344 (curr_insn_transform): Likewise.
345 (process_alt_operands): Use targetm.secondary_memory_needed
346 instead of TARGET_SECONDARY_MEMORY_NEEDED.
347 (check_secondary_memory_needed_p): Likewise.
348 (choose_split_class): Likewise.
349 * reload.c: Unconditionally include code that was previously
350 conditional on SECONDARY_MEMORY_NEEDED.
351 (push_secondary_reload): Use targetm.secondary_memory_needed
352 instead of TARGET_SECONDARY_MEMORY_NEEDED.
353 (push_reload): Likewise.
354 * reload1.c: Unconditionally include code that was previously
355 conditional on SECONDARY_MEMORY_NEEDED.
356 (choose_reload_regs): Use targetm.secondary_memory_needed
357 instead of TARGET_SECONDARY_MEMORY_NEEDED.
358 (gen_reload): Likewise.
359 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
361 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
362 Alan Hayward <alan.hayward@arm.com>
363 David Sherwood <david.sherwood@arm.com>
365 * target.def (secondary_memory_needed_mode): New hook:
366 * targhooks.c (default_secondary_memory_needed_mode): Declare.
367 * targhooks.h (default_secondary_memory_needed_mode): New function.
368 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
369 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
370 * doc/tm.texi: Regenerate.
371 * lra-constraints.c (check_and_process_move): Use
372 targetm.secondary_memory_needed_mode instead of
373 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
374 (curr_insn_transform): Likewise.
375 * reload.c (get_secondary_mem): Likewise.
376 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
377 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
379 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
380 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
381 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
382 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
383 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
385 * config/powerpcspe/powerpcspe-protos.h
386 (rs6000_secondary_memory_needed_mode): Delete.
387 * config/powerpcspe/powerpcspe.c
388 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
389 (rs6000_secondary_memory_needed_mode): Make static.
390 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
391 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
393 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
395 (rs6000_secondary_memory_needed_mode): Make static.
396 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
397 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
398 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
399 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
400 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
402 (sparc_secondary_memory_needed_mode): New function.
403 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
405 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
407 * config/aarch64/constraints.md (Umq): New constraint.
408 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
410 (mov<mode>): Update condition.
412 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
414 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
415 when bitposition is the same.
417 2017-09-13 Richard Biener <rguenther@suse.de>
419 * dwarf2out.c (output_die_symbol): Remove.
420 (output_die): Do not output a DIEs symbol.
422 2017-09-13 Richard Biener <rguenther@suse.de>
425 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
426 default-def to avoid breaking iterator update with the weird
427 interaction with cgraph_update_edges_for_call_stmt_node.
429 2017-09-13 Richard Biener <rguenther@suse.de>
431 * tree-cfg.c (verify_gimple_assign_binary): Add verification
432 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
433 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
434 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
436 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
438 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
439 Disable pc relative literal load irrespective of
440 TARGET_FIX_ERR_A53_84341 for default.
442 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
444 * config/sparc/sparc.c (output_return): Output the source location of
445 the insn in the delay slot, if any.
446 (output_sibcall): Likewise.
448 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
451 * common.opt (-static-pie): New alias.
452 (shared): Negate static-pie.
453 (-no-pie): Update help text.
455 (static-pie): New option.
456 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
458 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
459 (LINK_EH_SPEC): Likewise.
460 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
461 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
462 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
463 * gcc.c (LINK_COMMAND_SPEC): Likewise.
464 (init_gcc_specs): Likewise.
465 (init_spec): Likewise.
466 (display_help): Update help message for -pie.
467 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
470 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
472 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
473 (movdi_aarch64): Likewise.
474 (movti_aarch64): Likewise.
476 2017-09-12 Simon Wright <simon@pushface.org>
479 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
480 calculation of the minor version, always output as 0.
482 2017-09-12 Jakub Jelinek <jakub@redhat.com>
485 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
486 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
487 on it early, rather than manual conversion late. For
488 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
489 instead of performing manual conversion.
491 2017-09-12 Carl Love <cel@us.ibm.com>
493 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
494 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
496 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
497 VMULOSW): Add definitions.
498 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
499 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
500 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
501 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
502 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
504 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
506 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
508 (movti_aarch64): Likewise.
509 (movdf_aarch64): Likewise.
510 (movtf_aarch64): Likewise.
511 (load_pairdi): Likewise.
512 (store_pairdi): Likewise.
513 (load_pairdf): Likewise.
514 (store_pairdf): Likewise.
515 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
516 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
517 (ldr_got_small_<mode>): Likewise.
518 (ldr_got_small_28k_<mode>): Likewise.
519 (ldr_got_tiny): Likewise.
520 * config/aarch64/iterators.md (ldst_sz): New.
521 (ldpstp_sz): Likewise.
522 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
524 (thunderx_load): Split load_8 to load_16.
525 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
527 (thunderx2t99_storepair_basic): Split store_8 to store_16.
528 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
529 (xgene1_store_pair): Split store_8 to store_16.
530 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
531 (falkor_st_0_st_sd): Split store_8 to store_16.
533 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
535 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
536 and store1/2/3/4 to store_4/8/12/16.
537 * config/aarch64/aarch64.md: Update for rename.
538 * config/arm/arm.md: Likewise.: Likewise.
539 * config/arm/arm.c: Likewise.
540 * config/arm/thumb1.md: Likewise.
541 * config/arm/thumb2.md: Likewise.
542 * config/arm/vfp.md: Likewise.
543 * config/arm/arm-generic.md: Likewise.
544 * config/arm/arm1020e.md: Likewise.
545 * config/arm/arm1026ejs.md: Likewise.
546 * config/arm/arm1136jfs.md: Likewise.
547 * config/arm/arm926ejs.md: Likewise.
548 * config/arm/cortex-a15.md: Likewise.
549 * config/arm/cortex-a17.md: Likewise.
550 * config/arm/cortex-a5.md: Likewise.
551 * config/arm/cortex-a53.md: Likewise.
552 * config/arm/cortex-a57.md: Likewise.
553 * config/arm/cortex-a7.md: Likewise.
554 * config/arm/cortex-a8.md: Likewise.
555 * config/arm/cortex-a9.md: Likewise.
556 * config/arm/cortex-m4.md: Likewise.
557 * config/arm/cortex-m7.md: Likewise.
558 * config/arm/cortex-r4.md: Likewise.
559 * config/arm/exynos-m1.md: Likewise.
560 * config/arm/fa526.md: Likewise.
561 * config/arm/fa606te.md: Likewise.
562 * config/arm/fa626te.md: Likewise.
563 * config/arm/fa726te.md: Likewise.
564 * config/arm/fmp626.md: Likewise.
565 * config/arm/iwmmxt.md: Likewise.
566 * config/arm/ldmstm.md: Likewise.
567 * config/arm/marvell-pj4.md: Likewise.
568 * config/arm/xgene1.md: Likewise.
569 * config/aarch64/thunderx.md: Likewise.
570 * config/aarch64/thunderx2t99.md: Likewise.
571 * config/aarch64/falkor.md: Likewise.
573 2017-09-12 Martin Liska <mliska@suse.cz>
575 * attribs.c (private_lookup_attribute): New function.
576 * attribs.h (private_lookup_attribute): Declared here.
577 (lookup_attribute): Called from this place.
579 2017-09-12 Richard Biener <rguenther@suse.de>
581 PR tree-optimization/82157
582 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
583 stmts with side-effects.
585 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
586 Alan Hayward <alan.hayward@arm.com>
587 David Sherwood <david.sherwood@arm.com>
589 * target.def (hard_regno_nregs): New hook.
590 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
591 * targhooks.h (default_hard_regno_nregs): Declare.
592 * targhooks.c (default_hard_regno_nregs): New function.
593 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
594 (TARGET_HARD_REGNO_NREGS): ...this hook.
595 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
596 (CLASS_MAX_NREGS): Likewise.
597 * doc/tm.texi: Regenerate.
598 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
599 instead of HARD_REGNO_NREGS.
600 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
601 HARD_REGNO_NREGS in the comment.
602 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
603 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
604 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
605 Return an unsigned int.
606 (TARGET_HARD_REGNO_NREGS): Redefine.
607 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
608 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
609 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
610 (arc_hard_regno_nregs): New function.
611 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
612 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
613 (arm_hard_regno_nregs): New function.
614 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
615 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
616 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
617 (TARGET_HARD_REGNO_NREGS): Redefine.
618 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
619 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
620 (HARD_REGNO_NREGS): Delete.
621 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
622 (cr16_hard_regno_nregs): New function.
623 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
624 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
625 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
626 (cris_hard_regno_nregs): New function.
627 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
628 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
629 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
630 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
631 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
632 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
633 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
634 (frv_hard_regno_nregs): Make static. Take and return an
636 (frv_class_max_nregs): Remove outdated copy of documentation.
637 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
638 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
639 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
640 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
641 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
642 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
643 (TARGET_HARD_REGNO_NREGS): Redefine.
644 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
645 (CLASS_MAX_NREGS): Update comment.
646 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
647 (ia64_hard_regno_nregs): New function.
648 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
649 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
650 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
651 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
652 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
654 (m32c_hard_regno_nregs): Likewise. Make static.
655 (TARGET_HARD_REGNO_NREGS): Redefine.
656 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
657 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
658 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
659 (m68k_hard_regno_nregs): New function.
660 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
661 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
662 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
663 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
664 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
665 Take and return an unsigned int.
666 (TARGET_HARD_REGNO_NREGS): Redefine.
667 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
668 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
669 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
670 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
671 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
672 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
673 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
674 (msp430_hard_regno_nregs): Make static. Take and return an
676 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
677 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
678 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
679 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
680 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
681 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
682 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
683 (TARGET_HARD_REGNO_NREGS): Redefine.
684 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
685 (PA_HARD_REGNO_NREGS): ...this.
686 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
687 (PA_HARD_REGNO_NREGS): ...this.
688 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
689 (pa_hard_regno_nregs): New function.
690 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
691 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
692 (pdp11_hard_regno_nregs): New function.
693 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
694 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
695 (rs6000_hard_regno_nregs_hook): New function.
696 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
697 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
698 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
699 Take and return an unsigned int. Move earlier in file.
700 (TARGET_HARD_REGNO_NREGS): Redefine.
701 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
702 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
703 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
704 (rl78_hard_regno_nregs): Make static. Take and return an
706 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
707 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
708 (rs6000_hard_regno_nregs_hook): New function.
709 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
710 * config/rx/rx.c (rx_hard_regno_nregs): New function.
711 (TARGET_HARD_REGNO_NREGS): Redefine.
712 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
713 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
714 instead of HARD_REGNO_NREGS.
715 (s390_hard_regno_nregs): New function.
716 (s390_hard_regno_mode_ok): Add comment from s390.h.
717 (TARGET_HARD_REGNO_NREGS): Redefine.
718 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
719 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
720 (sh_hard_regno_nregs): New function.
721 (sh_pass_in_reg_p): Use it.
722 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
723 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
724 (sparc_hard_regno_nregs): New function.
725 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
726 * config/spu/spu.c (spu_hard_regno_nregs): New function.
727 (spu_function_arg_advance): Use it, supplying a valid register number.
728 (TARGET_HARD_REGNO_NREGS): Redefine.
729 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
730 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
731 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
732 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
733 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
734 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
735 (CLASS_MAX_NREGS): Remove copy of old documentation.
736 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
737 (visium_hard_regno_nregs): New function.
738 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
739 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
740 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
741 xtensa_hard_regno_nregs): New function.
742 * system.h (HARD_REGNO_NREGS): Poison.
744 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
746 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
747 hard_regno_nregs instead of HARD_REGNO_NREGS.
748 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
749 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
750 (c6x_expand_epilogue): Likewise.
751 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
752 (frv_read_iacc_argument): Likewise.
753 * config/sh/sh.c: Include regs.h.
754 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
755 (regs_used): Likewise.
756 (output_stack_adjust): Likewise.
757 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
758 * expmed.c: Include regs.h.
759 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
760 * ree.c: Include regs.h.
761 (combine_reaching_defs): Use hard_regno_nregs instead of
763 (add_removable_extension): Likewise.
765 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
767 * regs.h (hard_regno_nregs): Turn into a function.
768 (end_hard_regno): Update accordingly.
769 * caller-save.c (setup_save_areas): Likewise.
770 (save_call_clobbered_regs): Likewise.
771 (replace_reg_with_saved_mem): Likewise.
772 (insert_restore): Likewise.
773 (insert_save): Likewise.
774 * combine.c (can_change_dest_mode): Likewise.
775 (move_deaths): Likewise.
776 (distribute_notes): Likewise.
777 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
778 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
779 (rs6000_split_multireg_move): Likewise.
780 (rs6000_register_move_cost): Likewise.
781 (rs6000_memory_move_cost): Likewise.
782 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
783 (rs6000_split_multireg_move): Likewise.
784 (rs6000_register_move_cost): Likewise.
785 (rs6000_memory_move_cost): Likewise.
786 * cselib.c (cselib_reset_table): Likewise.
787 (cselib_lookup_1): Likewise.
788 * emit-rtl.c (set_mode_and_regno): Likewise.
789 * function.c (aggregate_value_p): Likewise.
790 * ira-color.c (setup_profitable_hard_regs): Likewise.
791 (check_hard_reg_p): Likewise.
792 (calculate_saved_nregs): Likewise.
793 (assign_hard_reg): Likewise.
794 (improve_allocation): Likewise.
795 (calculate_spill_cost): Likewise.
796 * ira-emit.c (modify_move_list): Likewise.
797 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
798 (ira_hard_reg_in_set_p): Likewise.
799 * ira.c (setup_reg_mode_hard_regset): Likewise.
800 (clarify_prohibited_class_mode_regs): Likewise.
801 (check_allocation): Likewise.
802 * lra-assigns.c (find_hard_regno_for_1): Likewise.
803 (lra_setup_reg_renumber): Likewise.
804 (setup_try_hard_regno_pseudos): Likewise.
805 (spill_for): Likewise.
806 (assign_hard_regno): Likewise.
807 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
808 * lra-constraints.c (in_class_p): Likewise.
809 (lra_constraint_offset): Likewise.
810 (simplify_operand_subreg): Likewise.
811 (lra_constraints): Likewise.
812 (split_reg): Likewise.
813 (split_if_necessary): Likewise.
814 (invariant_p): Likewise.
815 (inherit_in_ebb): Likewise.
816 * lra-lives.c (process_bb_lives): Likewise.
817 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
818 (get_hard_regs): Likewise.
819 (do_remat): Likewise.
820 * lra-spills.c (assign_spill_hard_regs): Likewise.
821 * mode-switching.c (create_pre_exit): Likewise.
822 * postreload.c (reload_combine_recognize_pattern): Likewise.
823 * recog.c (peep2_find_free_register): Likewise.
824 * regcprop.c (kill_value_regno): Likewise.
825 (set_value_regno): Likewise.
826 (copy_value): Likewise.
827 (maybe_mode_change): Likewise.
828 (find_oldest_value_reg): Likewise.
829 (copyprop_hardreg_forward_1): Likewise.
830 * regrename.c (check_new_reg_p): Likewise.
831 (regrename_do_replace): Likewise.
832 * reload.c (push_reload): Likewise.
833 (combine_reloads): Likewise.
834 (find_dummy_reload): Likewise.
835 (operands_match_p): Likewise.
836 (find_reloads): Likewise.
837 (find_equiv_reg): Likewise.
838 (reload_adjust_reg_for_mode): Likewise.
839 * reload1.c (count_pseudo): Likewise.
840 (count_spilled_pseudo): Likewise.
841 (find_reg): Likewise.
842 (clear_reload_reg_in_use): Likewise.
843 (free_for_value_p): Likewise.
844 (allocate_reload_reg): Likewise.
845 (choose_reload_regs): Likewise.
846 (reload_adjust_reg_for_temp): Likewise.
847 (emit_reload_insns): Likewise.
848 (delete_output_reload): Likewise.
849 * rtlanal.c (subreg_get_info): Likewise.
850 * sched-deps.c (sched_analyze_reg): Likewise.
851 * sel-sched.c (init_regs_for_mode): Likewise.
852 (mark_unavailable_hard_regs): Likewise.
853 (choose_best_reg_1): Likewise.
854 (verify_target_availability): Likewise.
855 * valtrack.c (dead_debug_insert_temp): Likewise.
856 * var-tracking.c (track_loc_p): Likewise.
857 (emit_note_insn_var_location): Likewise.
858 * varasm.c (make_decl_rtl): Likewise.
859 * reginfo.c (choose_hard_reg_mode): Likewise.
860 (init_reg_modes_target): Refer directly to
861 this_target_regs->x_hard_regno_nregs.
863 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
865 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
866 instead of hard_regno_nregs.
868 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
870 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
871 end_hard_regno instead of hard_regno_nregs.
872 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
873 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
874 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
875 * ira-color.c (improve_allocation): Likewise.
876 * lra-assigns.c (find_hard_regno_for_1): Likewise.
877 * lra-lives.c (mark_regno_live): Likewise.
878 (mark_regno_dead): Likewise.
879 * lra-remat.c (operand_to_remat): Likewise.
880 * lra.c (collect_non_operand_hard_regs): Likewise.
881 * postreload.c (reload_combine_note_store): Likewise.
882 (move2add_valid_value_p): Likewise.
883 * reload.c (regno_clobbered_p): Likewise.
885 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
887 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
889 * config/v850/v850.c (v850_reorg): Likewise.
890 * reload.c (refers_to_regno_for_reload_p): Likewise.
891 (find_equiv_reg): Likewise.
892 * reload1.c (reload_reg_reaches_end_p): Likewise.
894 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
896 * caller-save.c (add_used_regs): Use REG_NREGS instead of
898 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
899 * config/arm/arm.c (output_move_neon): Likewise.
900 (arm_attr_length_move_neon): Likewise.
901 (neon_split_vcombine): Likewise.
902 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
903 (c6x_mark_reg_written): Likewise.
904 (c6x_dwarf_register_span): Likewise.
905 * config/i386/i386.c (ix86_save_reg): Likewise.
906 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
907 (rws_access_reg): Likewise.
908 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
909 * mode-switching.c (create_pre_exit): Likewise.
910 * ree.c (combine_reaching_defs): Likewise.
911 (add_removable_extension): Likewise.
912 * regcprop.c (find_oldest_value_reg): Likewise.
913 (copyprop_hardreg_forward_1): Likewise.
914 * reload.c (reload_inner_reg_of_subreg): Likewise.
915 (push_reload): Likewise.
916 (combine_reloads): Likewise.
917 (find_dummy_reload): Likewise.
918 (reload_adjust_reg_for_mode): Likewise.
919 * reload1.c (find_reload_regs): Likewise.
920 (forget_old_reloads_1): Likewise.
921 (reload_reg_free_for_value_p): Likewise.
922 (reload_adjust_reg_for_temp): Likewise.
923 (emit_reload_insns): Likewise.
924 (delete_output_reload): Likewise.
925 * sel-sched.c (choose_best_reg_1): Likewise.
926 (choose_best_pseudo_reg): Likewise.
928 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
929 Alan Hayward <alan.hayward@arm.com>
930 David Sherwood <david.sherwood@arm.com>
932 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
933 * target.def (slow_unaligned_access): New hook.
934 * targhooks.h (default_slow_unaligned_access): Declare.
935 * targhooks.c (default_slow_unaligned_access): New function.
936 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
937 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
938 * doc/tm.texi: Regenerate.
939 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
940 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
941 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
943 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
944 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
946 (rs6000_slow_unaligned_access): New function.
947 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
948 (expand_block_compare): Likewise.
949 (expand_strn_compare): Likewise.
950 (rs6000_rtx_costs): Likewise.
951 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
952 (riscv_slow_unaligned_access): Likewise.
953 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
954 (riscv_slow_unaligned_access_p): ...this and make static.
955 (riscv_option_override): Update accordingly.
956 (riscv_slow_unaligned_access): New function.
957 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
958 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
959 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
960 (rs6000_slow_unaligned_access): New function.
961 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
962 (rs6000_rtx_costs): Likewise.
963 * config/rs6000/rs6000-string.c (expand_block_compare)
964 (expand_strn_compare): Use targetm.slow_unaligned_access instead
965 of SLOW_UNALIGNED_ACCESS.
966 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
967 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
968 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
969 of SLOW_UNALIGNED_ACCESS.
970 * expmed.c (simple_mem_bitfield_p): Likewise.
971 * expr.c (alignment_for_piecewise_move): Likewise.
972 (emit_group_load_1): Likewise.
973 (emit_group_store): Likewise.
974 (copy_blkmode_from_reg): Likewise.
975 (emit_push_insn): Likewise.
976 (expand_assignment): Likewise.
977 (store_field): Likewise.
978 (expand_expr_real_1): Likewise.
979 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
980 * lra-constraints.c (simplify_operand_subreg): Likewise.
981 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
982 * gimple-ssa-store-merging.c: Likewise in block comment at start
984 * tree-ssa-strlen.c: Include target.h.
985 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
986 of SLOW_UNALIGNED_ACCESS.
987 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
989 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
991 PR rtl-optimization/82185
992 * expmed.c (emit_store_flag_int): Only test tem if it has been
995 2017-09-12 Richard Biener <rguenther@suse.de>
998 * match.pd ((FTYPE) N CMP CST): Fix typo.
1000 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1002 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1004 (mips_near_type_p): Add 'short_call' attribute as a synonym
1006 * doc/extend.texi (short_call): Document new function attribute.
1008 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1011 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1012 assertion check that in the condition.
1013 (get_atomic_generic_size): Likewise. Before testing if parameter
1014 has pointer type, if it has array type, call for C++
1015 default_conversion to perform array-to-pointer conversion.
1017 2017-09-12 Richard Biener <rguenther@suse.de>
1019 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1020 for operations we cannot scalarize.
1022 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1024 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1025 vectors heap vectors. Clean up comments.
1026 Make visited_bbs a reference.
1027 (profitable_jump_thread_path): Make GC
1028 vectors heap vectors. Clean up comments.
1030 (convert_and_register_jump_thread_path): Make GC vectors heap
1032 (check_subpath_and_update_thread_path): Same. Clean up comments.
1033 Make visited_bbs a reference.
1034 (handle_phi): Abstract common code to to
1035 register_jump_thread_path_if_profitable.
1036 Rename VAR_BB to DEF_BB.
1038 Make GC vectors heap vectors.
1039 Make visited_bbs a reference.
1040 (handle_assignment): Same.
1041 (register_jump_thread_path_if_profitable): New.
1042 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1044 Make GC vectors heap vectors. Clean up comments.
1045 Make visited_bbs a reference.
1046 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1047 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1050 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1053 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1054 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1056 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1058 Revert r251800 and r251799.
1060 2017-09-11 Martin Jambor <mjambor@suse.cz>
1063 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1064 arguments in advance.
1065 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1066 use it to find predecessor edges.
1067 (naive_outof_ssa): Collect vector of predecessors.
1069 2017-09-08 Jason Merrill <jason@redhat.com>
1071 PR c++/70029 - ICE with ref-qualifier and -flto
1072 * langhooks.h (struct lang_hooks_for_types): Add
1073 copy_lang_qualifiers.
1074 * attribs.c (build_type_attribute_qual_variant): Use it.
1075 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1077 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1078 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1080 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1083 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1084 (*mulsi3_sp64): New instruction.
1085 (mulsi3): New expander.
1087 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1089 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1091 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1093 * sancov.c: Include memmodel.h.
1095 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1098 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1101 2017-09-07 Carl Love <cel@us.ibm.com>
1103 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1104 the sldi instruction.
1106 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1108 * sancov.c: Include tm_p.h.
1110 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1113 * output.h (switch_to_other_text_partition): New declaration.
1114 * varasm.c (switch_to_other_text_partition): New function.
1115 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1116 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1117 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1118 to the other text partition before emitting LCL label and switch back
1119 after emitting the word after it.
1121 2017-09-07 Richard Biener <rguenther@suse.de>
1123 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1124 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1125 critical edges here, after loop init.
1126 (pass_data_pre): Remove PROP_no_crit_edges flags.
1127 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1128 for valueization of call args to avoid leaking VN_TOP.
1129 (visit_use): Assert we do not visit default defs.
1130 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1131 Use error_mark_node to more easily detect leaking VN_TOP.
1132 All default-defs are varying, not VN_TOP. Mark them visited.
1133 (run_scc_vn): Make code match comment.
1135 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1137 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1138 OPTION_MASK_FLOAT128_KEYWORD.
1139 (POWERPC_MASKS): Likewise.
1140 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1141 support for the -mfloat128-type option, and make -mfloat128
1142 default on PowerPC Linux systems. Define or undefine
1143 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1144 Define __float128 to be __ieee128 if IEEE 128-bit support is
1145 enabled, or undefine it.
1146 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1147 Delete defining __FLOAT128_TYPE__.
1148 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1149 -mfloat128-type option and make -mfloat128 default on PowerPC
1151 (TARGET_FLOAT128_TYPE): Likewise.
1152 (-mfloat128-type): Likewise.
1153 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1154 Delete the -mfloat128-type option and make -mfloat128 default on
1155 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1156 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1157 128-bit floating point is enabled. Change tests from using
1158 -mfloat128-type to -mfloat128.
1159 (rs6000_mangle_type): Use the correct mangling for the __float128
1160 type even if normal long double is restricted to 64-bits.
1161 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1163 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1164 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1165 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1167 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1168 documentation for -mfloat128.
1170 2017-09-06 Olivier Hainque <hainque@adacore.com>
1172 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1174 2017-09-06 Wish Wu <wishwu007@gmail.com>
1175 Jakub Jelinek <jakub@redhat.com>
1177 * asan.c (initialize_sanitizer_builtins): Add
1178 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1179 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1180 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1181 BT_FN_VOID_UINT64_PTR variables.
1182 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1183 (BT_FN_VOID_UINT16_UINT16): Likewise.
1184 (BT_FN_VOID_UINT32_UINT32): Likewise.
1185 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1186 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1187 (BT_FN_VOID_UINT64_PTR): Likewise.
1188 * common.opt (flag_sanitize_coverage): New variable.
1189 (fsanitize-coverage=trace-pc): Remove.
1190 (fsanitize-coverage=): Add.
1191 * flag-types.h (enum sanitize_coverage_code): New enum.
1192 * fold-const.c (fold_range_test): Disable non-short-circuit
1193 optimization if flag_sanitize_coverage.
1194 (fold_truth_andor): Likewise.
1195 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1196 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1197 (coverage_sanitizer_opts): New array.
1198 (get_closest_sanitizer_option): Add OPTS argument, handle also
1199 OPT_fsanitize_coverage_.
1200 (parse_sanitizer_options): Adjusted to also handle
1201 OPT_fsanitize_coverage_.
1202 (common_handle_option): Add OPT_fsanitize_coverage_.
1203 * sancov.c (instrument_comparison, instrument_switch): New function.
1204 (sancov_pass): Add trace-cmp support.
1205 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1206 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1207 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1208 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1209 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1210 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1211 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1212 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1213 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1214 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1216 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1218 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1219 error. Only quit immediately if parsing is complete.
1220 (BEGIN): Initialize fatal_err and parse_done.
1221 (begin fpu, end fpu): Check number of arguments.
1222 (begin arch, end arch): Likewise.
1223 (begin cpu, end cpu): Likewise.
1224 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1225 (optalias): Likewise.
1227 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1229 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1230 * config/arm/arm-isa.h: Delete. Move definitions to ...
1231 * arm-cpus.in: ... here. Use new feature and fgroup values.
1232 * config/arm/arm.c (arm_option_override): Use lower case for feature
1234 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1235 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1236 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1238 (print_isa_bits_for): New function.
1239 (gen_isa): New function.
1240 (gen_comm_data): Use print_isa_bits_for.
1241 (define feature): New keyword.
1242 (define fgroup): New keyword.
1243 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1244 (arm-isa.h): Add rule to generate file.
1245 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1246 case for feature bit names.
1248 2017-09-06 Richard Biener <rguenther@suse.de>
1250 * tree-ssa-pre.c (NECESSARY): Remove.
1251 (create_expression_by_pieces): Do not touch pass-local flags.
1252 (insert_into_preds_of_block): Likewise.
1253 (do_pre_regular_insertion): Likewise.
1254 (eliminate_insert): Likewise.
1255 (eliminate_dom_walker::before_dom_children): Likewise.
1256 (fini_eliminate): Do not look at inserted_exprs.
1257 (mark_operand_necessary): Remove.
1258 (remove_dead_inserted_code): Replace with simple work-list
1259 algorithm based on inserted_exprs and SSA uses.
1260 (pass_pre::execute): Re-order fini_eliminate and
1261 remove_dead_inserted_code.
1263 2017-09-06 Olivier Hainque <hainque@adacore.com>
1265 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1266 for VxWorks 7. Adjust surrounding comments.
1268 2017-09-06 Richard Biener <rguenther@suse.de>
1270 * gimple-ssa-strength-reduction.c
1271 (find_candidates_dom_walker::before_dom_children): Also allow
1274 2017-09-06 Richard Biener <rguenther@suse.de>
1276 PR tree-optimization/82108
1277 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1278 for gap in the non-permutation SLP case.
1280 2017-09-06 Martin Jambor <mjambor@suse.cz>
1282 PR tree-optimization/82078
1283 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1284 add_access_to_work_queue...
1285 (build_accesses_from_assign): ...here.
1286 (propagate_all_subaccesses): Make sure racc is the group
1287 representative, if there is one.
1289 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1292 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1295 2017-09-06 Richard Biener <rguenther@suse.de>
1297 * gimple-ssa-strength-reduction.c
1298 (find_candidates_dom_walker::before_doom_children): Use a
1299 type and not a mode check.
1301 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1304 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1306 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1308 2017-09-05 Jeff Law <law@redhat.com>
1310 PR tree-optimization/64910
1311 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1312 cases where we have 3 or more operands.
1314 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1317 * omp-low.c (lower_omp_for): Recompute tree invariant if
1318 gimple_omp_for_initial/final is ADDR_EXPR.
1321 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1322 into gimple val before gimplification fo the COND_EXPR.
1324 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1326 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1327 REGION_COPY argument.
1328 (thread_through_all_blocks): Remove unused argument to
1329 duplicate_thread_path.
1331 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1332 Alan Hayward <alan.hayward@arm.com>
1333 David Sherwood <david.sherwood@arm.com>
1335 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1336 Take a scalar_mode rather than a machine_mode.
1337 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1338 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1339 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1340 (aarch64_gen_adjusted_ldpstp): Likewise.
1341 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1343 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1344 Alan Hayward <alan.hayward@arm.com>
1345 David Sherwood <david.sherwood@arm.com>
1347 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1348 Take a scalar_int_mode instead of a machine_mode.
1349 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1350 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1351 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1352 (aarch64_simd_attr_length_rglist): Delete.
1353 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1354 a scalar_int_mode instead of a machine_mode.
1355 (aarch64_add_offset): Likewise.
1356 (aarch64_internal_mov_immediate): Likewise
1357 (aarch64_add_constant_internal): Likewise.
1358 (aarch64_add_constant): Likewise.
1359 (aarch64_movw_imm): Likewise.
1360 (aarch64_rtx_arith_op_extract_p): Likewise.
1361 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1362 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1363 Remove assert that the mode isn't a vector.
1364 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1365 (aarch64_expand_mov_immediate): Update calls after above changes.
1366 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1367 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1368 (aarch64_move_imm): Likewise.
1369 (aarch64_can_const_movi_rtx_p): Likewise.
1370 (aarch64_strip_extend): Likewise.
1371 (aarch64_extr_rtx_p): Likewise.
1372 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1373 a CONST_INT when the mode parameter is VOIDmode.
1374 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1376 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1378 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1379 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1380 (bitwise_type_for_mode): Update accordingly.
1382 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1384 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1385 * stor-layout.c (mode_for_size_tree): Likewise.
1386 (mode_for_array): Update accordingly.
1387 (layout_decl): Likewise.
1388 (compute_record_mode): Likewise. Only set the mode once.
1390 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1392 * target.def (get_mask_mode): Change return type to opt_mode.
1394 * doc/tm.texi: Regenerate.
1395 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1396 * targhooks.c (default_get_mask_mode): Likewise.
1397 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1398 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1399 targetm.get_mask_mode.
1400 * tree.c (build_truth_vector_type): Likewise.
1402 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1404 * machmode.h (mode_for_vector): Return an opt_mode.
1405 * stor-layout.c (mode_for_vector): Likewise.
1406 (mode_for_int_vector): Update accordingly.
1407 (layout_type): Likewise.
1408 * config/i386/i386.c (emit_memmov): Likewise.
1409 (ix86_expand_set_or_movmem): Likewise.
1410 (ix86_expand_vector_init): Likewise.
1411 (ix86_get_mask_mode): Likewise.
1412 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1414 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1415 * expmed.c (extract_bit_field_1): Likewise.
1416 * expr.c (expand_expr_real_2): Likewise.
1417 * optabs-query.c (can_vec_perm_p): Likewise.
1418 (can_vec_mask_load_store_p): Likewise.
1419 * optabs.c (expand_vec_perm): Likewise.
1420 * targhooks.c (default_get_mask_mode): Likewise.
1421 * tree-vect-stmts.c (vectorizable_store): Likewise.
1422 (vectorizable_load): Likewise.
1423 (get_vectype_for_scalar_type_and_size): Likewise.
1425 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1427 * machmode.h (mode_for_int_vector): New function.
1428 * stor-layout.c (mode_for_int_vector): Likewise.
1429 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1430 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1431 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1432 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1433 (s390_expand_vcond): Likewise.
1435 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1437 * machmode.h (opt_machine_mode): New type.
1438 (opt_mode<T>): Allow construction from anything that can be
1440 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1441 (mode_for_size): Return an opt_machine_mode.
1442 * stor-layout.c (mode_for_size): Likewise.
1443 (mode_for_size_tree): Update call accordingly.
1444 (bitwise_mode_for_mode): Likewise.
1445 (make_fract_type): Likewise.
1446 (make_accum_type): Likewise.
1447 * caller-save.c (replace_reg_with_saved_mem): Update call
1449 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1450 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1451 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1452 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1453 * expmed.c (extract_bit_field_1): Likewise.
1454 * reload.c (get_secondary_mem): Likewise.
1455 * varasm.c (assemble_integer): Likewise.
1456 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1459 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1461 * machmode.h (decimal_float_mode_for_size): New function.
1462 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1463 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1464 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1465 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1466 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1467 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1469 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1471 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1472 (get_builtin_sync_mode): Likewise.
1473 (expand_ifn_atomic_compare_exchange): Likewise.
1474 (expand_builtin_atomic_clear): Likewise.
1475 (expand_builtin_atomic_test_and_set): Likewise.
1476 (fold_builtin_atomic_always_lock_free): Likewise.
1477 * calls.c (compute_argument_addresses): Likewise.
1478 (emit_library_call_value_1): Likewise.
1479 (store_one_arg): Likewise.
1480 * combine.c (combine_instructions): Likewise.
1481 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1482 * config/arm/arm.c (arm_function_value): Likewise.
1483 (aapcs_allocate_return_reg): Likewise.
1484 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1485 * config/i386/i386.c (construct_container): Likewise.
1486 (ix86_gimplify_va_arg): Likewise.
1487 (ix86_expand_sse_cmp): Likewise.
1488 (emit_memmov): Likewise.
1489 (emit_memset): Likewise.
1490 (expand_small_movmem_or_setmem): Likewise.
1491 (ix86_expand_pextr): Likewise.
1492 (ix86_expand_pinsr): Likewise.
1493 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1494 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1496 * config/mips/mips.c (mips_function_value_1) Likewise.
1497 (mips_block_move_straight): Likewise.
1498 (mips_expand_ins_as_unaligned_store): Likewise.
1499 * config/powerpcspe/powerpcspe.c
1500 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1501 (rs6000_darwin64_record_arg_flush): Likewise.
1502 * config/rs6000/rs6000.c
1503 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1504 (rs6000_darwin64_record_arg_flush): Likewise.
1505 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1506 (sparc_function_value_1): Likewise.
1507 * config/spu/spu.c (adjust_operand): Likewise.
1508 (spu_emit_branch_or_set): Likewise.
1509 (arith_immediate_p): Likewise.
1510 * emit-rtl.c (gen_lowpart_common): Likewise.
1511 * expr.c (expand_expr_real_1): Likewise.
1512 * function.c (assign_parm_setup_block): Likewise.
1513 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1514 * reload1.c (alter_reg): Likewise.
1515 * stor-layout.c (mode_for_vector): Likewise.
1516 (layout_type): Likewise.
1518 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1520 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1521 (spu_convert_move): Likewise.
1522 * lower-subreg.c (resolve_simple_move): Likewise.
1524 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1527 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1528 define_insn to a define_expand.
1529 (altivec_vsum2sws_direct): New define_insn.
1530 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1532 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1534 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1537 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1542 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1543 realign stack if stack alignment needed is less than incoming
1546 2017-09-05 Marek Polacek <polacek@redhat.com>
1549 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1552 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1554 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1556 2017-09-05 Richard Biener <rguenther@suse.de>
1558 PR tree-optimization/82084
1559 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1561 2017-09-05 Richard Biener <rguenther@suse.de>
1563 PR tree-optimization/82102
1564 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1566 2017-09-05 Martin Liska <mliska@suse.cz>
1568 PR tree-optimization/82032
1569 * tree-cfg.c (generate_range_test): New function.
1570 * tree-cfg.h (generate_range_test): Declared here.
1571 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1572 (cleanup_control_expr_graph): Use it.
1573 * tree-switch-conversion.c (try_switch_expansion): Remove
1575 (emit_case_nodes): Use generate_range_test.
1577 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1580 * config/i386/i386.md (*<btsc><mode>_mask): Add
1581 TARGET_USE_BT to insn constraint.
1582 (*btr<mode>_mask): Ditto.
1584 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1586 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1587 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1589 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1592 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1593 TARGET_NEON and TARGET_IWMMXT.
1594 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1595 TARGET_NEON and TARGET_IWMMXT.
1596 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1598 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1600 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1601 (ix86_rewrite_tls_address): Ditto.
1602 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1603 (ix86_rewrite_tls_address_1): Ditto.
1604 (ix86_rewrite_tls_address): Ditto.
1605 * config/i386/predicates.md (tls_address_pattern): New predicate.
1606 * config/i386/i386.md (TLS address splitter): New splitter.
1608 2017-09-04 Richard Biener <rguenther@suse.de>
1610 PR tree-optimization/82084
1611 * fold-const.h (can_native_encode_string_p): Declare.
1612 * fold-const.c (can_native_encode_string_p): Factor out from ...
1613 (native_encode_string): ... here.
1614 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1615 vectorizing stores from constants we later cannot handle.
1617 2017-09-04 Marek Polacek <polacek@redhat.com>
1620 * doc/invoke.texi: Update -Wtautological-compare documentation.
1622 2017-09-04 Jeff Law <law@redhat.com>
1624 PR tree-optimization/64910
1625 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1626 swap the first and last operand if the last is a constant.
1628 2017-09-04 Marek Polacek <polacek@redhat.com>
1631 * convert.c (do_narrow): When sanitizing signed integer overflows,
1632 bail out for signed types.
1633 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1635 2017-09-04 Richard Biener <rguenther@suse.de>
1637 PR tree-optimization/82060
1638 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1639 Move devirtualization after stmt folding and before EH/AB/noreturn
1640 cleanup to get the stmt refs canonicalized. Use a bool instead
1641 of gimple_modified_p since that doesn't work for NOPs. Schedule
1642 NOPs generated by folding for removal.
1644 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1645 Alan Hayward <alan.hayward@arm.com>
1646 David Sherwood <david.sherwood@arm.com>
1648 * coretypes.h (pad_direction): New enum.
1649 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1650 (FUNCTION_ARG_PADDING): Likewise.
1651 * target.def (function_arg_padding): New hook.
1652 * targhooks.h (default_function_arg_padding): Declare.
1653 * targhooks.c (default_function_arg_padding): New function.
1654 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1655 (TARGET_FUNCTION_ARG_PADDING): ...this.
1656 * doc/tm.texi: Regenerate.
1657 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1658 instead of direction.
1659 (compute_argument_addresses): Likewise.
1660 (load_register_parameters): Likewise.
1661 (emit_library_call_value_1): Likewise.
1662 (store_one_arg): Use targetm.calls.function_arg_padding instead
1663 of FUNCTION_ARG_PADDING.
1664 (must_pass_in_stack_var_size_or_pad): Likewise.
1665 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1666 (emit_group_store): Likewise.
1667 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1668 instead of FUNCTION_ARG_PADDING.
1669 (emit_push_insn): Likewise, and propagate enum change throughout
1671 * function.h (direction): Delete.
1672 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1674 * function.c (assign_parm_find_stack_rtl): Likewise.
1675 (assign_parm_setup_block_p): Likewise.
1676 (assign_parm_setup_block): Likewise.
1677 (gimplify_parameters): Likewise.
1678 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1679 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1681 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1682 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1683 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1684 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1685 (aarch64_function_arg_padding): ...this new function.
1686 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1687 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1688 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1689 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1690 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1691 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1692 (arm_pad_arg_upward): Replace with...
1693 (arm_function_arg_padding): ...this new function.
1694 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1696 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1697 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1698 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1699 (ia64_hpux_function_arg_padding): Replace with...
1700 (ia64_function_arg_padding): ...this new function. Use pad_direction
1701 instead of direction. Check for TARGET_HPUX.
1702 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1703 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1704 (iq2000_function_arg_padding): New function.
1705 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1706 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1707 (mips_function_arg_padding): ...this new function.
1708 (mips_pad_reg_upward): Update accordingly.
1709 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1710 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1711 targetm.calls.function_arg_padding.
1712 (FUNCTION_ARG_PADDING): Delete.
1713 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1714 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1715 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1716 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1717 (nios2_block_reg_padding): Return pad_direction instead of direction.
1718 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1719 instead of direction.
1720 (nios2_function_arg_padding): Likewise. Make static.
1721 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1722 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1723 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1724 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1725 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1726 (pa_function_arg_padding): Make static. Return pad_direction instead
1728 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1729 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1730 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1731 instead of direction. Use targetm.calls.function_arg_padding.
1732 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1733 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1734 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1735 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1736 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1738 (function_arg_padding): Rename to...
1739 (rs6000_function_arg_padding): ...this. Make static. Return
1740 pad_direction instead of direction.
1741 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1742 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1743 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1744 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1745 instead of direction. Use targetm.calls.function_arg_padding.
1746 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1747 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1748 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1749 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1750 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1751 (function_arg_padding): Rename to...
1752 (rs6000_function_arg_padding): ...this. Make static. Return
1753 pad_direction instead of direction.
1754 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1755 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1756 * config/s390/s390.c (s390_function_arg_padding): New function.
1757 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1758 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1759 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1760 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1761 (function_arg_padding): Rename to...
1762 (sparc_function_arg_padding): ...this. Make static. Return
1763 pad_direction instead of direction.
1764 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1765 * config/spu/spu.c (spu_function_arg_padding): New function.
1766 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1767 * system.h (FUNCTION_ARG_PADDING): Poison.
1769 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1770 Alan Hayward <alan.hayward@arm.com>
1771 David Sherwood <david.sherwood@arm.com>
1773 * target.def (modes_tieable_p): New hook.
1774 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1775 (TARGET_MODES_TIEABLE_P): ...this.
1776 * doc/tm.texi.in: Regenerate.
1777 * hooks.h (hook_bool_mode_mode_true): Declare.
1778 * hooks.c (hook_bool_mode_mode_true): New function.
1779 * combine.c (subst): Use targetm.modes_tieable_p instead of
1781 * dse.c (find_shift_sequence): Likewise.
1782 * expmed.c (extract_low_bits): Likewise.
1783 * lower-subreg.c: Include target.h.
1784 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1786 * rtlanal.c (rtx_cost): Likewise.
1787 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1788 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1789 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1790 (TARGET_MODES_TIEABLE_P): Redefine.
1791 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1792 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1793 (TARGET_MODES_TIEABLE_P): Redefine.
1794 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1795 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1796 (arc_modes_tieable_p): New function.
1797 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1798 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1799 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1800 (arm_modes_tieable_p): Make static.
1801 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1802 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1803 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1804 (TARGET_MODES_TIEABLE_P): Redefine.
1805 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1806 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1807 (TARGET_MODES_TIEABLE_P): Redefine.
1808 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1809 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1810 (cr16_modes_tieable_p): New function.
1811 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1812 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1813 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1814 (TRULY_NOOP_TRUNCATION): Update comment.
1815 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1816 (TRULY_NOOP_TRUNCATION): Update comment.
1817 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1818 (frv_modes_tieable_p): New function.
1819 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1820 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1821 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1822 (TARGET_MODES_TIEABLE_P): Redefine.
1823 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1824 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1825 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1826 (TARGET_MODES_TIEABLE_P): Redefine.
1827 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1828 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1829 (ia64_modes_tieable_p): New function.
1830 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1831 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1832 (iq2000_modes_tieable_p): New function.
1833 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1834 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1835 (lm32_modes_tieable_p): New function.
1836 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1837 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1838 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1839 (TARGET_MODES_TIEABLE_P): Redefine.
1840 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1841 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1842 (m32r_modes_tieable_p): New function.
1843 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1844 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1845 (m68k_modes_tieable_p): New function.
1846 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1847 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1848 (mcore_modes_tieable_p): New function.
1849 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1850 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1852 (TARGET_MODES_TIEABLE_P): Redefine.
1853 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1854 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1855 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1856 (TARGET_MODES_TIEABLE_P): Redefine.
1857 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1858 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1859 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1860 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1861 (mn10300_modes_tieable_p): ...this and make static.
1862 (TARGET_MODES_TIEABLE_P): Redefine.
1863 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1864 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1865 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1866 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1867 (msp430_modes_tieable_p): Make static.
1868 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1869 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1870 (TARGET_MODES_TIEABLE_P): Redefine.
1871 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1872 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1873 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1874 (TARGET_MODES_TIEABLE_P): Redefine.
1875 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1876 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1877 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1878 (TARGET_MODES_TIEABLE_P): Redefine.
1879 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1880 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1881 (pdp11_modes_tieable_p): New function.
1882 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1883 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1884 (rs6000_modes_tieable_p): New function.
1885 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1886 * config/powerpcspe/powerpcspe.md: Update comment.
1887 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1888 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1889 (TARGET_MODES_TIEABLE_P): Redefine.
1890 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1891 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1892 (rl78_modes_tieable_p): New function.
1893 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1894 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1895 (rs6000_modes_tieable_p): New function.
1896 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1897 * config/rs6000/rs6000.md: Update comment.
1898 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1899 * config/rx/rx.c (rx_modes_tieable_p): New function.
1900 (TARGET_MODES_TIEABLE_P): Redefine.
1901 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1902 * config/s390/s390.c (s390_modes_tieable_p): New function.
1903 (TARGET_MODES_TIEABLE_P): Redefine.
1904 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1905 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1906 (sh_modes_tieable_p): New function.
1907 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1908 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1909 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1910 (sparc_modes_tieable_p): Make static.
1911 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1912 * config/spu/spu.c (spu_modes_tieable_p): New function.
1913 (TARGET_MODES_TIEABLE_P): Redefine.
1914 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1915 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1916 (TARGET_MODES_TIEABLE_P): Redefine.
1917 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1918 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1919 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1920 * config/v850/v850.c (v850_modes_tieable_p): New function.
1921 (TARGET_MODES_TIEABLE_P): Redefine.
1922 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1923 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1924 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1925 (visium_modes_tieable_p): New function.
1926 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1927 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1928 (xtensa_modes_tieable_p): New function.
1929 * system.h (MODES_TIEABLE_P): Poison.
1931 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1932 Alan Hayward <alan.hayward@arm.com>
1933 David Sherwood <david.sherwood@arm.com>
1935 * target.def (hard_regno_mode_ok): New hook.
1936 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1937 (TARGET_HARD_REGNO_MODE_OK): ...this.
1938 * doc/tm.texi.in: Regenerate.
1939 * hooks.h (hook_bool_uint_mode_true): Declare.
1940 * hooks.c (hook_bool_uint_mode_true): New function.
1941 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1943 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1944 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1945 instead of HARD_REGNO_MODE_OK.
1946 * caller-save.c: Include target.h.
1947 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1949 * combine.c (can_combine_p): Likewise.
1950 (combinable_i3pat): Likewise.
1951 (can_change_dest_mode): Likewise.
1952 * expr.c (init_expr_target): Likewise.
1953 (convert_move): Likewise.
1954 (convert_modes): Likewise.
1955 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1956 (setup_prohibited_mode_move_regs): Likewise.
1957 * ira.h (target_ira): Likewise.
1958 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1959 * lra-constraints.c (process_alt_operands): Likewise.
1960 (split_reg): Likewise.
1961 * recog.c (peep2_find_free_register): Likewise.
1962 * ree.c (combine_reaching_defs): Likewise.
1963 * regcprop.c (maybe_mode_change): Likewise.
1964 * reginfo.c (init_reg_sets_1): Likewise.
1965 (choose_hard_reg_mode): Likewise.
1966 (simplifiable_subregs): Likewise.
1967 * regrename.c (check_new_reg_p): Likewise.
1968 * reload.c (find_valid_class): Likewise.
1969 (find_valid_class_1): Likewise.
1970 (reload_inner_reg_of_subreg): Likewise.
1971 (push_reload): Likewise.
1972 (combine_reloads): Likewise.
1973 (find_dummy_reload): Likewise.
1974 (find_reloads): Likewise.
1975 * reload1.c (find_reg): Likewise.
1976 (set_reload_reg): Likewise.
1977 (allocate_reload_reg): Likewise.
1978 (choose_reload_regs): Likewise.
1979 (reload_adjust_reg_for_temp): Likewise.
1980 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1981 (simplify_subreg_regno): Likewise.
1982 * sel-sched.c (init_regs_for_mode): Likewise.
1983 * varasm.c (make_decl_rtl): Likewise.
1984 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1985 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1987 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1988 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1989 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1990 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1991 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1993 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1994 (arc_mode_class): Delete.
1995 (HARD_REGNO_MODE_OK): Delete.
1996 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1997 (arc_hard_regno_mode_ok): Rename old array to...
1998 (arc_hard_regno_mode_ok_modes): ...this.
1999 (arc_conditional_register_usage): Update accordingly.
2000 (arc_mode_class): Make static.
2001 (arc_hard_regno_mode_ok): New function.
2002 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2003 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2004 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2005 (arm_hard_regno_mode_ok): Make static.
2006 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2008 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2009 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2010 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2012 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2013 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2014 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2015 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2016 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2018 * config/bfin/predicates.md (valid_reg_operand): Use
2019 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2020 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2021 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2022 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2023 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2024 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2025 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2026 (cr16_hard_regno_mode_ok): Make static and return a bool.
2027 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2028 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2029 (cris_hard_regno_mode_ok): New function.
2030 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2031 (epiphany_mode_class): Delete.
2032 (HARD_REGNO_MODE_OK): Delete.
2033 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2034 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2035 (hard_regno_mode_ok): Rename to...
2036 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2037 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2038 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2040 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2041 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2042 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2043 (frv_hard_regno_mode_ok): Make static and return a bool.
2044 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2046 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2047 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2048 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2049 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2051 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2052 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2053 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2054 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2056 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2057 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2058 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2059 (ia64_hard_regno_mode_ok): New function.
2060 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2061 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2062 (iq2000_hard_regno_mode_ok): New function.
2063 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2064 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2065 (lm32_hard_regno_mode_ok): New function.
2066 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2067 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2068 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2069 instead of HARD_REGNO_MODE_OK.
2070 (m32c_hard_regno_ok): Rename to...
2071 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2072 (m32c_cannot_change_mode_class): Update accordingly.
2073 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2074 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2075 (m32r_mode_class): Delete.
2076 (HARD_REGNO_MODE_OK): Delete.
2077 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2078 (m32r_hard_regno_mode_ok): Rename to...
2079 (m32r_hard_regno_modes): ...this.
2080 (m32r_mode_class): Make static.
2081 (m32r_hard_regno_mode_ok): New function.
2082 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2083 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2084 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2085 (m68k_hard_regno_mode_ok): Make static.
2086 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2087 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2088 (mcore_hard_regno_mode_ok): New function.
2089 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2090 (HARD_REGNO_MODE_OK): Delete.
2091 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2093 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2094 (microblaze_hard_regno_mode_ok): New function.
2095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2096 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2097 (mips_hard_regno_mode_ok): Delete.
2098 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2099 (mips_hard_regno_mode_ok_p): ...this and make static.
2100 (mips_hard_regno_mode_ok_p): Rename to...
2101 (mips_hard_regno_mode_ok_uncached): ...this.
2102 (mips_hard_regno_mode_ok): New function.
2103 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2104 of HARD_REGNO_MODE_OK.
2105 (mips_option_override): Update after above name changes.
2106 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2107 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2108 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2109 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2110 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2111 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2112 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2113 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2114 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2115 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2116 (msp430_hard_regno_mode_ok): Make static and return a bool.
2117 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2118 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2119 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2121 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2122 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2123 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2124 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2125 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2126 (PA_HARD_REGNO_MODE_OK): ...this
2127 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2128 (PA_HARD_REGNO_MODE_OK): ...this.
2129 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2130 (pa_hard_regno_mode_ok): New function.
2131 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2132 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2133 (pdp11_hard_regno_mode_ok): New function.
2134 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2135 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2137 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2140 (rs6000_hard_regno_mode_ok): Rename to...
2141 (rs6000_hard_regno_mode_ok_uncached): ...this.
2142 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2143 (rs6000_hard_regno_mode_ok): New function.
2144 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2145 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2146 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2147 (riscv_hard_regno_mode_ok): ...this and make static.
2148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2149 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2150 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2151 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2152 (rl78_hard_regno_mode_ok): Make static and return bool.
2153 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2154 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2156 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2158 (rs6000_hard_regno_mode_ok): Rename to...
2159 (rs6000_hard_regno_mode_ok_uncached): ...this.
2160 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2161 (rs6000_hard_regno_mode_ok): New function.
2162 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2163 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2164 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2165 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2166 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2167 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2169 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2170 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2171 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2172 (sh_hard_regno_mode_ok): Make static.
2173 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2174 instead of HARD_REGNO_MODE_OK.
2175 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2176 (sparc_mode_class): Delete.
2177 (HARD_REGNO_MODE_OK): Delete.
2178 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2179 (hard_regno_mode_classes): Make static.
2180 (sparc_mode_class): Likewise.
2181 (sparc_hard_regno_mode_ok): New function.
2182 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2183 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2184 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2186 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2187 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2188 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2189 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2190 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2191 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2192 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2193 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2194 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2195 (visium_hard_regno_mode_ok): New function.
2196 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2197 instead of HARD_REGNO_MODE_OK.
2198 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2199 (HARD_REGNO_MODE_OK): Delete.
2200 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2201 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2202 (xtensa_option_override): Update accordingly.
2203 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2204 (xtensa_hard_regno_mode_ok): New function.
2205 * system.h (HARD_REGNO_MODE_OK): Poison.
2207 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2208 Alan Hayward <alan.hayward@arm.com>
2209 David Sherwood <david.sherwood@arm.com>
2211 * target.def (hard_regno_call_part_clobbered): New hook.
2212 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2213 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2214 * doc/tm.texi: Regenerate.
2215 * hooks.h (hook_bool_uint_mode_false): Declare.
2216 * hooks.c (hook_bool_uint_mode_false): New function.
2217 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2218 * cselib.c (cselib_process_insn): Use
2219 targetm.hard_regno_call_part_clobbered instead of
2220 HARD_REGNO_CALL_PART_CLOBBERED.
2221 * ira-conflicts.c (ira_build_conflicts): Likewise.
2222 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2223 * lra-constraints.c (need_for_call_save_p): Likewise.
2224 * lra-lives.c: Include target.h.
2225 (check_pseudos_live_through_calls): Use
2226 targetm.hard_regno_call_part_clobbered instead of
2227 HARD_REGNO_CALL_PART_CLOBBERED.
2228 * regcprop.c: Include target.h.
2229 (copyprop_hardreg_forward_1): Use
2230 targetm.hard_regno_call_part_clobbered instead of
2231 HARD_REGNO_CALL_PART_CLOBBERED.
2232 * reginfo.c (choose_hard_reg_mode): Likewise.
2233 * regrename.c (check_new_reg_p): Likewise.
2234 * reload.c (find_equiv_reg): Likewise.
2235 * reload1.c (emit_reload_insns): Likewise.
2236 * sched-deps.c (deps_analyze_insn): Likewise.
2237 * sel-sched.c (init_regs_for_mode): Likewise.
2238 (mark_unavailable_hard_regs): Likewise.
2239 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2240 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2241 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2243 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2244 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2245 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2247 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2249 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2250 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2251 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2253 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2254 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2255 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2257 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2258 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2260 * config/powerpcspe/powerpcspe.c
2261 (rs6000_hard_regno_call_part_clobbered): New function.
2262 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2263 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2264 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2266 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2267 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2268 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2270 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2271 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2272 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2274 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2275 Alan Hayward <alan.hayward@arm.com>
2276 David Sherwood <david.sherwood@arm.com>
2278 * rtl.h (subreg_memory_offset): Declare.
2279 * emit-rtl.c (subreg_memory_offset): New function.
2280 * expmed.c (store_bit_field_1): Use it.
2281 * expr.c (undefined_operand_subword_p): Likewise.
2282 * simplify-rtx.c (simplify_subreg): Likewise.
2284 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2286 PR rtl-optimization/57448
2289 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2290 using atomic_load pattern.
2291 (expand_atomic_store): Likewise.
2293 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2296 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2297 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2298 handling. Use replace_call_with_value with NULL instead of
2299 gsi_replace, unlink_stmt_vdef and release_defs.
2301 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2304 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2306 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2309 * rtl.h (emit_library_call_value_1): Declare.
2310 (emit_library_call): Replace declaration with a series of overloads.
2311 Remove the parameter count argument.
2312 (emit_library_call_value): Likewise.
2313 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2314 with an "rtx_mode_t *".
2315 (emit_library_call_value): Delete.
2316 (emit_library_call): Likewise.
2317 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2318 (asan_emit_allocas_unpoison): Likewise.
2319 * builtins.c (expand_builtin_powi): Likewise.
2320 (expand_asan_emit_allocas_unpoison): Likewise.
2321 * cfgexpand.c (expand_main_function): Likewise.
2322 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2323 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2324 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2325 * config/arm/arm.c (arm_trampoline_init): Likewise.
2326 (arm_call_tls_get_addr): Likewise.
2327 (arm_expand_divmod_libfunc): Likewise.
2328 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2329 (smulsi3_highpart): Likewise.
2330 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2331 (c6x_expand_compare): Likewise.
2332 (c6x_expand_movmem): Likewise.
2333 * config/frv/frv.c (frv_trampoline_init): Likewise.
2334 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2335 (ix86_expand_divmod_libfunc): Likewise.
2336 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2337 (ia64_expand_compare): Likewise.
2338 (ia64_profile_hook): Likewise.
2339 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2340 (nonlocal_goto): Likewise.
2341 (restore_stack_nonlocal): Likewise.
2342 * config/m32r/m32r.c (block_move_call): Likewise.
2343 (m32r_trampoline_init): Likewise.
2344 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2345 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2346 (m68k_call_m68k_read_tp): Likewise.
2347 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2348 (microblaze_expand_divide): Likewise.
2349 * config/mips/mips.h (mips_args): Likewise.
2350 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2351 (MIPS_ICACHE_SYNC): Likewise.
2352 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2353 (nios2_trampoline_init): Likewise.
2354 * config/pa/pa.c (hppa_tls_call): Likewise.
2355 (pa_trampoline_init): Likewise.
2356 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2357 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2358 (expand_strn_compare): Likewise.
2359 (rs6000_generate_compare): Likewise.
2360 (rs6000_expand_float128_convert): Likewise.
2361 (output_profile_hook): Likewise.
2362 (rs6000_trampoline_init): Likewise.
2363 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2364 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2365 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2366 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2367 (rs6000_generate_compare): Likewise.
2368 (rs6000_expand_float128_convert): Likewise.
2369 (output_profile_hook): Likewise.
2370 (rs6000_trampoline_init): Likewise.
2371 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2372 * config/sh/sh.c (sh_trampoline_init): Likewise.
2373 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2374 (sparc_emit_float_lib_cmp): Likewise.
2375 (sparc32_initialize_trampoline): Likewise.
2376 (sparc64_initialize_trampoline): Likewise.
2377 (sparc_profile_hook): Likewise.
2378 * config/spu/spu.c (ea_load_store): Likewise.
2379 * config/spu/spu.md (floatunssidf2): Likewise.
2380 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2381 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2382 * config/visium/visium.c (expand_block_move_4): Likewise.
2383 (expand_block_move_2): Likewise.
2384 (expand_block_move_1): Likewise.
2385 (expand_block_set_4): Likewise.
2386 (expand_block_set_2): Likewise.
2387 (expand_block_set_1): Likewise.
2388 (visium_trampoline_init): Likewise.
2389 (visium_profile_hook): Likewise.
2390 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2391 (xtensa_setup_frame_addresses): Likewise.
2392 (xtensa_trampoline_init): Likewise.
2393 * except.c (sjlj_emit_function_enter): Likewise.
2394 (sjlj_emit_function_exit): Likewise.
2395 * explow.c (allocate_dynamic_stack_space): Likewise.
2396 (probe_stack_range): Likewise.
2397 * expr.c (convert_mode_scalar): Likewise.
2398 * optabs.c (expand_binop): Likewise.
2399 (expand_twoval_binop_libfunc): Likewise.
2400 (expand_unop): Likewise.
2401 (prepare_cmp_insn): Likewise.
2402 (prepare_float_lib_cmp): Likewise.
2403 (expand_float): Likewise.
2404 (expand_fix): Likewise.
2405 (expand_fixed_convert): Likewise.
2406 (maybe_emit_sync_lock_test_and_set): Likewise.
2407 (expand_atomic_compare_and_swap): Likewise.
2408 (expand_mem_thread_fence): Likewise.
2409 (expand_atomic_fetch_op): Likewise.
2411 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2413 * doc/generic.texi (OpenACC): Adjust URL.
2414 * doc/invoke.texi (C Dialect Options): Ditto.
2416 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2418 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2419 predicate for operand 1. Add (m,<S>) constraint.
2420 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2421 Prevent memory operand 1 with register operand 2.
2423 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2425 PR rtl-optimization/82024
2426 * combine.c (try_combine): If the combination result is a PARALLEL,
2427 and we only need to retain the SET in there that would be placed
2428 at I2, check that we can place that at I3 instead, before doing so.
2430 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2433 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2435 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2436 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2439 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2440 Jeff Law <law@redhat.com>
2442 * varasm.c (bss_initializer_p): Do not put constants into .bss
2443 (categorize_decl_for_section): Handle bss_initializer_p returning
2444 false when DECL_INITIAL is NULL.
2446 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2449 * config/s390/s390.c (s390_can_inline_p): New function.
2451 2017-09-01 Jeff Law <law@redhat.com>
2453 PR tree-optimization/82052
2454 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2455 Always initialize the returned slot after a hash table miss
2456 when INSERT is true.
2458 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2460 * config/s390/s390.md (mem_signal_fence): Remove.
2461 * doc/md.texi (mem_signal_fence): Remove.
2462 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2464 * target-insns.def (mem_signal_fence): Remove.
2466 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2469 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2472 * asan.c (create_odr_indicator): Strip name encoding from assembler
2473 name before appending it after __odr_asan_.
2475 2017-09-01 Martin Liska <mliska@suse.cz>
2477 PR tree-optimization/82059
2478 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2479 frequency only when an edge is redirected.
2481 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2483 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2484 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2485 (arc_conditional_register_usage): Remove ARC600 lp_count
2487 (arc_file_start): Emit Tag_ARC_CPU_variation.
2488 (arc_can_use_doloop_p): New conditions to use ZOLs.
2489 (hwloop_fail): New function.
2490 (hwloop_optimize): Likewise.
2491 (hwloop_pattern_reg): Likewise.
2492 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2493 (arc_reorg_loops): New function, calls reorg_loops.
2494 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2495 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2497 (arc_loop_hazard): Remove function, functionality moved into
2499 (arc_hazard): Remove arc_loop_hazard call.
2500 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2501 into hwloop_optimize.
2502 (arc_label_align): Remove ZOL handling.
2503 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2504 * config/arc/arc.md (doloop_begin): Remove pattern.
2505 (doloop_begin_i): Likewise.
2506 (doloop_end_i): Likewise.
2507 (doloop_fallback): Likewise.
2508 (doloop_fallback_m): Likewise.
2509 (doloop_end): Reimplement expand.
2510 (arc_lp): New pattern for LP instruction.
2511 (loop_end): New pattern.
2512 (loop_fail): Likewise.
2513 (decrement_and_branch_until_zero): Likewise.
2514 * config/arc/arc.opt (mlpc-width): New option.
2515 * doc/invoke.texi (mlpc-width): Document option.
2517 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2519 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2520 (arc_ccfsm_advance): Fix checking for delay slots.
2521 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2523 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2525 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2527 (movhi_insn): Update store instruction constraint which are saving
2528 6-bit short immediates.
2529 (movsi_insn): Consider also short scaled load operations.
2530 (zero_extendhisi2_i): Use Usd constraint instead of T.
2531 (extendhisi2_i): Add q constraint.
2532 (arc_clzsi2): Add type and length attributes.
2533 (arc_ctzsi2): Likewise.
2534 * config/arc/constraints.md (Usc): Update constraint, the
2535 assembler can parse two relocations for a single instruction.
2537 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2539 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2540 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2542 2017-08-31 Olivier Hainque <hainque@adacore.com>
2544 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2545 match as powerpc-wrs-vxworks*.
2547 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2549 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2550 register constraint for by-element operand.
2551 (aarch64_mls_elt_merge<mode>): Likewise.
2553 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2555 * config/arc/arc.c (arc_can_follow_jump): Check for short
2558 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2560 * config.gcc: Use g.opt for arc.
2561 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2562 functionality moved to ...
2563 (legitimate_scaled_address_p): New function, ...here.
2564 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2565 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2566 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2568 (arc_override_options): Handle G option.
2569 (arc_output_pic_addr_const): Correct function definition.
2570 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2571 (arc_decl_anon_ns_mem_p): Delete.
2572 (arc_in_small_data_p): Overhaul this function to take into
2573 consideration the value given via G option.
2574 (arc_rewrite_small_data_1): Renamed and corrected old
2575 arc_rewrite_small_data function.
2576 (arc_rewrite_small_data): New function.
2577 (small_data_pattern): Don't use pic_offset_table_rtx.
2578 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2579 * config/arc/simdext.md (movmisalignv2hi): Use
2580 prepare_move_operands function.
2582 (movmisalign*): Likewise.
2583 * doc/invoke.texi (ARC options): Document -G option.
2585 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2587 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2589 * config/arc/arc.c (arc_print_operand): Output scalled address for
2590 sdata whenever is possible.
2591 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2592 load/stores are available.
2593 (compact_sda_memory_operand): Check for the alignment required by
2594 code density instructions.
2595 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2597 * config/arc/constraints.md (Usd): Update constraint.
2598 (Us0): New constraint.
2599 (Usc): Update constraint.
2601 2017-08-31 Richard Biener <rguenther@suse.de>
2604 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2607 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2609 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2610 Resize type_signature.
2612 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2613 Alan Hayward <alan.hayward@arm.com>
2614 David Sherwood <david.sherwood@arm.com>
2616 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2617 subregs whose inner modes can be stored in GPRs.
2618 (aarch64_classify_index): Likewise.
2620 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2621 Alan Hayward <alan.hayward@arm.com>
2622 David Sherwood <david.sherwood@arm.com>
2624 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2625 (V_INT_EQUIV): ...this.
2626 (v_cmp_result): Rename to...
2627 (v_int_equiv): ...this.
2628 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2629 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2630 (copysign<mode>3): Likewise.
2631 (aarch64_simd_bsl<mode>_internal): Likewise.
2632 (aarch64_simd_bsl<mode>): Likewise.
2633 (vec_cmp<mode><mode>): Likewise.
2634 (vcond<mode><mode>): Likewise.
2635 (vcond<v_cmp_mixed><mode>): Likewise.
2636 (vcondu<mode><v_cmp_mixed>): Likewise.
2637 (aarch64_cm<optab><mode>): Likewise.
2638 (aarch64_cmtst<mode>): Likewise.
2639 (aarch64_fac<optab><mode>): Likewise.
2640 (vec_perm_const<mode>): Likewise.
2641 (vcond_mask_<mode><v_cmp_result>): Rename to...
2642 (vcond_mask_<mode><v_int_equiv>): ...this.
2643 (vec_cmp<mode><v_cmp_result>): Rename to...
2644 (vec_cmp<mode><v_int_equiv>): ...this.
2646 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2647 Alan Hayward <alan.hayward@arm.com>
2648 David Sherwood <david.sherwood@arm.com>
2650 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2652 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2653 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2654 (UNSPEC_LD4_DREG): New unspecs.
2655 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2656 (aarch64_ld2<mode>_dreg_be): Replace with...
2657 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2659 (aarch64_ld3<mode>_dreg_le)
2660 (aarch64_ld3<mode>_dreg_be): Replace with...
2661 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2663 (aarch64_ld4<mode>_dreg_le)
2664 (aarch64_ld4<mode>_dreg_be): Replace with...
2665 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2668 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2670 PR tree-optimization/81987
2671 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2672 insert an initializer in a location not dominated by the stride
2675 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2677 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2678 on the entire header of the finally block in the fallthru case.
2680 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2682 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2684 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2686 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2687 rs6000_emit_move_from_cr and call renamed function.
2688 (rs6000_emit_prologue): Call renamed functions.
2689 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2690 movesi_from_cr, remove volatile CRs.
2692 2017-08-30 Jon Beniston <jon@beniston.com>
2693 Richard Biener <rguenther@suse.de>
2695 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2696 of VECTOR_MODE_P check.
2697 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2698 element vector types.
2700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2702 * df.h (df_read_modify_subreg_p): Remove in favor of...
2703 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2704 const_rtx instead of an rtx.
2705 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2706 * df-problems.c (df_word_lr_mark_ref): Likewise.
2707 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2708 (mark_pseudo_reg_dead): Likewise.
2709 (mark_ref_dead): Likewise.
2710 * reginfo.c (init_subregs_of_mode): Likewise.
2711 * sched-deps.c (sched_analyze_1): Likewise.
2712 * df-scan.c (df_def_record_1): Likewise.
2713 (df_uses_record): Likewise.
2714 (df_read_modify_subreg_p): Remove in favor of...
2715 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2716 const_rtx instead of an rtx.
2718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2719 Alan Hayward <alan.hayward@arm.com>
2720 David Sherwood <david.sherwood@arm.com>
2722 * rtl.h (partial_subreg_p): New function.
2723 * caller-save.c (save_call_clobbered_regs): Use it.
2724 * calls.c (expand_call): Likewise.
2725 * combine.c (combinable_i3pat): Likewise.
2726 (simplify_set): Likewise.
2727 (make_extraction): Likewise.
2728 (make_compound_operation_int): Likewise.
2729 (gen_lowpart_or_truncate): Likewise.
2730 (force_to_mode): Likewise.
2731 (make_field_assignment): Likewise.
2732 (reg_truncated_to_mode): Likewise.
2733 (record_truncated_value): Likewise.
2734 (move_deaths): Likewise.
2735 * cse.c (record_jump_cond): Likewise.
2736 (cse_insn): Likewise.
2737 * cselib.c (cselib_lookup_1): Likewise.
2738 * expmed.c (extract_bit_field_using_extv): Likewise.
2739 * function.c (assign_parm_setup_reg): Likewise.
2740 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2741 * ira-build.c (create_insn_allocnos): Likewise.
2742 * lra-coalesce.c (merge_pseudos): Likewise.
2743 * lra-constraints.c (match_reload): Likewise.
2744 (simplify_operand_subreg): Likewise.
2745 (curr_insn_transform): Likewise.
2746 * lra-lives.c (process_bb_lives): Likewise.
2747 * lra.c (new_insn_reg): Likewise.
2748 (lra_substitute_pseudo): Likewise.
2749 * regcprop.c (mode_change_ok): Likewise.
2750 (maybe_mode_change): Likewise.
2751 (copyprop_hardreg_forward_1): Likewise.
2752 * reload.c (push_reload): Likewise.
2753 (find_reloads): Likewise.
2754 (find_reloads_subreg_address): Likewise.
2755 * reload1.c (alter_reg): Likewise.
2756 (eliminate_regs_1): Likewise.
2757 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2759 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2761 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2762 back to if statements, including unpack.
2764 2017-08-30 Martin Liska <mliska@suse.cz>
2767 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2769 (func_checker::compare_asm_inputs_outputs): ... this function.
2770 (func_checker::compare_gimple_asm): Use the function to compare
2771 also ASM constrains.
2772 * ipa-icf-gimple.h: Rename the function.
2774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2775 Alan Hayward <alan.hayward@arm.com>
2776 David Sherwood <david.sherwood@arm.com>
2778 * coretypes.h (complex_mode): New type.
2779 * gdbhooks.py (build_pretty_printer): Handle it.
2780 * machmode.h (complex_mode): New class.
2781 (complex_mode::includes_p): New function.
2782 (is_complex_int_mode): Likewise.
2783 (is_complex_float_mode): Likewise.
2784 * genmodes.c (get_mode_class): Handle complex mode classes.
2785 * function.c (expand_function_end): Use is_complex_int_mode.
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 * coretypes.h (scalar_mode_pod): New typedef.
2792 * gdbhooks.py (build_pretty_printer): Handle it.
2793 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2794 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2795 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2796 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2800 Alan Hayward <alan.hayward@arm.com>
2801 David Sherwood <david.sherwood@arm.com>
2803 * machmode.h (mode_for_vector): Take a scalar_mode instead
2805 * stor-layout.c (mode_for_vector): Likewise.
2806 * explow.c (promote_mode): Use as_a <scalar_mode>.
2807 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2810 Alan Hayward <alan.hayward@arm.com>
2811 David Sherwood <david.sherwood@arm.com>
2813 * target.def (preferred_simd_mode): Take a scalar_mode
2814 instead of a machine_mode.
2815 * targhooks.h (default_preferred_simd_mode): Likewise.
2816 * targhooks.c (default_preferred_simd_mode): Likewise.
2817 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2818 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2819 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2820 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2821 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2822 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2823 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2824 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2826 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2827 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2828 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2829 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2830 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2831 * doc/tm.texi: Regenerate.
2832 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2836 Alan Hayward <alan.hayward@arm.com>
2837 David Sherwood <david.sherwood@arm.com>
2839 * target.def (scalar_mode_supported_p): Take a scalar_mode
2840 instead of a machine_mode.
2841 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2842 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2843 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2844 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2845 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2846 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2847 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2848 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2849 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2850 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2851 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2852 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2853 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2854 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2856 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2857 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2858 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2859 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2860 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2862 * doc/tm.texi: Regenerate.
2864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2865 Alan Hayward <alan.hayward@arm.com>
2866 David Sherwood <david.sherwood@arm.com>
2868 * coretypes.h (opt_scalar_mode): New typedef.
2869 * gdbhooks.py (build_pretty_printers): Handle it.
2870 * machmode.h (mode_iterator::get_2xwider): Add overload for
2872 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2874 * expr.c (convert_mode_scalar): Likewise.
2875 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2876 * optabs.c (expand_float): Likewise.
2877 (expand_fix): Likewise.
2878 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2881 Alan Hayward <alan.hayward@arm.com>
2882 David Sherwood <david.sherwood@arm.com>
2884 * optabs.c (expand_float): Explicitly check for scalars before
2885 using a branching expansion.
2886 (expand_fix): Likewise.
2888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2889 Alan Hayward <alan.hayward@arm.com>
2890 David Sherwood <david.sherwood@arm.com>
2892 * expr.c (convert_mode): Split scalar handling out into...
2893 (convert_mode_scalar): ...this new function. Treat the modes
2896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2897 Alan Hayward <alan.hayward@arm.com>
2898 David Sherwood <david.sherwood@arm.com>
2900 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2902 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2905 Alan Hayward <alan.hayward@arm.com>
2906 David Sherwood <david.sherwood@arm.com>
2908 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2909 rather than a machine_mode.
2910 (fixed_from_string): Likewise.
2911 (fixed_convert): Likewise.
2912 (fixed_convert_from_int): Likewise.
2913 (fixed_convert_from_real): Likewise.
2914 (real_convert_from_fixed): Likewise.
2915 * fixed-value.c (fixed_from_double_int): Likewise.
2916 (fixed_from_string): Likewise.
2917 (fixed_convert): Likewise.
2918 (fixed_convert_from_int): Likewise.
2919 (fixed_convert_from_real): Likewise.
2920 (real_convert_from_fixed): Likewise.
2921 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2924 Alan Hayward <alan.hayward@arm.com>
2925 David Sherwood <david.sherwood@arm.com>
2927 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2928 of separate mode class checks. Do not allow vector modes here.
2929 (immed_wide_int_const): Use as_a <scalar_mode>.
2930 * explow.c (trunc_int_for_mode): Likewise.
2931 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2932 (wi::shwi): Likewise.
2933 (wi::min_value): Likewise.
2934 (wi::max_value): Likewise.
2935 * dwarf2out.c (loc_descriptor): Likewise.
2936 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2940 Alan Hayward <alan.hayward@arm.com>
2941 David Sherwood <david.sherwood@arm.com>
2943 * tree.h (SCALAR_TYPE_MODE): New macro.
2944 * expr.c (expand_expr_addr_expr_1): Use it.
2945 (expand_expr_real_2): Likewise.
2946 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2947 (fold_convert_const_fixed_from_int): Likewise.
2948 (fold_convert_const_fixed_from_real): Likewise.
2949 (native_encode_fixed): Likewise
2950 (native_encode_complex): Likewise
2951 (native_encode_vector): Likewise.
2952 (native_interpret_fixed): Likewise.
2953 (native_interpret_real): Likewise.
2954 (native_interpret_complex): Likewise.
2955 (native_interpret_vector): Likewise.
2956 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2957 (simd_clone_adjust_argument_types): Likewise.
2958 (simd_clone_init_simd_arrays): Likewise.
2959 (simd_clone_adjust): Likewise.
2960 * stor-layout.c (layout_type): Likewise.
2961 * tree.c (build_minus_one_cst): Likewise.
2962 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2963 * tree-inline.c (estimate_move_cost): Likewise.
2964 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2965 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2966 (vectorizable_reduction): Likewise.
2967 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2968 (vect_recog_mixed_size_cond_pattern): Likewise.
2969 (check_bool_pattern): Likewise.
2970 (adjust_bool_pattern): Likewise.
2971 (search_type_for_mask_1): Likewise.
2972 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2973 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2974 (vectorizable_load): Likewise.
2975 (vectorizable_store): Likewise.
2976 * ubsan.c (ubsan_encode_value): Likewise.
2977 * varasm.c (output_constant): Likewise.
2979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2980 Alan Hayward <alan.hayward@arm.com>
2981 David Sherwood <david.sherwood@arm.com>
2983 * coretypes.h (scalar_mode): New class.
2984 * machmode.h (scalar_mode): Likewise.
2985 (scalar_mode::includes_p): New function.
2986 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2987 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2988 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2989 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2990 * expmed.c (store_bit_field_1): Likewise.
2991 (extract_bit_field_1): Likewise.
2992 * expr.c (write_complex_part): Likewise.
2993 (read_complex_part): Likewise.
2994 (emit_move_complex_push): Likewise.
2995 (expand_expr_real_2): Likewise.
2996 * function.c (assign_parm_setup_reg): Likewise.
2997 (assign_parms_unsplit_complex): Likewise.
2998 * optabs.c (expand_binop): Likewise.
2999 * rtlanal.c (subreg_get_info): Likewise.
3000 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3001 * varasm.c (output_constant_pool_2): Likewise.
3003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3004 Alan Hayward <alan.hayward@arm.com>
3005 David Sherwood <david.sherwood@arm.com>
3007 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3009 (expmed_mult_highpart_optab): Likewise.
3010 (expmed_mult_highpart): Likewise.
3012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3013 Alan Hayward <alan.hayward@arm.com>
3014 David Sherwood <david.sherwood@arm.com>
3016 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3017 instead of a machine_mode.
3018 (builtin_memset_read_str): Likewise.
3019 * builtins.c (c_readstr): Likewise.
3020 (builtin_memcpy_read_str): Likewise.
3021 (builtin_strncpy_read_str): Likewise.
3022 (builtin_memset_read_str): Likewise.
3023 (builtin_memset_gen_str): Likewise.
3024 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3025 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3026 instead of a machine_mode.
3027 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3029 (make_extraction): Likewise.
3030 (try_widen_shift_mode): Take and return scalar_int_modes instead
3032 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3033 a scalar_int_mode instead of a machine_mode.
3034 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3035 (avr_addr_space_pointer_mode): Likewise.
3036 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3037 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3038 (msp430_unwind_word_mode): Likewise.
3039 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3040 (spu_addr_space_pointer_mode): Likewise.
3041 (spu_addr_space_address_mode): Likewise.
3042 (spu_libgcc_cmp_return_mode): Likewise.
3043 (spu_libgcc_shift_count_mode): Likewise.
3044 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3045 (rl78_addr_space_pointer_mode): Likewise.
3046 (fl78_unwind_word_mode): Likewise.
3047 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3049 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3050 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3051 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3052 (mips_valid_pointer_mode): Likewise.
3053 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3054 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3055 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3057 (ft32_addr_space_address_mode): Likewise.
3058 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3059 scalar_int_mode instead of a machine_mode.
3060 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3062 (m32c_addr_space_address_mode): Likewise.
3063 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3064 (rs6000_eh_return_filter_mode): Likewise.
3065 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3066 (rs6000_eh_return_filter_mode): Likewise.
3067 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3068 (s390_libgcc_shift_count_mode): Likewise.
3069 (s390_unwind_word_mode): Likewise.
3070 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3072 * target.def (mode_rep_extended): Likewise.
3073 (valid_pointer_mode): Likewise.
3074 (addr_space.valid_pointer_mode): Likewise.
3075 (eh_return_filter_mode): Return a scalar_int_mode rather than
3077 (libgcc_cmp_return_mode): Likewise.
3078 (libgcc_shift_count_mode): Likewise.
3079 (unwind_word_mode): Likewise.
3080 (addr_space.pointer_mode): Likewise.
3081 (addr_space.address_mode): Likewise.
3082 * doc/tm.texi: Regenerate.
3083 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3085 (do_jump): Use scalar_int_mode for local variables.
3086 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3087 rather than a machine_mode.
3088 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3089 (scompare_loc_descriptor_wide): Likewise.
3090 (scompare_loc_descriptor_narrow): Likewise.
3091 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3093 * except.c (sjlj_emit_dispatch_table): Likewise.
3094 (expand_builtin_eh_copy_values): Likewise.
3095 * explow.c (convert_memory_address_addr_space_1): Likewise.
3096 Take a scalar_int_mode rather than a machine_mode.
3097 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3098 than a machine_mode.
3099 (memory_address_addr_space): Use scalar_int_mode for local variables.
3100 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3101 rather than a machine_mode.
3102 * expmed.c (mask_rtx): Likewise.
3103 (init_expmed_one_conv): Likewise.
3104 (expand_mult_highpart_adjust): Likewise.
3105 (extract_high_half): Likewise.
3106 (expmed_mult_highpart_optab): Likewise.
3107 (expmed_mult_highpart): Likewise.
3108 (expand_smod_pow2): Likewise.
3109 (expand_sdiv_pow2): Likewise.
3110 (emit_store_flag_int): Likewise.
3111 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3113 (extract_low_bits): Likewise.
3114 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3116 * expr.c (pieces_addr::adjust): Likewise.
3117 (can_store_by_pieces): Likewise.
3118 (store_by_pieces): Likewise.
3119 (clear_by_pieces_1): Likewise.
3120 (expand_expr_addr_expr_1): Likewise.
3121 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3122 (expand_expr_real_1): Likewise.
3123 (try_casesi): Likewise.
3124 * final.c (shorten_branches): Likewise.
3125 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3126 type of "mode" to machine_mode.
3127 * internal-fn.c (expand_arith_overflow_result_store): Take a
3128 scalar_int_mode rather than a machine_mode.
3129 (expand_mul_overflow): Use scalar_int_mode for local variables.
3130 * loop-doloop.c (doloop_modify): Likewise.
3131 (doloop_optimize): Likewise.
3132 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3133 than a machine_mode.
3134 (expand_doubleword_shift_condmove): Likewise.
3135 (expand_doubleword_shift): Likewise.
3136 (expand_doubleword_clz): Likewise.
3137 (expand_doubleword_popcount): Likewise.
3138 (expand_doubleword_parity): Likewise.
3139 (expand_absneg_bit): Use scalar_int_mode for local variables.
3140 (prepare_float_lib_cmp): Likewise.
3141 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3142 rather than a machine_mode.
3143 (convert_memory_address_addr_space): Likewise.
3144 (get_mode_bounds): Likewise.
3145 (get_address_mode): Return a scalar_int_mode rather than a
3147 * rtlanal.c (get_address_mode): Likewise.
3148 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3149 than a machine_mode.
3150 * targhooks.c (default_mode_rep_extended): Likewise.
3151 (default_valid_pointer_mode): Likewise.
3152 (default_addr_space_valid_pointer_mode): Likewise.
3153 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3154 than a machine_mode.
3155 (default_libgcc_cmp_return_mode): Likewise.
3156 (default_libgcc_shift_count_mode): Likewise.
3157 (default_unwind_word_mode): Likewise.
3158 (default_addr_space_pointer_mode): Likewise.
3159 (default_addr_space_address_mode): Likewise.
3160 * targhooks.h (default_eh_return_filter_mode): Likewise.
3161 (default_libgcc_cmp_return_mode): Likewise.
3162 (default_libgcc_shift_count_mode): Likewise.
3163 (default_unwind_word_mode): Likewise.
3164 (default_addr_space_pointer_mode): Likewise.
3165 (default_addr_space_address_mode): Likewise.
3166 (default_mode_rep_extended): Take a scalar_int_mode rather than
3168 (default_valid_pointer_mode): Likewise.
3169 (default_addr_space_valid_pointer_mode): Likewise.
3170 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3172 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3173 rather than a machine_mode.
3174 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3175 for local variables.
3176 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3177 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3178 than a machine_mode.
3180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3181 Alan Hayward <alan.hayward@arm.com>
3182 David Sherwood <david.sherwood@arm.com>
3184 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3185 the mode argument to scalar_int_mode.
3186 (do_jump_by_parts_zero_rtx): Likewise.
3187 (do_jump_by_parts_equality_rtx): Likewise.
3188 (do_jump_by_parts_greater): Take a mode argument.
3189 (do_jump_by_parts_equality): Likewise.
3190 (do_jump_1): Update calls accordingly.
3192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3193 Alan Hayward <alan.hayward@arm.com>
3194 David Sherwood <david.sherwood@arm.com>
3196 * is-a.h (safe_dyn_cast): New function.
3197 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3198 (jump_table_for_label): Likewise.
3199 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3200 instead of an rtx_insn *.
3201 (shorten_branches): Use dyn_cast instead of LABEL_P and
3202 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3203 rtx_jump_table_data::get_data_mode.
3204 (final_scan_insn): Likewise.
3206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3207 Alan Hayward <alan.hayward@arm.com>
3208 David Sherwood <david.sherwood@arm.com>
3210 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3211 trying to combine a full-register integer set with a subreg
3214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3215 Alan Hayward <alan.hayward@arm.com>
3216 David Sherwood <david.sherwood@arm.com>
3218 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3219 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3226 when the two are known to be equal.
3228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3229 Alan Hayward <alan.hayward@arm.com>
3230 David Sherwood <david.sherwood@arm.com>
3232 * simplify-rtx.c (simplify_const_unary_operation): Use
3233 is_a <scalar_int_mode> instead of checking for a nonzero
3234 precision. Forcibly convert op_mode to a scalar_int_mode
3235 in that case. More clearly differentiate the operand and
3236 result modes and use the former when deciding what the value
3237 of a count-bits operation should be. Use is_int_mode instead
3238 of checking for a MODE_INT. Remove redundant check for whether
3239 this mode has a zero precision.
3241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3242 Alan Hayward <alan.hayward@arm.com>
3243 David Sherwood <david.sherwood@arm.com>
3245 * optabs.c (widen_leading): Change the type of the mode argument
3246 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3247 (widen_bswap): Likewise.
3248 (expand_parity): Likewise.
3249 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3250 (expand_ffs): Likewise.
3251 (epand_unop): Check for scalar integer modes before calling the
3254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3255 Alan Hayward <alan.hayward@arm.com>
3256 David Sherwood <david.sherwood@arm.com>
3258 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3260 (expand_expr_real_1): Update call accordingly.
3262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3263 Alan Hayward <alan.hayward@arm.com>
3264 David Sherwood <david.sherwood@arm.com>
3266 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3267 value_mode arguments. Use scalar_int_mode internally.
3268 (store_bit_field_1): Rename the new integer mode from imode
3269 to op0_mode and use it instead of GET_MODE (op0). Update calls
3270 to store_split_bit_field, store_bit_field_using_insv and
3271 store_fixed_bit_field.
3272 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3273 Use scalar_int_mode internally. Use a bit count rather than a mode
3274 when calculating the largest bit size for get_best_mode.
3275 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3276 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3277 Remove assertion that OP0 has a scalar integer mode.
3278 (store_split_bit_field): Add op0_mode and value_mode arguments.
3279 Update calls to extract_fixed_bit_field.
3280 (extract_bit_field_using_extv): Add an op0_mode argument.
3281 Use scalar_int_mode internally.
3282 (extract_bit_field_1): Rename the new integer mode from imode to
3283 op0_mode and use it instead of GET_MODE (op0). Update calls to
3284 extract_split_bit_field, extract_bit_field_using_extv and
3285 extract_fixed_bit_field.
3286 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3287 to extract_split_bit_field and extract_fixed_bit_field_1.
3288 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3289 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3291 (extract_split_bit_field): Add an op0_mode argument. Update call
3292 to extract_fixed_bit_field.
3294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3295 Alan Hayward <alan.hayward@arm.com>
3296 David Sherwood <david.sherwood@arm.com>
3298 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3299 * explow.c (hard_function_value): Likewise.
3300 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3301 convert_to_mode call outside the loop.
3302 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3303 for the mode iterator. Require the mode specified by max_pieces
3305 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3307 (copy_blkmode_to_reg): Likewise.
3308 (set_storage_via_setmem): Likewise.
3309 * optabs.c (prepare_cmp_insn): Likewise.
3310 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3311 * stor-layout.c (finish_bitfield_representative): Likewise.
3313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3314 Alan Hayward <alan.hayward@arm.com>
3315 David Sherwood <david.sherwood@arm.com>
3317 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3318 * expr.c (convert_move): Use them.
3319 (convert_modes): Likewise.
3320 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3327 parameter for the mode of "x". Remove the "known_x", "known_mode"
3328 and "known_ret" arguments. Change the type of the mode argument
3330 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3331 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3332 (reg_num_sign_bit_copies_for_combine): Likewise.
3333 * rtlanal.c (nonzero_bits1): Likewise.
3334 (num_sign_bit_copies1): Likewise.
3335 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3336 (reg_num_sign_bit_copies_general): Likewise.
3337 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3338 (reg_nonzero_bits_general): Likewise.
3340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3341 Alan Hayward <alan.hayward@arm.com>
3342 David Sherwood <david.sherwood@arm.com>
3344 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3345 than in subroutines. Return 1 for non-integer modes.
3346 (cached_num_sign_bit_copies): Change the type of the mode parameter
3348 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3349 classes. Handle CONST_INT_P first and then check whether X also
3350 has a scalar integer mode. Check the same thing for inner registers
3351 of a SUBREG and for values that are being extended or truncated.
3353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3354 Alan Hayward <alan.hayward@arm.com>
3355 David Sherwood <david.sherwood@arm.com>
3357 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3358 in subroutines. Return the mode mask for non-integer modes.
3359 (cached_nonzero_bits): Change the type of the mode parameter
3361 (nonzero_bits1): Likewise. Remove early exit for other mode
3362 classes. Handle CONST_INT_P first and then check whether X
3363 also has a scalar integer mode.
3365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3366 Alan Hayward <alan.hayward@arm.com>
3367 David Sherwood <david.sherwood@arm.com>
3369 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3370 Return a scalar_int_mode and assert that the size is greater than
3372 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3373 (op_by_pieces_d::op_by_pieces_d): Likewise.
3374 (op_by_pieces_d::run): Likewise.
3375 (can_store_by_pieces): Likewise.
3377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3378 Alan Hayward <alan.hayward@arm.com>
3379 David Sherwood <david.sherwood@arm.com>
3381 * combine.c (extract_left_shift): Add a mode argument and update
3383 (make_compound_operation_int): Change the type of the mode parameter
3384 to scalar_int_mode and update the call to extract_left_shift.
3386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3387 Alan Hayward <alan.hayward@arm.com>
3388 David Sherwood <david.sherwood@arm.com>
3390 * combine.c (simplify_and_const_int): Change the type of the mode
3391 parameter to scalar_int_mode.
3392 (simplify_and_const_int_1): Likewise. Update recursive call.
3394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3395 Alan Hayward <alan.hayward@arm.com>
3396 David Sherwood <david.sherwood@arm.com>
3398 * combine.c (simplify_compare_const): Check that the mode is a
3399 scalar_int_mode (rather than VOIDmode) before testing its
3401 (simplify_comparison): Move COMPARISON_P handling out of the
3402 loop and restrict the latter part of the loop to scalar_int_modes.
3403 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3404 and when considering SUBREG_REGs. Use is_int_mode instead of
3405 checking GET_MODE_CLASS against MODE_INT.
3407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3408 Alan Hayward <alan.hayward@arm.com>
3409 David Sherwood <david.sherwood@arm.com>
3411 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3412 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3413 shift_unit_mode and for modes involved in scalar shifts.
3415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3416 Alan Hayward <alan.hayward@arm.com>
3417 David Sherwood <david.sherwood@arm.com>
3419 * combine.c (force_int_to_mode): New function, split out from...
3420 (force_to_mode): ...here. Keep xmode up-to-date and use it
3421 instead of GET_MODE (x).
3423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3424 Alan Hayward <alan.hayward@arm.com>
3425 David Sherwood <david.sherwood@arm.com>
3427 * optabs-query.h (extraction_insn::struct_mode): Change type to
3428 opt_scalar_int_mode and update comment.
3429 (extraction_insn::field_mode): Change type to scalar_int_mode.
3430 (extraction_insn::pos_mode): Likewise.
3431 * combine.c (make_extraction): Update accordingly.
3432 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3433 (get_optab_extraction_insn): Likewise.
3434 * recog.c (simplify_while_replacing): Likewise.
3435 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3436 parameter to opt_scalar_int_mode.
3438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3439 Alan Hayward <alan.hayward@arm.com>
3440 David Sherwood <david.sherwood@arm.com>
3442 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3443 to a scalar_int_mode instead of a machine_mode.
3444 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3445 (get_best_mode): Return a boolean and use a pointer argument to store
3446 the selected mode. Replace the limit mode parameter with a bit limit.
3447 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3448 for the values returned by bit_field_mode_iterator::next_mode.
3449 (store_bit_field): Update call to get_best_mode.
3450 (store_fixed_bit_field): Likewise.
3451 (extract_fixed_bit_field): Likewise.
3452 * expr.c (optimize_bitfield_assignment_op): Likewise.
3453 * fold-const.c (optimize_bit_field_compare): Likewise.
3454 (fold_truth_andor_1): Likewise.
3455 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3456 Update for new type of m_mode.
3457 (get_best_mode): As above.
3459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3460 Alan Hayward <alan.hayward@arm.com>
3461 David Sherwood <david.sherwood@arm.com>
3463 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3464 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3465 (store_bit_field): Check is_a <scalar_int_mode> before calling
3466 strict_volatile_bitfield_p.
3467 (extract_bit_field): Likewise.
3469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3470 Alan Hayward <alan.hayward@arm.com>
3471 David Sherwood <david.sherwood@arm.com>
3473 * target.def (cstore_mode): Return a scalar_int_mode.
3474 * doc/tm.texi: Regenerate.
3475 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3476 * targhooks.h (default_cstore_mode): Likewise.
3477 * targhooks.c (default_cstore_mode): Likewise, using a forced
3479 * expmed.c (emit_cstore): Expect the target of the cstore to be
3482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3483 Alan Hayward <alan.hayward@arm.com>
3484 David Sherwood <david.sherwood@arm.com>
3486 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3488 (niter_desc): Likewise mode.
3489 (iv_analyze): Add a mode parameter.
3491 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3492 and change its type to scalar_int_mode.
3493 * loop-iv.c: Update commentary at head of file.
3494 (iv_constant): Pass the mode paraeter before the rtx it describes
3495 and change its type to scalar_int_mode. Remove VOIDmode handling.
3496 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3497 (iv_extend): Likewise.
3498 (shorten_into_mode): Likewise.
3499 (iv_add): Use scalar_int_mode.
3500 (iv_mult): Likewise.
3501 (iv_shift): Likewise.
3502 (canonicalize_iv_subregs): Likewise.
3503 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3504 it describes and change its mode to scalar_int_mode. Also change
3505 the type of the returned inner_mode to scalar_int_mode.
3506 (get_biv_step): Likewise, turning outer_mode from a pointer
3507 into a direct parameter. Update call to get_biv_step_1.
3508 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3509 iv_constant and get_biv_step.
3510 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3511 and change its type to scalar_int_mode. Don't initialise iv->mode
3512 to VOIDmode and remove later checks for its still being VOIDmode.
3513 Update calls to iv_analyze_op and iv_analyze_expr. Check
3514 is_a <scalar_int_mode> when changing the mode under consideration.
3515 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3516 Update call to iv_analyze_expr.
3517 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3518 inner register is not also a scalar_int_mode. Update call to
3520 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3521 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3522 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3523 separate mode class checks. Update calls to iv_analyze. Remove
3524 fix-up of VOIDmodes after iv_analyze_biv.
3525 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3526 don't have a scalar_int_mode. Update call to biv_p.
3528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3529 Alan Hayward <alan.hayward@arm.com>
3530 David Sherwood <david.sherwood@arm.com>
3532 * cfgexpand.c (convert_debug_memory_address): Use
3533 as_a <scalar_int_mode>.
3534 * combine.c (expand_compound_operation): Likewise.
3535 (make_extraction): Likewise.
3536 (change_zero_ext): Likewise.
3537 (simplify_comparison): Likewise.
3538 * cse.c (cse_insn): Likewise.
3539 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3540 (mem_loc_descriptor): Likewise.
3541 (loc_descriptor): Likewise.
3542 * expmed.c (init_expmed_one_mode): Likewise.
3543 (synth_mult): Likewise.
3544 (emit_store_flag_1): Likewise.
3545 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3546 of a comparison with size.
3547 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3548 (reduce_to_bit_field_precision): Likewise.
3549 * function.c (expand_function_end): Likewise.
3550 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3551 * loop-doloop.c (doloop_modify): Likewise.
3552 * optabs.c (expand_binop): Likewise.
3553 (expand_unop): Likewise.
3554 (expand_copysign_absneg): Likewise.
3555 (prepare_cmp_insn): Likewise.
3556 (maybe_legitimize_operand): Likewise.
3557 * recog.c (const_scalar_int_operand): Likewise.
3558 * rtlanal.c (get_address_mode): Likewise.
3559 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3560 (simplify_cond_clz_ctz): Likewise.
3561 * tree-nested.c (get_nl_goto_field): Likewise.
3562 * tree.c (build_vector_type_for_mode): Likewise.
3563 * var-tracking.c (use_narrower_mode): Likewise.
3565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3566 Alan Hayward <alan.hayward@arm.com>
3567 David Sherwood <david.sherwood@arm.com>
3569 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3570 * builtins.c (expand_builtin_signbit): Use it.
3571 * cfgexpand.c (expand_debug_expr): Likewise.
3572 * dojump.c (do_jump): Likewise.
3573 (do_compare_and_jump): Likewise.
3574 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3575 * expmed.c (make_tree): Likewise.
3576 * expr.c (expand_expr_real_2): Likewise.
3577 (expand_expr_real_1): Likewise.
3578 (try_casesi): Likewise.
3579 * fold-const-call.c (fold_const_call_ss): Likewise.
3580 * fold-const.c (unextend): Likewise.
3581 (extract_muldiv_1): Likewise.
3582 (fold_single_bit_test): Likewise.
3583 (native_encode_int): Likewise.
3584 (native_encode_string): Likewise.
3585 (native_interpret_int): Likewise.
3586 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3587 * internal-fn.c (expand_addsub_overflow): Likewise.
3588 (expand_neg_overflow): Likewise.
3589 (expand_mul_overflow): Likewise.
3590 (expand_arith_overflow): Likewise.
3591 * match.pd: Likewise.
3592 * stor-layout.c (layout_type): Likewise.
3593 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3594 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3595 * tree-ssanames.c (get_range_info): Likewise.
3596 * tree-switch-conversion.c (array_value_type) Likewise.
3597 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3598 (vect_recog_divmod_pattern): Likewise.
3599 (vect_recog_mixed_size_cond_pattern): Likewise.
3600 * tree-vrp.c (extract_range_basic): Likewise.
3601 (simplify_float_conversion_using_ranges): Likewise.
3602 * tree.c (int_fits_type_p): Likewise.
3603 * ubsan.c (instrument_bool_enum_load): Likewise.
3604 * varasm.c (mergeable_string_section): Likewise.
3605 (narrowing_initializer_constant_valid_p): Likewise.
3606 (output_constant): Likewise.
3608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3609 Alan Hayward <alan.hayward@arm.com>
3610 David Sherwood <david.sherwood@arm.com>
3612 * machmode.h (NARROWEST_INT_MODE): New macro.
3613 * expr.c (alignment_for_piecewise_move): Use it instead of
3614 GET_CLASS_NARROWEST_MODE (MODE_INT).
3615 (push_block): Likewise.
3616 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3618 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3621 Alan Hayward <alan.hayward@arm.com>
3622 David Sherwood <david.sherwood@arm.com>
3624 * postreload.c (move2add_valid_value_p): Change the type of the
3625 mode parameter to scalar_int_mode.
3626 (move2add_use_add2_insn): Add a mode parameter and use it instead
3628 (move2add_use_add3_insn): Likewise.
3629 (reload_cse_move2add): Update accordingly.
3631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3632 Alan Hayward <alan.hayward@arm.com>
3633 David Sherwood <david.sherwood@arm.com>
3635 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3637 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3638 * optabs.c (expand_unop): Likewise.
3640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3641 Alan Hayward <alan.hayward@arm.com>
3642 David Sherwood <david.sherwood@arm.com>
3644 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3645 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3646 (popcount_loc_descriptor): Likewise.
3647 (bswap_loc_descriptor): Likewise.
3648 (rotate_loc_descriptor): Likewise.
3649 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3650 calling the functions above.
3652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3653 Alan Hayward <alan.hayward@arm.com>
3654 David Sherwood <david.sherwood@arm.com>
3656 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3658 (try_combine): Likewise.
3659 (simplify_if_then_else): Likewise.
3660 * cse.c (cse_insn): Likewise.
3661 * dwarf2out.c (mem_loc_descriptor): Likewise.
3662 * emit-rtl.c (gen_lowpart_common): Likewise.
3663 * simplify-rtx.c (simplify_truncation): Likewise.
3664 (simplify_binary_operation_1): Likewise.
3665 (simplify_const_relational_operation): Likewise.
3666 (simplify_ternary_operation): Likewise.
3667 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3670 Alan Hayward <alan.hayward@arm.com>
3671 David Sherwood <david.sherwood@arm.com>
3673 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3674 * reload.c (push_reload): Likewise.
3675 (find_reloads): Likewise.
3677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3678 Alan Hayward <alan.hayward@arm.com>
3679 David Sherwood <david.sherwood@arm.com>
3681 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3682 (make_compound_operation_int): Likewise.
3683 (change_zero_ext): Likewise.
3684 * expr.c (convert_move): Likewise.
3685 (convert_modes): Likewise.
3686 * fwprop.c (forward_propagate_subreg): Likewise.
3687 * loop-iv.c (get_biv_step_1): Likewise.
3688 * optabs.c (widen_operand): Likewise.
3689 * postreload.c (move2add_valid_value_p): Likewise.
3690 * recog.c (simplify_while_replacing): Likewise.
3691 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3692 (simplify_binary_operation_1): Likewise. Remove redundant
3693 mode equality check.
3695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3696 Alan Hayward <alan.hayward@arm.com>
3697 David Sherwood <david.sherwood@arm.com>
3699 * combine.c (combine_simplify_rtx): Add checks for
3700 is_a <scalar_int_mode>.
3701 (simplify_if_then_else): Likewise.
3702 (make_field_assignment): Likewise.
3703 (simplify_comparison): Likewise.
3704 * ifcvt.c (noce_try_bitop): Likewise.
3705 * loop-invariant.c (canonicalize_address_mult): Likewise.
3706 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3709 Alan Hayward <alan.hayward@arm.com>
3710 David Sherwood <david.sherwood@arm.com>
3712 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3713 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3720 instead of != VOIDmode.
3721 * combine.c (if_then_else_cond): Likewise.
3722 (change_zero_ext): Likewise.
3723 * dwarf2out.c (mem_loc_descriptor): Likewise.
3724 (loc_descriptor): Likewise.
3725 * rtlanal.c (canonicalize_condition): Likewise.
3726 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3729 Alan Hayward <alan.hayward@arm.com>
3730 David Sherwood <david.sherwood@arm.com>
3732 * simplify-rtx.c (simplify_binary_operation_1): Use
3733 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3736 Alan Hayward <alan.hayward@arm.com>
3737 David Sherwood <david.sherwood@arm.com>
3739 * wide-int.h (int_traits<unsigned char>) New class.
3740 (int_traits<unsigned short>) Likewise.
3741 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3742 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3744 * combine.c (set_nonzero_bits_and_sign_copies): Use
3745 is_a <scalar_int_mode>.
3746 (find_split_point): Likewise.
3747 (combine_simplify_rtx): Likewise.
3748 (simplify_logical): Likewise.
3749 (expand_compound_operation): Likewise.
3750 (expand_field_assignment): Likewise.
3751 (make_compound_operation): Likewise.
3752 (extended_count): Likewise.
3753 (change_zero_ext): Likewise.
3754 (simplify_comparison): Likewise.
3755 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3756 (ucompare_loc_descriptor): Likewise.
3757 (minmax_loc_descriptor): Likewise.
3758 (mem_loc_descriptor): Likewise.
3759 (loc_descriptor): Likewise.
3760 * expmed.c (init_expmed_one_mode): Likewise.
3761 * lra-constraints.c (lra_constraint_offset): Likewise.
3762 * optabs.c (prepare_libcall_arg): Likewise.
3763 * postreload.c (move2add_note_store): Likewise.
3764 * reload.c (operands_match_p): Likewise.
3765 * rtl.h (load_extend_op): Likewise.
3766 * rtlhooks.c (gen_lowpart_general): Likewise.
3767 * simplify-rtx.c (simplify_truncation): Likewise.
3768 (simplify_unary_operation_1): Likewise.
3769 (simplify_binary_operation_1): Likewise.
3770 (simplify_const_binary_operation): Likewise.
3771 (simplify_const_relational_operation): Likewise.
3772 (simplify_subreg): Likewise.
3773 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3774 * var-tracking.c (adjust_mems): Likewise.
3775 (prepare_call_arguments): Likewise.
3777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3778 Alan Hayward <alan.hayward@arm.com>
3779 David Sherwood <david.sherwood@arm.com>
3781 * machmode.h (is_int_mode): New fuction.
3782 * combine.c (find_split_point): Use it.
3783 (combine_simplify_rtx): Likewise.
3784 (simplify_if_then_else): Likewise.
3785 (simplify_set): Likewise.
3786 (simplify_shift_const_1): Likewise.
3787 (simplify_comparison): Likewise.
3788 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3789 * cse.c (notreg_cost): Likewise.
3790 (cse_insn): Likewise.
3791 * cselib.c (cselib_lookup_1): Likewise.
3792 * dojump.c (do_jump_1): Likewise.
3793 (do_compare_rtx_and_jump): Likewise.
3794 * dse.c (get_call_args): Likewise.
3795 * dwarf2out.c (rtl_for_decl_init): Likewise.
3796 (native_encode_initializer): Likewise.
3797 * expmed.c (emit_store_flag_1): Likewise.
3798 (emit_store_flag): Likewise.
3799 * expr.c (convert_modes): Likewise.
3800 (store_field): Likewise.
3801 (expand_expr_real_1): Likewise.
3802 * fold-const.c (fold_read_from_constant_string): Likewise.
3803 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3804 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3805 * optabs.c (expand_binop): Likewise.
3806 (expand_unop): Likewise.
3807 (expand_abs_nojump): Likewise.
3808 (expand_one_cmpl_abs_nojump): Likewise.
3809 * simplify-rtx.c (mode_signbit_p): Likewise.
3810 (val_signbit_p): Likewise.
3811 (val_signbit_known_set_p): Likewise.
3812 (val_signbit_known_clear_p): Likewise.
3813 (simplify_relational_operation_1): Likewise.
3814 * tree.c (vector_type_mode): Likewise.
3816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3817 Alan Hayward <alan.hayward@arm.com>
3818 David Sherwood <david.sherwood@arm.com>
3820 * machmode.h (smallest_mode_for_size): Fix formatting.
3821 (smallest_int_mode_for_size): New function.
3822 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3823 instead of smallest_mode_for_size.
3824 * combine.c (make_extraction): Likewise.
3825 * config/arc/arc.c (arc_expand_movmem): Likewise.
3826 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3827 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3828 * config/s390/s390.c (s390_expand_insv): Likewise.
3829 * config/sparc/sparc.c (assign_int_registers): Likewise.
3830 * config/spu/spu.c (spu_function_value): Likewise.
3831 (spu_function_arg): Likewise.
3832 * coverage.c (get_gcov_type): Likewise.
3833 (get_gcov_unsigned_t): Likewise.
3834 * dse.c (find_shift_sequence): Likewise.
3835 * expmed.c (store_bit_field_1): Likewise.
3836 * expr.c (convert_move): Likewise.
3837 (store_field): Likewise.
3838 * internal-fn.c (expand_arith_overflow): Likewise.
3839 * optabs-query.c (get_best_extraction_insn): Likewise.
3840 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3841 * stor-layout.c (layout_type): Likewise.
3842 (initialize_sizetypes): Likewise.
3843 * targhooks.c (default_get_mask_mode): Likewise.
3844 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3847 Alan Hayward <alan.hayward@arm.com>
3848 David Sherwood <david.sherwood@arm.com>
3850 * machmode.h (opt_mode::else_blk): New function.
3851 (int_mode_for_mode): Declare.
3852 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3853 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3855 * cfgexpand.c (expand_debug_expr): Likewise.
3856 * combine.c (gen_lowpart_or_truncate): Likewise.
3857 (gen_lowpart_for_combine): Likewise.
3858 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3859 * config/avr/avr.c (avr_to_int_mode): Likewise.
3860 (avr_out_plus_1): Likewise.
3861 (avr_out_plus): Likewise.
3862 (avr_out_round): Likewise.
3863 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3864 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3865 (s390_expand_vcond): Likewise.
3866 * config/spu/spu.c (spu_split_immediate): Likewise.
3867 (spu_expand_mov): Likewise.
3868 * dse.c (get_stored_val): Likewise.
3869 * expmed.c (store_bit_field_1): Likewise.
3870 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3872 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3873 (extract_low_bits): Likewise.
3874 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3875 handling rather than repeating the check.
3876 (emit_group_store): Likewise.
3877 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3878 * optabs.c (expand_absneg_bit): Likewise.
3879 (expand_copysign_absneg): Likewise.
3880 (expand_copysign_bit): Likewise.
3881 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3882 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3883 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3884 * var-tracking.c (prepare_call_arguments): Likewise.
3885 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3886 int_mode_for_mode instead of mode_for_size.
3887 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3890 Alan Hayward <alan.hayward@arm.com>
3891 David Sherwood <david.sherwood@arm.com>
3893 * machmode.h (int_mode_for_size): New function.
3894 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3895 instead of mode_for_size.
3896 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3898 * combine.c (expand_field_assignment): Use int_mode_for_size
3899 instead of mode_for_size.
3900 (make_extraction): Likewise.
3901 (simplify_shift_const_1): Likewise.
3902 (simplify_comparison): Likewise.
3903 * dojump.c (do_jump): Likewise.
3904 * dwarf2out.c (mem_loc_descriptor): Likewise.
3905 * emit-rtl.c (init_derived_machine_modes): Likewise.
3906 * expmed.c (flip_storage_order): Likewise.
3907 (convert_extracted_bit_field): Likewise.
3908 * expr.c (copy_blkmode_from_reg): Likewise.
3909 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3910 * internal-fn.c (expand_mul_overflow): Likewise.
3911 * lower-subreg.c (simple_move): Likewise.
3912 * optabs-libfuncs.c (init_optabs): Likewise.
3913 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3914 * tree.c (vector_type_mode): Likewise.
3915 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3916 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3917 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3918 * tree-vect-stmts.c (vectorizable_load): Likewise.
3919 (vectorizable_store): Likewise.
3921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3922 Alan Hayward <alan.hayward@arm.com>
3923 David Sherwood <david.sherwood@arm.com>
3925 * coretypes.h (pod_mode): New type.
3926 (scalar_int_mode_pod): New typedef.
3927 * machmode.h (pod_mode): New class.
3928 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3929 * genmodes.c (emit_mode_int_n): Update accordingly.
3930 * lower-subreg.h (target_lower_subreg): Change type to
3931 scalar_int_mode_pod.
3932 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3933 scalar_int_mode_pod.
3935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3936 Alan Hayward <alan.hayward@arm.com>
3937 David Sherwood <david.sherwood@arm.com>
3939 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3940 machine_mode to scalar_int_mode.
3941 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3942 (rs6000_option_override_internal): Remove cast to int.
3943 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3944 machine_mode to scalar_int_mode.
3945 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3946 (rs6000_option_override_internal): Remove cast to int.
3947 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3948 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3950 * config/s390/s390.c (s390_expand_builtin): Likewise.
3951 * coretypes.h (scalar_int_mode): New type.
3952 (opt_scalar_int_mode): New typedef.
3953 * machmode.h (scalar_int_mode): New class.
3954 (scalar_int_mode::includes_p): New function.
3955 (byte_mode): Change type to scalar_int_mode.
3956 (word_mode): Likewise.
3957 (ptr_mode): Likewise.
3958 * emit-rtl.c (byte_mode): Likewise.
3959 (word_mode): Likewise.
3960 (ptr_mode): Likewise.
3961 (init_derived_machine_modes): Update accordingly.
3962 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3963 and MODE_PARTIAL_INT.
3964 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3965 opt_scalar_int_mode.
3967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3968 Alan Hayward <alan.hayward@arm.com>
3969 David Sherwood <david.sherwood@arm.com>
3971 * target.def (libgcc_floating_mode_supported_p): Take a
3973 * doc/tm.texi: Regenerate.
3974 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3976 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3977 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3985 * doc/tm.texi: Regenerate.
3986 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3987 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3988 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3989 * targhooks.h (default_floatn_mode): Likewise.
3990 * targhooks.c (default_floatn_mode): Likewise.
3991 * tree.c (build_common_tree_nodes): Update accordingly.
3993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3994 Alan Hayward <alan.hayward@arm.com>
3995 David Sherwood <david.sherwood@arm.com>
3997 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3998 (mode_iterator::iterate_p): Likewise.
3999 (mode_iterator::get_wider): Likewise.
4000 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4003 Alan Hayward <alan.hayward@arm.com>
4004 David Sherwood <david.sherwood@arm.com>
4006 * coretypes.h (opt_scalar_float_mode): New typedef.
4007 * machmode.h (float_mode_for_size): New function.
4008 * emit-rtl.c (double_mode): Delete.
4009 (init_emit_once): Use float_mode_for_size.
4010 * stor-layout.c (layout_type): Likewise.
4011 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4014 Alan Hayward <alan.hayward@arm.com>
4015 David Sherwood <david.sherwood@arm.com>
4017 * output.h (assemble_real): Take a scalar_float_mode.
4018 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4019 * config/arm/arm.md (consttable_4): Likewise.
4020 (consttable_8): Likewise.
4021 (consttable_16): Likewise.
4022 * config/mips/mips.md (consttable_float): Likewise.
4023 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4024 * varasm.c (assemble_real): Take a scalar_float_mode.
4025 (output_constant_pool_2): Update accordingly.
4026 (output_constant): Likewise.
4028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4029 Alan Hayward <alan.hayward@arm.com>
4030 David Sherwood <david.sherwood@arm.com>
4032 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4033 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4034 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4035 (native_encode_real): Likewise.
4036 (native_interpret_real): Likewise.
4037 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4038 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4045 <scalar_float_mode>. Simplify.
4046 (gen_extend_conv_libfunc): Likewise.
4048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4049 Alan Hayward <alan.hayward@arm.com>
4050 David Sherwood <david.sherwood@arm.com>
4052 * coretypes.h (scalar_float_mode): New type.
4053 * machmode.h (mode_traits::from_int): Use machine_mode if
4054 USE_ENUM_MODES is defined.
4055 (is_a): New function.
4057 (dyn_cast): Likewise.
4058 (scalar_float_mode): New class.
4059 (scalar_float_mode::includes_p): New function.
4060 (is_float_mode): Likewise.
4061 * gdbhooks.py (MachineModePrinter): New class.
4062 (build_pretty_printer): Use it for scalar_float_mode.
4063 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4064 (format_helper::format_helper): Turn into a template.
4065 * genmodes.c (get_mode_class): New function.
4066 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4067 or machine_mode if none.
4068 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4069 as_a <scalar_float_mode>.
4070 * dwarf2out.c (mem_loc_descriptor): Likewise.
4071 (insert_float): Likewise.
4072 (add_const_value_attribute): Likewise.
4073 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4074 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4075 (expand_unop): Update accordingly.
4076 (expand_abs_nojump): Likewise.
4077 (expand_copysign_absneg): Take a scalar_float_mode.
4078 (expand_copysign_bit): Likewise.
4079 (expand_copysign): Update accordingly.
4081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4082 Alan Hayward <alan.hayward@arm.com>
4083 David Sherwood <david.sherwood@arm.com>
4085 * coretypes.h (opt_mode): New class.
4086 * machmode.h (opt_mode): Likewise.
4087 (opt_mode::else_void): New function.
4088 (opt_mode::require): Likewise.
4089 (opt_mode::exists): Likewise.
4090 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4091 (GET_MODE_2XWIDER_MODE): Likewise.
4092 (mode_iterator::get_wider): Update accordingly.
4093 (mode_iterator::get_2xwider): Likewise.
4094 (mode_iterator::get_known_wider): Likewise, turning into a template.
4095 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4096 forcing a wider mode to exist.
4097 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4098 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4099 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4100 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4101 * lower-subreg.c (init_lower_subreg): Likewise.
4102 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4103 on the final iteration.
4104 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4105 a wider mode exists before asking for a move pattern.
4106 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4107 forcing a wider mode to exist.
4108 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4109 returning false if no such mode exists.
4110 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4111 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4112 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4113 Avoid checking for a MODE_INT if we already know the mode is not a
4115 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4116 forcing a wider mode to exist.
4117 (expmed_mult_highpart_optab): Likewise.
4118 (expmed_mult_highpart): Likewise.
4119 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4121 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4122 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4123 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4124 * internal-fn.c (expand_mul_overflow): Update use of
4125 GET_MODE_2XWIDER_MODE.
4126 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4127 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4128 GET_MODE_WIDER_MODE.
4129 (convert_plusminus_to_widen): Likewise.
4130 * tree-switch-conversion.c (array_value_type): Likewise.
4131 * var-tracking.c (emit_note_insn_var_location): Likewise.
4132 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4133 Return false inside rather than outside the loop if no wider mode
4135 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4136 and GET_MODE_2XWIDER_MODE
4137 (can_compare_p): Use else_void.
4138 * gdbhooks.py (OptMachineModePrinter): New class.
4139 (build_pretty_printer): Use it for opt_mode.
4141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4142 Alan Hayward <alan.hayward@arm.com>
4143 David Sherwood <david.sherwood@arm.com>
4145 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4146 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4149 Alan Hayward <alan.hayward@arm.com>
4150 David Sherwood <david.sherwood@arm.com>
4152 * machmode.h (mode_traits): New structure.
4153 (get_narrowest_mode): New function.
4154 (mode_iterator::start): Likewise.
4155 (mode_iterator::iterate_p): Likewise.
4156 (mode_iterator::get_wider): Likewise.
4157 (mode_iterator::get_known_wider): Likewise.
4158 (mode_iterator::get_2xwider): Likewise.
4159 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4160 (FOR_EACH_MODE): Likewise.
4161 (FOR_EACH_MODE_FROM): Likewise.
4162 (FOR_EACH_MODE_UNTIL): Likewise.
4163 (FOR_EACH_WIDER_MODE): Likewise.
4164 (FOR_EACH_2XWIDER_MODE): Likewise.
4165 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4166 * combine.c (simplify_comparison): Likewise
4167 * config/i386/i386.c (type_natural_mode): Likewise.
4168 * cse.c (cse_insn): Likewise.
4169 * dse.c (find_shift_sequence): Likewise.
4170 * emit-rtl.c (init_derived_machine_modes): Likewise.
4171 (init_emit_once): Likewise.
4172 * explow.c (hard_function_value): Likewise.
4173 * expmed.c (extract_fixed_bit_field_1): Likewise.
4174 (extract_bit_field_1): Likewise.
4175 (expand_divmod): Likewise.
4176 (emit_store_flag_1): Likewise.
4177 * expr.c (init_expr_target): Likewise.
4178 (convert_move): Likewise.
4179 (alignment_for_piecewise_move): Likewise.
4180 (widest_int_mode_for_size): Likewise.
4181 (emit_block_move_via_movmem): Likewise.
4182 (copy_blkmode_to_reg): Likewise.
4183 (set_storage_via_setmem): Likewise.
4184 (compress_float_constant): Likewise.
4185 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4186 * optabs-query.c (get_best_extraction_insn): Likewise.
4187 * optabs.c (expand_binop): Likewise.
4188 (expand_twoval_unop): Likewise.
4189 (expand_twoval_binop): Likewise.
4190 (widen_leading): Likewise.
4191 (widen_bswap): Likewise.
4192 (expand_parity): Likewise.
4193 (expand_unop): Likewise.
4194 (prepare_cmp_insn): Likewise.
4195 (prepare_float_lib_cmp): Likewise.
4196 (expand_float): Likewise.
4197 (expand_fix): Likewise.
4198 (expand_sfix_optab): Likewise.
4199 * postreload.c (move2add_use_add2_insn): Likewise.
4200 * reg-stack.c (reg_to_stack): Likewise.
4201 * reginfo.c (choose_hard_reg_mode): Likewise.
4202 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4203 * stor-layout.c (mode_for_size): Likewise.
4204 (smallest_mode_for_size): Likewise.
4205 (mode_for_vector): Likewise.
4206 (finish_bitfield_representative): Likewise.
4207 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4208 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4209 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4210 * var-tracking.c (prepare_call_arguments): Likewise.
4212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4213 Alan Hayward <alan.hayward@arm.com>
4214 David Sherwood <david.sherwood@arm.com>
4216 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4217 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4218 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4219 * machmode.h (mode_size): Move earlier in file.
4220 (mode_precision): Likewise.
4221 (mode_inner): Likewise.
4222 (mode_nunits): Likewise.
4223 (mode_unit_size): Likewise.
4224 (unit_unit_precision): Likewise.
4225 (mode_wider): Likewise.
4226 (mode_2xwider): Likewise.
4227 (machine_mode): New class.
4228 (mode_to_bytes): New function.
4229 (mode_to_bits): Likewise.
4230 (mode_to_precision): Likewise.
4231 (mode_to_inner): Likewise.
4232 (mode_to_unit_size): Likewise.
4233 (mode_to_unit_precision): Likewise.
4234 (mode_to_nunits): Likewise.
4235 (GET_MODE_SIZE): Use mode_to_bytes.
4236 (GET_MODE_BITSIZE): Use mode_to_bits.
4237 (GET_MODE_PRECISION): Use mode_to_precision.
4238 (GET_MODE_INNER): Use mode_to_inner.
4239 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4240 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4241 (GET_MODE_NUNITS): Use mode_to_nunits.
4242 * system.h (ALWAYS_INLINE): New macro.
4243 * config/powerpcspe/powerpcspe-c.c
4244 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4245 int for arg1_mode and arg2_mode.
4247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4248 Alan Hayward <alan.hayward@arm.com>
4249 David Sherwood <david.sherwood@arm.com>
4251 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4252 Prefix mode names with E_ in case statements.
4253 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4254 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4255 (aarch64_split_simd_move): Likewise.
4256 (aarch64_gen_storewb_pair): Likewise.
4257 (aarch64_gen_loadwb_pair): Likewise.
4258 (aarch64_gen_store_pair): Likewise.
4259 (aarch64_gen_load_pair): Likewise.
4260 (aarch64_get_condition_code_1): Likewise.
4261 (aarch64_constant_pool_reload_icode): Likewise.
4262 (get_rsqrte_type): Likewise.
4263 (get_rsqrts_type): Likewise.
4264 (get_recpe_type): Likewise.
4265 (get_recps_type): Likewise.
4266 (aarch64_gimplify_va_arg_expr): Likewise.
4267 (aarch64_simd_container_mode): Likewise.
4268 (aarch64_emit_load_exclusive): Likewise.
4269 (aarch64_emit_store_exclusive): Likewise.
4270 (aarch64_expand_compare_and_swap): Likewise.
4271 (aarch64_gen_atomic_cas): Likewise.
4272 (aarch64_emit_bic): Likewise.
4273 (aarch64_emit_atomic_swap): Likewise.
4274 (aarch64_emit_atomic_load_op): Likewise.
4275 (aarch64_evpc_trn): Likewise.
4276 (aarch64_evpc_uzp): Likewise.
4277 (aarch64_evpc_zip): Likewise.
4278 (aarch64_evpc_ext): Likewise.
4279 (aarch64_evpc_rev): Likewise.
4280 (aarch64_evpc_dup): Likewise.
4281 (aarch64_gen_ccmp_first): Likewise.
4282 (aarch64_gen_ccmp_next): Likewise.
4283 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4284 (alpha_emit_xfloating_libcall): Likewise.
4285 (emit_insxl): Likewise.
4286 (alpha_arg_type): Likewise.
4287 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4288 (arc_preferred_simd_mode): Likewise.
4289 (arc_secondary_reload): Likewise.
4290 (get_arc_condition_code): Likewise.
4291 (arc_print_operand): Likewise.
4292 (arc_legitimate_constant_p): Likewise.
4293 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4294 * config/arc/arc.md (casesi_load): Likewise.
4295 (casesi_compact_jump): Likewise.
4296 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4297 (cc_use_register): Likewise.
4298 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4299 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4300 (arm_init_iwmmxt_builtins): Likewise.
4301 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4302 (neon_expand_vector_init): Likewise.
4303 (arm_attr_length_move_neon): Likewise.
4304 (maybe_get_arm_condition_code): Likewise.
4305 (arm_emit_vector_const): Likewise.
4306 (arm_preferred_simd_mode): Likewise.
4307 (arm_output_iwmmxt_tinsr): Likewise.
4308 (thumb1_output_casesi): Likewise.
4309 (thumb2_output_casesi): Likewise.
4310 (arm_emit_load_exclusive): Likewise.
4311 (arm_emit_store_exclusive): Likewise.
4312 (arm_expand_compare_and_swap): Likewise.
4313 (arm_evpc_neon_vuzp): Likewise.
4314 (arm_evpc_neon_vzip): Likewise.
4315 (arm_evpc_neon_vrev): Likewise.
4316 (arm_evpc_neon_vtrn): Likewise.
4317 (arm_evpc_neon_vext): Likewise.
4318 (arm_validize_comparison): Likewise.
4319 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4320 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4321 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4322 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4323 (c6x_preferred_simd_mode): Likewise.
4324 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4325 (epiphany_rtx_costs): Likewise.
4326 * config/epiphany/predicates.md (proper_comparison_operator):
4328 * config/frv/frv.c (condexec_memory_operand): Likewise.
4329 (frv_emit_move): Likewise.
4330 (output_move_single): Likewise.
4331 (output_condmove_single): Likewise.
4332 (frv_hard_regno_mode_ok): Likewise.
4333 (frv_matching_accg_mode): Likewise.
4334 * config/h8300/h8300.c (split_adds_subs): Likewise.
4335 (h8300_rtx_costs): Likewise.
4336 (h8300_print_operand): Likewise.
4337 (compute_mov_length): Likewise.
4338 (output_logical_op): Likewise.
4339 (compute_logical_op_length): Likewise.
4340 (compute_logical_op_cc): Likewise.
4341 (h8300_shift_needs_scratch_p): Likewise.
4342 (output_a_shift): Likewise.
4343 (compute_a_shift_length): Likewise.
4344 (compute_a_shift_cc): Likewise.
4345 (expand_a_rotate): Likewise.
4346 (output_a_rotate): Likewise.
4347 * config/i386/i386.c (classify_argument): Likewise.
4348 (function_arg_advance_32): Likewise.
4349 (function_arg_32): Likewise.
4350 (function_arg_64): Likewise.
4351 (function_value_64): Likewise.
4352 (ix86_gimplify_va_arg): Likewise.
4353 (ix86_legitimate_constant_p): Likewise.
4354 (put_condition_code): Likewise.
4355 (split_double_mode): Likewise.
4356 (ix86_avx256_split_vector_move_misalign): Likewise.
4357 (ix86_expand_vector_logical_operator): Likewise.
4358 (ix86_split_idivmod): Likewise.
4359 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4360 (ix86_build_const_vector): Likewise.
4361 (ix86_build_signbit_mask): Likewise.
4362 (ix86_match_ccmode): Likewise.
4363 (ix86_cc_modes_compatible): Likewise.
4364 (ix86_expand_branch): Likewise.
4365 (ix86_expand_sse_cmp): Likewise.
4366 (ix86_expand_sse_movcc): Likewise.
4367 (ix86_expand_int_sse_cmp): Likewise.
4368 (ix86_expand_vec_perm_vpermi2): Likewise.
4369 (ix86_expand_vec_perm): Likewise.
4370 (ix86_expand_sse_unpack): Likewise.
4371 (ix86_expand_int_addcc): Likewise.
4372 (ix86_split_to_parts): Likewise.
4373 (ix86_vectorize_builtin_gather): Likewise.
4374 (ix86_vectorize_builtin_scatter): Likewise.
4375 (avx_vpermilp_parallel): Likewise.
4376 (inline_memory_move_cost): Likewise.
4377 (ix86_tieable_integer_mode_p): Likewise.
4378 (x86_maybe_negate_const_int): Likewise.
4379 (ix86_expand_vector_init_duplicate): Likewise.
4380 (ix86_expand_vector_init_one_nonzero): Likewise.
4381 (ix86_expand_vector_init_one_var): Likewise.
4382 (ix86_expand_vector_init_concat): Likewise.
4383 (ix86_expand_vector_init_interleave): Likewise.
4384 (ix86_expand_vector_init_general): Likewise.
4385 (ix86_expand_vector_set): Likewise.
4386 (ix86_expand_vector_extract): Likewise.
4387 (emit_reduc_half): Likewise.
4388 (ix86_emit_i387_round): Likewise.
4389 (ix86_mangle_type): Likewise.
4390 (ix86_expand_round_sse4): Likewise.
4391 (expand_vec_perm_blend): Likewise.
4392 (canonicalize_vector_int_perm): Likewise.
4393 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4394 (expand_vec_perm_1): Likewise.
4395 (expand_vec_perm_interleave3): Likewise.
4396 (expand_vec_perm_even_odd_pack): Likewise.
4397 (expand_vec_perm_even_odd_1): Likewise.
4398 (expand_vec_perm_broadcast_1): Likewise.
4399 (ix86_vectorize_vec_perm_const_ok): Likewise.
4400 (ix86_expand_vecop_qihi): Likewise.
4401 (ix86_expand_mul_widen_hilo): Likewise.
4402 (ix86_expand_sse2_abs): Likewise.
4403 (ix86_expand_pextr): Likewise.
4404 (ix86_expand_pinsr): Likewise.
4405 (ix86_preferred_simd_mode): Likewise.
4406 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4407 * config/i386/sse.md (*andnot<mode>3): Likewise.
4408 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4409 (*<code><mode>3): Likewise.
4410 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4411 (ia64_expand_atomic_op): Likewise.
4412 (ia64_arg_type): Likewise.
4413 (ia64_mode_to_int): Likewise.
4414 (ia64_scalar_mode_supported_p): Likewise.
4415 (ia64_vector_mode_supported_p): Likewise.
4416 (expand_vec_perm_broadcast): Likewise.
4417 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4418 (iq2000_function_arg_advance): Likewise.
4419 (iq2000_function_arg): Likewise.
4420 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4421 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4422 (m68k_libcall_value): Likewise.
4423 (m68k_function_value): Likewise.
4424 (sched_attr_op_type): Likewise.
4425 * config/mcore/mcore.c (mcore_output_move): Likewise.
4426 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4428 (microblaze_function_arg): Likewise.
4429 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4430 (mips_print_operand): Likewise.
4431 (mips_mode_ok_for_mov_fmt_p): Likewise.
4432 (mips_vector_mode_supported_p): Likewise.
4433 (mips_preferred_simd_mode): Likewise.
4434 (mips_expand_vpc_loongson_even_odd): Likewise.
4435 (mips_expand_vec_unpack): Likewise.
4436 (mips_expand_vi_broadcast): Likewise.
4437 (mips_expand_vector_init): Likewise.
4438 (mips_expand_vec_reduc): Likewise.
4439 (mips_expand_msa_cmp): Likewise.
4440 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4441 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4442 (cc_flags_for_mode): Likewise.
4443 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4444 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4445 (nds32_output_casesi_pc_relative): Likewise.
4446 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4447 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4448 (nvptx_gen_unpack): Likewise.
4449 (nvptx_gen_pack): Likewise.
4450 (nvptx_gen_shuffle): Likewise.
4451 (nvptx_gen_wcast): Likewise.
4452 (nvptx_preferred_simd_mode): Likewise.
4453 * config/pa/pa.c (pa_secondary_reload): Likewise.
4454 * config/pa/predicates.md (base14_operand): Likewise.
4455 * config/powerpcspe/powerpcspe-c.c
4456 (altivec_resolve_overloaded_builtin): Likewise.
4457 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4459 (rs6000_preferred_simd_mode): Likewise.
4460 (output_vec_const_move): Likewise.
4461 (rs6000_expand_vector_extract): Likewise.
4462 (rs6000_split_vec_extract_var): Likewise.
4463 (reg_offset_addressing_ok_p): Likewise.
4464 (rs6000_legitimate_offset_address_p): Likewise.
4465 (rs6000_legitimize_address): Likewise.
4466 (rs6000_emit_set_const): Likewise.
4467 (rs6000_const_vec): Likewise.
4468 (rs6000_emit_move): Likewise.
4469 (spe_build_register_parallel): Likewise.
4470 (rs6000_darwin64_record_arg_recurse): Likewise.
4471 (swap_selector_for_mode): Likewise.
4472 (spe_init_builtins): Likewise.
4473 (paired_init_builtins): Likewise.
4474 (altivec_init_builtins): Likewise.
4475 (do_load_for_compare): Likewise.
4476 (rs6000_generate_compare): Likewise.
4477 (rs6000_expand_float128_convert): Likewise.
4478 (emit_load_locked): Likewise.
4479 (emit_store_conditional): Likewise.
4480 (rs6000_output_function_epilogue): Likewise.
4481 (rs6000_handle_altivec_attribute): Likewise.
4482 (rs6000_function_value): Likewise.
4483 (emit_fusion_gpr_load): Likewise.
4484 (emit_fusion_p9_load): Likewise.
4485 (emit_fusion_p9_store): Likewise.
4486 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4487 (fusion_gpr_mem_load): Likewise.
4488 (fusion_addis_mem_combo_load): Likewise.
4489 (fusion_addis_mem_combo_store): Likewise.
4490 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4491 (fusion_gpr_mem_load): Likewise.
4492 (fusion_addis_mem_combo_load): Likewise.
4493 (fusion_addis_mem_combo_store): Likewise.
4494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4496 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4497 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4498 (rs6000_preferred_simd_mode): Likewise.
4499 (output_vec_const_move): Likewise.
4500 (rs6000_expand_vector_extract): Likewise.
4501 (rs6000_split_vec_extract_var): Likewise.
4502 (reg_offset_addressing_ok_p): Likewise.
4503 (rs6000_legitimate_offset_address_p): Likewise.
4504 (rs6000_legitimize_address): Likewise.
4505 (rs6000_emit_set_const): Likewise.
4506 (rs6000_const_vec): Likewise.
4507 (rs6000_emit_move): Likewise.
4508 (rs6000_darwin64_record_arg_recurse): Likewise.
4509 (swap_selector_for_mode): Likewise.
4510 (paired_init_builtins): Likewise.
4511 (altivec_init_builtins): Likewise.
4512 (rs6000_expand_float128_convert): Likewise.
4513 (emit_load_locked): Likewise.
4514 (emit_store_conditional): Likewise.
4515 (rs6000_output_function_epilogue): Likewise.
4516 (rs6000_handle_altivec_attribute): Likewise.
4517 (rs6000_function_value): Likewise.
4518 (emit_fusion_gpr_load): Likewise.
4519 (emit_fusion_p9_load): Likewise.
4520 (emit_fusion_p9_store): Likewise.
4521 * config/rx/rx.c (rx_gen_move_template): Likewise.
4522 (flags_from_mode): Likewise.
4523 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4524 (s390_slb_comparison): Likewise.
4525 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4526 (s390_vector_mode_supported_p): Likewise.
4527 (s390_cc_modes_compatible): Likewise.
4528 (s390_match_ccmode_set): Likewise.
4529 (s390_canonicalize_comparison): Likewise.
4530 (s390_emit_compare_and_swap): Likewise.
4531 (s390_branch_condition_mask): Likewise.
4532 (s390_rtx_costs): Likewise.
4533 (s390_secondary_reload): Likewise.
4534 (__SECONDARY_RELOAD_CASE): Likewise.
4535 (s390_expand_cs): Likewise.
4536 (s390_preferred_simd_mode): Likewise.
4537 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4538 * config/sh/sh.c (sh_print_operand): Likewise.
4539 (dump_table): Likewise.
4540 (sh_secondary_reload): Likewise.
4541 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4542 * config/sh/sh.md (casesi_worker_1): Likewise.
4543 (casesi_worker_2): Likewise.
4544 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4545 (fcc_comparison_operator): Likewise.
4546 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4547 (emit_soft_tfmode_cvt): Likewise.
4548 (sparc_preferred_simd_mode): Likewise.
4549 (output_cbranch): Likewise.
4550 (sparc_print_operand): Likewise.
4551 (sparc_expand_vec_perm_bmask): Likewise.
4552 (vector_init_bshuffle): Likewise.
4553 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4554 (spu_vector_mode_supported_p): Likewise.
4555 (spu_expand_insv): Likewise.
4556 (spu_emit_branch_or_set): Likewise.
4557 (spu_handle_vector_attribute): Likewise.
4558 (spu_builtin_splats): Likewise.
4559 (spu_builtin_extract): Likewise.
4560 (spu_builtin_promote): Likewise.
4561 (spu_expand_sign_extend): Likewise.
4562 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4563 (tilegx_simd_int): Likewise.
4564 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4565 (tilepro_simd_int): Likewise.
4566 * config/v850/v850.c (const_double_split): Likewise.
4567 (v850_print_operand): Likewise.
4568 (ep_memory_offset): Likewise.
4569 * config/vax/vax.c (vax_rtx_costs): Likewise.
4570 (vax_output_int_move): Likewise.
4571 (vax_output_int_add): Likewise.
4572 (vax_output_int_subtract): Likewise.
4573 * config/visium/predicates.md (visium_branch_operator): Likewise.
4574 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4575 (visium_print_operand_address): Likewise.
4576 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4577 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4578 (xtensa_expand_conditional_branch): Likewise.
4579 (xtensa_copy_incoming_a7): Likewise.
4580 (xtensa_output_literal): Likewise.
4581 * dfp.c (decimal_real_maxval): Likewise.
4582 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4585 Alan Hayward <alan.hayward@arm.com>
4586 David Sherwood <david.sherwood@arm.com>
4588 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4589 (mode_nunits_inline): Likewise.
4590 (mode_inner_inline): Likewise.
4591 (mode_unit_size_inline): Likewise.
4592 (mode_unit_precision_inline): Likewise.
4593 (emit_insn_modes_h): Likewise. Also emit a #define of the
4595 (emit_mode_wider): Add an E_ prefix to mode names.
4596 (emit_mode_complex): Likewise.
4597 (emit_mode_inner): Likewise.
4598 (emit_mode_adjustments): Likewise.
4599 (emit_mode_int_n): Likewise.
4600 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4601 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4602 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4603 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4604 (CRC32_BUILTIN, ENTRY): Likewise.
4605 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4606 (aarch64_pop_regs): Likewise.
4607 (aarch64_process_components): Likewise.
4608 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4609 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4610 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4611 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4612 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4613 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4615 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4616 (pcmpistr): Likewise.
4617 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4618 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4619 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4621 * config/rl78/rl78.c (mduc_regs): Likewise.
4622 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4623 (htm_expand_builtin): Likewise.
4624 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4625 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4626 * config/xtensa/xtensa.c (print_operand): Likewise.
4627 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4628 (NUM_MODE_VECTOR_INT): Likewise.
4629 * genoutput.c (null_operand): Likewise.
4630 (output_operand_data): Likewise.
4631 * genrecog.c (print_parameter_value): Likewise.
4632 * lra.c (debug_operand_data): Likewise.
4634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4636 * dwarf2out.c (scompare_loc_descriptor_wide)
4637 (scompare_loc_descriptor_narrow): New functions, split out from...
4638 (scompare_loc_descriptor): ...here.
4639 * expmed.c (emit_store_flag_int): New function, split out from...
4640 (emit_store_flag): ...here.
4642 2017-08-30 Richard Biener <rguenther@suse.de>
4644 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4645 (dwarf2out_early_finish): Move setting of AT_pubnames from
4646 early debug output to early finish.
4648 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4650 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4651 and -mdata-region to the assembler.
4653 2017-08-30 Richard Biener <rguenther@suse.de>
4655 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4657 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4658 (dwarf2out_early_global_decl): Only generate a DIE for the
4659 abstract origin if it doesn't already exist or is a declaration DIE.
4660 (resolve_addr): Do not add the linkage name twice when
4661 generating a stub DIE for the DW_TAG_GNU_call_site target.
4663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4665 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4666 Use machine_mode rather than int for arg1_mode.
4668 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4671 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4672 that the second argument of the built-in functions to unpack
4673 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4674 switch statement instead a lot of if statements.
4675 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4676 Allow 64-bit values to be in Altivec registers as well as
4677 traditional floating point registers.
4678 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4680 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4682 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4683 MAX_REGS_PER_ADDRESS == 1.
4685 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4687 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4688 * config/i386/i386.c (ix86_option_override_internal): Simplify
4689 setting of opts->x_flag_entry.
4691 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4692 Jakub Jelinek <jakub@redhat.com>
4693 Richard Biener <rguenther@suse.de>
4695 PR tree-optimization/81503
4696 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4697 folded constant fits in the target type; reorder tests for clarity.
4699 2017-08-29 Martin Liska <mliska@suse.cz>
4701 * passes.def: Include pass_lower_switch.
4702 * stmt.c (dump_case_nodes): Remove and move to
4703 tree-switch-conversion.
4704 (case_values_threshold): Likewise.
4705 (expand_switch_as_decision_tree_p): Likewise.
4706 (emit_case_decision_tree): Likewise.
4707 (expand_case): Likewise.
4708 (balance_case_nodes): Likewise.
4709 (node_has_low_bound): Likewise.
4710 (node_has_high_bound): Likewise.
4711 (node_is_bounded): Likewise.
4712 (emit_case_nodes): Likewise.
4713 (struct simple_case_node): New struct.
4714 (add_case_node): Remove.
4715 (emit_case_dispatch_table): Use vector instead of case_list.
4716 (reset_out_edges_aux): Remove.
4717 (compute_cases_per_edge): Likewise.
4718 (expand_case): Build list of simple_case_node.
4719 (expand_sjlj_dispatch_table): Use it.
4720 * tree-switch-conversion.c (struct case_node): Moved from
4721 stmt.c and adjusted.
4722 (emit_case_nodes): Likewise.
4723 (node_has_low_bound): Likewise.
4724 (node_has_high_bound): Likewise.
4725 (node_is_bounded): Likewise.
4726 (case_values_threshold): Likewise.
4727 (reset_out_edges_aux): Likewise.
4728 (compute_cases_per_edge): Likewise.
4729 (add_case_node): Likewise.
4730 (dump_case_nodes): Likewise.
4731 (balance_case_nodes): Likewise.
4732 (expand_switch_as_decision_tree_p): Likewise.
4733 (emit_jump): Likewise.
4734 (emit_case_decision_tree): Likewise.
4735 (try_switch_expansion): Likewise.
4736 (do_jump_if_equal): Likewise.
4737 (emit_cmp_and_jump_insns): Likewise.
4738 (fix_phi_operands_for_edge): New function.
4739 (record_phi_operand_mapping): Likewise.
4740 (class pass_lower_switch): New pass.
4741 (pass_lower_switch::execute): New function.
4742 (make_pass_lower_switch): Likewise.
4743 (conditional_probability):
4744 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4745 * tree-pass.h: Add make_pass_lower_switch.
4747 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4750 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4753 2017-08-29 Richard Biener <rguenther@suse.de>
4755 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4756 we do not add a DW_AT_inline attribute twice.
4757 (gen_subprogram_die): Remove code setting DW_AT_inline on
4758 DECL_ABSTRACT_P nodes.
4760 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4762 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4763 calls to internal functions.
4764 (gimplify_modify_expr): Likewise.
4765 * tree-call-cdce.c (use_internal_fn): Likewise.
4766 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4767 (convert_to_divmod): Set the nothrow flag.
4768 * tree-if-conv.c (predicate_mem_writes): Likewise.
4769 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4770 (vectorizable_call): Likewise.
4771 (vectorizable_store): Likewise.
4772 (vectorizable_load): Likewise.
4773 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4774 (vect_recog_mask_conversion_pattern): Likewise.
4776 2017-08-29 Martin Liska <mliska@suse.cz>
4779 * gcc.c (driver_handle_option): Add new argument.
4780 * opts-common.c (handle_option): Pass
4781 target_option_override_hook.
4782 * opts-global.c (lang_handle_option): Add new option.
4783 (set_default_handlers): Add new argument.
4784 (decode_options): Likewise.
4785 * opts.c (target_handle_option): Likewise.
4786 (common_handle_option): Call target_option_override_hook.
4787 * opts.h (struct cl_option_handler_func): Add hook for
4788 target option override.
4789 (struct cl_option_handlers): Likewise.
4790 (set_default_handlers): Add new argument.
4791 (decode_options): Likewise.
4792 (common_handle_option): Likewise.
4793 (target_handle_option): Likewise.
4794 * toplev.c (toplev::main): Pass targetm.target_option.override
4797 2017-08-29 Richard Biener <rguenther@suse.de>
4798 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4800 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4801 life to the active subtree.
4803 2017-08-28 Jeff Law <law@redhat.com>
4805 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4806 derive_equivalences.
4807 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4809 (edge_info::derive_equivalences): New private member function
4811 * tree-ssa-dom.c (class edge_info): Changed from a struct
4812 to a class. Add ctor/dtor, methods and data members.
4813 (edge_info::edge_info): Renamed from allocate_edge_info.
4814 Initialize additional members.
4815 (edge_info::~edge_info): New.
4816 (free_dom_edge_info): Delete the edge info.
4817 (record_edge_info): Use new class & associated member functions.
4818 Tighten forms for testing for edge equivalences.
4819 (record_temporary_equivalences): Iterate over the simple
4820 equivalences rather than assuming there's only one per edge.
4821 (cprop_into_successor_phis): Iterate over the simple
4822 equivalences rather than assuming there's only one per edge.
4823 (optimize_stmt): Use operand_equal_p rather than pointer
4824 equality for mini-DSE code.
4826 2017-08-28 Nathan Sidwell <nathan@acm.org>
4828 * gcc.c (execute): Fold SIGPIPE handling into switch
4829 statement. Adjust internal error message.
4831 2017-08-28 Richard Biener <rguenther@suse.de>
4834 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4835 Do nothing for removed DIEs.
4837 2017-08-28 Richard Biener <rguenther@suse.de>
4839 PR tree-optimization/81977
4840 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4843 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4846 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4847 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4848 using targetm.gen_mem_thread_fence.
4850 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4853 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4854 predicate to register_operand. Reorder operands.
4855 (*btr<mode>): Ditto.
4856 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4857 (*btr<mode>_mask): Ditto.
4859 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4861 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4862 * config/rs6000/xmmintrin.h: New file.
4863 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4865 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4868 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4869 parameter and_insn and return it.
4870 (recombine_lvx_pattern): Insert a copy to ensure availability of
4871 the base register of the copied masking operation at the point of
4872 the instruction replacement.
4873 (recombine_stvx_pattern): Likewise.
4875 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4877 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4878 undocumented switches.
4879 (-mpower9-dform-vector): Likewise.
4880 (-mpower9-dform): Likewise.
4881 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4882 comments to delete references to -mpower9-dform* switches.
4883 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4884 Delete reference to -mpower9-dform* switches, test for
4885 -mpower9-vector instead.
4886 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4887 (OTHER_P9_VECTOR_MASKS): Likewise.
4888 (POWERPC_MASKS): Likewise.
4889 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4890 tests against -mpower9-dform* to -mpower9-vector. Delete code
4891 that checked for -mpower9-dform* consistancy with other options.
4892 Add test for -mpower9-misc to enable other power9 switches.
4893 (rs6000_init_hard_regno_mode_ok): Likewise.
4894 (rs6000_option_override_internal): Likewise.
4895 (rs6000_emit_prologue): Likewise.
4896 (rs6000_emit_epilogue): Likewise.
4897 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4898 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4899 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4901 (emit_fusion_p9_store): Likewise.
4902 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4903 resetting these macros if the assembler does not support ISA 3.0
4905 (TARGET_P9_DFORM_VECTOR): Likewise.
4906 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4907 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4909 2017-08-25 Alan Modra <amodra@gmail.com>
4911 PR rtl-optimization/81747
4912 * cse.c (cse_extended_basic_block): Don't attempt to record
4913 equivalences for degenerate conditional branches that branch
4914 to their fall-through.
4916 2017-08-24 Martin Sebor <msebor@redhat.com>
4919 * gimple-fold.c (size_must_be_zero_p): New function.
4920 (gimple_fold_builtin_memory_op): Call it.
4922 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4924 * config/rs6000/mm_malloc.h: New file.
4926 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4928 PR tree-optimization/81913
4929 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4930 analysis when either IVs in condition can wrap.
4932 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4934 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4935 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4937 2017-08-24 Richard Biener <rguenther@suse.de>
4940 * targhooks.c (default_target_can_inline_p): Properly
4941 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4942 is present and always compare.
4943 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4944 infer -mfpmath=sse from TARGET_SSE_P.
4945 (ix86_can_inline_p): Properly use target_option_default_node when
4946 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4948 2017-08-24 Richard Biener <rguenther@suse.de>
4951 * dwarf2out.c (output_die): Handle flag_generate_offload like
4953 (output_comp_unit): Likewise.
4954 (gen_array_type_die): Likewise.
4955 (dwarf2out_early_finish): Likewise.
4956 (note_variable_value_in_expr): Likewise.
4957 (dwarf2out_finish): Likewise. Adjust assert.
4958 * cgraphunit.c (symbol_table::compile): Move setting of
4959 flag_generate_offload earlier ...
4960 (symbol_table::finalize_compilation_unit): ... here, before
4961 early debug finalization.
4963 2017-08-24 Richard Biener <rguenther@suse.de>
4965 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4966 and ipa-fnsummary.h.
4967 (ix86_can_inline_p): When ix86_fpmath flags do not match
4968 check whether the callee uses FP math at all.
4970 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4973 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4974 instead of TYPE_PRECISION.
4976 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4977 Alan Hayward <alan.hayward@arm.com>
4978 David Sherwood <david.sherwood@arm.com>
4980 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4982 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4984 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4986 (ix86_frame::stack_realign_allocate): New field.
4987 (struct machine_frame_state): Modify comments.
4988 (machine_frame_state::sp_realigned_fp_end): New field.
4989 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4991 (sp_valid_at): Add assertion to assure no attempt to access invalid
4992 offset of a realigned stack.
4993 (fp_valid_at): Likewise.
4994 (choose_baseaddr): Modify comments.
4995 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4996 ix86_expand_prologue.
4997 (ix86_expand_prologue): Modify stack realignment and allocation.
4998 (ix86_expand_epilogue): Modify comments.
4999 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5000 avx2_runtime, avx512f, and avx512f_runtime.
5002 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5004 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5005 (mstackrealign): Do not init to -1.
5006 * config/i386/i386.c (ix86_option_override_internal):
5007 Check opts_set, not opts when setting default value of
5008 opts->x_ix86_force_align_arg_pointer.
5010 2017-08-23 Richard Biener <rguenther@suse.de>
5012 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5013 lang_hooks.decl_printable_name.
5014 * print-rtl-function.c (print_rtx_function): Likewise.
5015 * tree-pretty-print.c (dump_function_header): Likewise.
5017 2017-08-23 Richard Biener <rguenther@suse.de>
5020 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5021 -g0 at compile-time.
5023 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5026 * doc/sourcebuild.texi (Other hardware attributes):
5029 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5032 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5033 Fix single-use check.
5035 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5037 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5039 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5041 * doc/install.texi: Modify to add more details on running selected
5044 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5046 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5047 is combined with -mabi=ms.
5048 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5051 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5053 PR tree-optimization/81488
5054 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5055 and cached_basis fields.
5056 (MAX_SPREAD): New constant.
5057 (alloc_cand_and_find_basis): Initialize new fields.
5058 (clear_visited): New function.
5059 (create_phi_basis_1): Rename from create_phi_basis, set visited
5060 and cached_basis fields.
5061 (create_phi_basis): New wrapper function.
5062 (phi_add_costs_1): Rename from phi_add_costs, add spread
5063 parameter, set visited field, short-circuit when limits reached.
5064 (phi_add_costs): New wrapper function.
5065 (record_phi_increments_1): Rename from record_phi_increments, set
5067 (record_phi_increments): New wrapper function.
5068 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5069 (phi_incr_cost): New wrapper function.
5070 (all_phi_incrs_profitable_1): Rename from
5071 all_phi_incrs_profitable, set visited field.
5072 (all_phi_incrs_profitable): New wrapper function.
5074 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5075 Alan Hayward <alan.hayward@arm.com>
5076 David Sherwood <david.sherwood@arm.com>
5078 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5079 that takes the outer and inner modes.
5080 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5081 comparison as the canonical test for a paradoxical subreg.
5082 * combine.c (simplify_set): Use paradoxical_subreg_p.
5083 (make_extraction): Likewise.
5084 (force_to_mode): Likewise.
5085 (rtx_equal_for_field_assignment_p): Likewise.
5086 (gen_lowpart_for_combine): Likewise.
5087 (simplify_comparison): Likewise.
5088 * cse.c (equiv_constant): Likewise.
5089 * expmed.c (store_bit_field_1): Likewise.
5090 * final.c (alter_subreg): Likewise.
5091 * fwprop.c (propagate_rtx): Likewise.
5092 (forward_propagate_subreg): Likewise.
5093 * ira-conflicts.c (ira_build_conflicts): Likewise.
5094 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5095 * lra-constraints.c (curr_insn_transform): Likewise.
5096 (split_reg): Likewise.
5097 * lra-eliminations.c (move_plus_up): Likewise.
5098 (lra_eliminate_regs_1): Likewise.
5099 * recog.c (general_operand): Likewise.
5100 * ree.c (combine_reaching_defs): Likewise.
5101 * reload.c (push_reload): Likewise.
5102 (find_reloads): Likewise.
5103 * reload1.c (elimination_effects): Likewise.
5104 (compute_reload_subreg_offset): Likewise.
5105 (choose_reload_regs): Likewise.
5106 * rtlanal.c (subreg_lsb_1): Likewise.
5107 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5108 (simplify_subreg): Likewise.
5109 * var-tracking.c (track_loc_p): Likewise.
5110 * emit-rtl.c (byte_lowpart_offset): Likewise.
5111 (paradoxical_subreg_p): Delete out-of-line definition.
5113 2017-08-22 Jeff Law <law@redhat.com>
5115 PR tree-optimization/81741
5116 PR tree-optimization/71947
5117 * tree-ssa-dom.c: Include tree-inline.h.
5118 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5119 equivalences if one is more expensive to compute than the other.
5120 * tree-ssa-scopedtables.h (class const_or_copies): Make
5121 record_const_or_copy_raw method private.
5122 (class avail_exprs_stack): New method simplify_binary_operation.
5123 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5124 avail_exprs_stack::simplify_binary_operation as needed.
5125 (avail_exprs_stack::simplify_binary_operation): New function.
5127 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5129 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5130 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5131 (DOT_SYMBOLS): Likewise.
5132 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5133 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5134 (RS6000_ABI_NAME): Likewise.
5135 (TARGET_CMODEL): Likewise.
5136 (TOC_SECTION_ASM_OP): Likewise.
5137 (SET_CMODEL): New macro.
5138 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5140 2017-08-22 Richard Biener <rguenther@suse.de>
5142 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5143 to allow for free-lang-data replacements similar to verify_type_variant.
5145 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5146 Alan Hayward <alan.hayward@arm.com>
5147 David Sherwood <david.sherwood@arm.com>
5149 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5150 VOIDmode for the LABEL_REF.
5152 2017-08-22 Richard Biener <rguenther@suse.de>
5154 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5156 2017-08-22 Richard Biener <rguenther@suse.de>
5158 * common.opt (feliminate-dwarf2-dups): Ignore.
5159 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5160 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5161 same_die_p_wrap, compute_section_prefix,
5162 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5163 (comdat_symbol_id, comdat_symbol_number): Likewise.
5164 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5166 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5167 (output_die): Mark unreachable path unreachable.
5168 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5169 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5170 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5171 (dwarf2out_early_finish): Likewise.
5173 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5175 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5177 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5180 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5181 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5182 (avr_attribute_table) <io, io_low, address>: Initialize
5183 .decl_required with true.
5185 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5187 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5188 undocumented debugging options.
5189 (-mvsx-scalar-double): Likewise.
5190 (-mallow-df-permute): Likewise.
5191 (-mvectorize-builtins): Likewise.
5192 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5193 (rs6000_builtin_vectorized_function): Likewise.
5194 (rs6000_builtin_md_vectorized_function): Likewise.
5195 (rs6000_opt_vars): Likewise.
5197 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5200 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5201 (*btrq_imm): Rename from *btrq.
5202 (*btcq_imm): Rename from *btcq.
5203 (btsc): New code attribute.
5204 (*<btsc><mode>): New insn pattern.
5205 (*btr<mode>): Ditto.
5206 (*<btsc><mode>_mask): New insn_and_split pattern.
5207 (*btr<mode>_mask): Ditto.
5209 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5210 Alan Hayward <alan.hayward@arm.com>
5211 David Sherwood <david.sherwood@arm.com>
5213 * function.c (pad_below): Simplify padding calculation.
5215 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5216 Alan Hayward <alan.hayward@arm.com>
5217 David Sherwood <david.sherwood@arm.com>
5219 * target.def (function_prologue): Remove frame size argument.
5220 (function_epilogue): Likewise.
5221 * doc/tm.texi: Regenerate.
5222 * final.c (final_start_function): Update call to function_prologue.
5223 (final_end_function): Update call to function_epilogue.
5224 (default_function_pro_epilogue): Remove frame size argument.
5225 * output.h (default_function_pro_epilogue): Likewise.
5226 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5227 (arm_output_function_prologue): Likewise.
5228 * config/frv/frv.c (frv_function_prologue): Likewise.
5229 (frv_function_epilogue): Likewise.
5230 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5231 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5232 (ia64_output_function_epilogue): Likewise.
5233 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5234 (m32r_output_function_epilogue): Likewise.
5235 * config/microblaze/microblaze.c (microblaze_function_prologue)
5236 (microblaze_function_epilogue): Likewise.
5237 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5238 (mips_output_function_epilogue): Likewise.
5239 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5240 (mmix_target_asm_function_epilogue): Likewise.
5241 * config/msp430/msp430.c (msp430_start_function): Likewise.
5242 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5243 (nds32_asm_function_epilogue): Likewise.
5244 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5245 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5246 (pa_output_function_epilogue): Likewise.
5247 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5248 (rs6000_output_function_epilogue): Likewise.
5249 * config/rl78/rl78.c (rl78_start_function): Likewise.
5250 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5251 (rs6000_output_function_epilogue): Likewise.
5252 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5253 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5254 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5255 (sparc_asm_function_epilogue): Likewise.
5257 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5259 * tree.h (type_has_mode_precision_p): New function.
5260 * convert.c (convert_to_integer_1): Use it.
5261 * expr.c (expand_expr_real_2): Likewise.
5262 (expand_expr_real_1): Likewise.
5263 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5264 * match.pd: Likewise.
5265 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5266 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5267 * tree-tailcall.c (process_assignment): Likewise.
5268 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5269 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5270 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5271 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5272 (vectorizable_assignment): Likewise.
5273 (vectorizable_shift): Likewise.
5274 (vectorizable_operation): Likewise.
5275 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5277 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5279 * match.pd: Add pow (C, x) simplification.
5281 2017-08-21 Richard Biener <rguenther@suse.de>
5283 PR tree-optimization/81900
5284 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5285 for blocks with abnormal predecessors.
5286 (compute_antic): Do not set visited flag prematurely.
5288 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5291 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5293 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5295 * stor-layout.h (vector_type_mode): Move to...
5296 * tree.h (vector_type_mode): ...here.
5297 * stor-layout.c (vector_type_mode): Move to...
5298 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5300 2017-08-21 Richard Biener <rguenther@suse.de>
5302 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5303 register_external_die hooks.
5304 (debug_false_tree_charstarstar_uhwistar): Declare.
5305 (debug_nothing_tree_charstar_uhwi): Likewise.
5306 * debug.c (do_nothing_debug_hooks): Adjust.
5307 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5308 (debug_nothing_tree_charstar_uhwi): Likewise.
5309 * dbxout.c (dbx_debug_hooks): Adjust.
5310 (xcoff_debug_hooks): Likewise.
5311 * sdbout.c (sdb_debug_hooks): Likewise.
5312 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5313 * dwarf2out.c (macinfo_label_base): New global.
5314 (dwarf2out_register_external_die): New function for the
5315 register_external_die hook.
5316 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5317 (dwarf2_debug_hooks): Use them.
5318 (dwarf2_lineno_debug_hooks): Adjust.
5319 (struct die_struct): Add with_offset flag.
5320 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5321 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5322 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5323 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5324 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5325 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5326 defining section names for the early LTO debug variants.
5327 (reset_indirect_string): New helper.
5328 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5329 (print_dw_val): Add support for offsetted symbol references.
5330 (get_ultimate_context): Split out from is_cxx.
5331 (is_cxx): Use get_ultimate_context.
5332 (is_fortran): Add decl overload.
5333 (compute_comp_unit_symbol): Split out worker from
5334 compute_section_prefix.
5335 (compute_section_prefix): Call compute_comp_unit_symbol and
5336 set comdat_type_p here.
5337 (output_die): Skip DIE symbol output for the LTO added one.
5338 Handle DIE symbol references with offset.
5339 (output_comp_unit): Guard section name mangling properly.
5340 For LTO debug sections emit a symbol at the section beginning
5341 which we use to refer to its DIEs.
5342 (add_abstract_origin_attribute): For DIEs registered via
5343 dwarf2out_register_external_die directly refer to the early
5344 DIE rather than indirectly through the shadow one we created.
5345 Remove obsolete call to dwarf2out_abstract_function for
5346 non-function/block origins.
5347 (gen_array_type_die): When generating early LTO debug do
5348 not emit DW_AT_string_length.
5349 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5350 late when in LTO. As suggested place a gcc_unreachable for
5351 the DECL_ABSTRACT_P case.
5352 (gen_subprogram_die): Avoid another specification DIE
5353 for early built declarations/definitions for the late LTO case.
5354 (gen_variable_die): Add type references for late duplicated VLA dies
5356 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5357 we have the abstract instance already.
5358 (process_scope_var): Adjust decl DIE contexts in LTO which
5359 first puts them in limbo.
5360 (gen_decl_die): Do not generate type DIEs late apart from
5361 types for VLAs or for decls we do not yet have a DIE. Do not
5362 call dwarf2out_abstract_function late.
5363 (dwarf2out_early_global_decl): Make sure to create DIEs
5364 for abstract instances of a decl first.
5365 (dwarf2out_late_global_decl): Adjust comment.
5366 (output_macinfo_op): With multiple macro sections use
5367 macinfo_label_base to distinguish labels.
5368 (output_macinfo): Likewise. Update macinfo_label_base.
5369 Pass in the line info label.
5370 (note_variable_value_in_expr): When generating LTO resolve
5371 all variable values here by generating DIEs as needed.
5372 (init_sections_and_labels): Add early LTO debug flag parameter
5373 and generate different sections and names if set. Add generation
5374 counter for the labels so we can have multiple of them.
5375 (reset_dies): Helper to allow DIEs to be output multiple times.
5376 (dwarf2out_finish): When outputting DIEs to the fat part of an
5377 LTO object first reset DIEs.
5378 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5379 (modified_type_die): Check for decl_ultimate_origin being self
5381 (gen_type_die_with_usage): Likewise.
5382 (gen_typedef_die): Allow decl_ultimate_origin being self.
5383 (set_decl_abstract_flags): Remove.
5384 (set_block_abstract_flags): Likewise.
5385 (dwarf2out_abstract_function): Treat the early generated DIEs
5386 as the abstract copy and only add DW_AT_inline and
5387 DW_AT_artificial here and call set_decl_origin_self.
5388 If the DIE has an abstract origin don't do anything.
5389 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5390 if we have none yet (Go fails to build one, PR78628).
5391 (variably_modified_type_p): Prevent endless recursion for Ada
5392 cyclic pointer types.
5393 * lto-streamer-in.c: Include debug.h.
5394 (dref_queue): New global.
5395 (lto_read_tree_1): Stream in DIE references.
5396 (lto_input_tree): Register DIE references.
5397 (input_function): Stream DECL_DEBUG_ARGS.
5398 * lto-streamer-out.c: Include debug.h.
5399 (lto_write_tree_1): Output DIE references.
5400 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5401 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5402 (output_function): Stream DECL_DEBUG_ARGS.
5403 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5404 Stream DECL_ABSTRACT_ORIGIN.
5405 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5406 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5407 DECL_CONTEXT for file-scope decls.
5408 * lto-streamer.h (struct dref_entry): Declare.
5409 (dref_queue): Likewise.
5410 * cfgexpand.c (pass_expand::execute): Do not call the
5411 outlining_inline_function hook here.
5412 * lto-wrapper.c (debug_obj): New global.
5413 (tool_cleanup): Unlink it if required.
5414 (debug_objcopy): New function.
5415 (run_gcc): Handle early debug sections in the IL files by
5416 extracting them to separate files, partially linkin them and
5417 feeding the result back as result to the linker.
5418 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5419 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5420 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5421 sections into a separate segment.
5422 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5424 (darwin_asm_dwarf_section): Likewise.
5425 (darwin_asm_output_dwarf_offset): Likewise.
5426 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5428 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5429 Alan Hayward <alan.hayward@arm.com>
5430 David Sherwood <david.sherwood@arm.com>
5432 * read-md.h (md_reader::record_potential_iterator_use): Replace
5433 pointer argument with an rtx and an index.
5434 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5435 (apply_mode_iterator): Likewise.
5436 (apply_code_iterator): Likewise.
5437 (apply_int_iterator): Likewise.
5438 (apply_subst_iterator): Likewise.
5439 (record_iterator_use): Likewise.
5440 (record_attribute_use): Likewise.
5441 (md_reader::record_potential_iterator_use): Likewise. Update calls
5442 to record_iterator_use and apply_iterator.
5443 (iterator_use): Replace ptr with x and index.
5444 (attribute_use): Likewise.
5445 (apply_attribute_uses): Update calls to apply_iterator.
5446 (apply_iterators): Likewise. Update initialization of iterator_use.
5447 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5448 and record_potential_iterator_use.
5449 (rtx_reader::read_rtx_operand): Likewise.
5451 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5452 Alan Hayward <alan.hayward@arm.com>
5453 David Sherwood <david.sherwood@arm.com>
5455 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5458 2017-08-21 Richard Biener <rguenther@suse.de>
5461 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5462 at struct end conservatively when comparing common bases.
5464 2017-08-21 Richard Biener <rguenther@suse.de>
5466 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5467 (mem_ref_in_stmt): Remove.
5468 (determine_max_movement): Use ref index to get at the reference.
5469 (invariantness_dom_walker::before_dom_children): Deal with
5470 lim data already initialized.
5471 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5473 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5475 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5476 (TARGET_ISA_ROUND): Ditto.
5477 (TARGET_ROUND): Ditto.
5478 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5479 * config/i386/i386.md: Ditto.
5480 * config/i386/sse.md: Ditto.
5481 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5482 with OPTION_MASK_ISA_SSE4_1.
5484 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5487 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5488 __builtin_ia32_lzcnt_u16.
5490 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5493 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5494 (rs6000_set_current_function): Rewrite function to use it.
5496 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5499 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5500 and TYPE_WARN_IF_NOT_ALIGN.
5501 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5502 (handle_warn_if_not_align): New.
5503 (place_union_field): Call handle_warn_if_not_align.
5504 (place_field): Call handle_warn_if_not_align.
5505 Copy TYPE_WARN_IF_NOT_ALIGN.
5506 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5507 (layout_type): Likewise.
5508 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5510 (tree_decl_common): Add warn_if_not_align.
5511 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5512 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5513 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5514 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5515 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5516 * doc/extend.texi: Document warn_if_not_aligned attribute.
5517 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5519 2017-08-17 Martin Liska <mliska@suse.cz>
5522 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5523 (get_data_dependence): Use it as pointer type.
5524 (distribute_loop): Likewise.
5526 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5528 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5529 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5530 (p8_vmrgow_<mode>_direct): New define_insn.
5531 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5532 handle endianness for vmrgew and vmrgow permute patterns.
5534 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5536 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5537 * config/rs6000/rs6000-cpus.def: Remove comment.
5538 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5539 (POWERPC_MASKS): Likewise.
5540 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5541 use of TARGET_VSX_TIMODE.
5542 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5543 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5544 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5545 (rs6000_option_override_internal): Remove dead code.
5546 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5547 (rs6000_legitimize_reload_address): Likewise.
5548 (rs6000_legitimate_address_p): Likewise.
5549 (rs6000_opt_masks): Delete "vsx-timode".
5550 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5551 from function comment.
5552 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5553 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5554 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5556 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5557 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5558 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5561 (define_peephole2): Likewise.
5563 2017-08-17 Martin Sebor <msebor@redhat.com>
5566 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5567 past the end of an array.
5568 (test_pp_format): Add test cases.
5570 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5572 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5573 missing ECF_NOTHROW flags.
5575 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5578 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5579 operands residing in integer registers.
5580 (*vsx_le_perm_load_<mode>): Likewise.
5581 (*vsx_le_perm_store_<mode>): Likewise.
5582 (define_peephole2): Add peepholes to optimize the above.
5584 2017-08-17 Marek Polacek <polacek@redhat.com>
5587 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5588 to mimic what shorten_compare did. Change the return type to bool.
5589 (fold_cond_expr_with_comparison): Update call to
5590 operand_equal_for_comparison_p.
5591 (fold_ternary_loc): Likewise.
5593 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5595 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5597 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5599 2017-08-17 Richard Biener <rguenther@suse.de>
5601 * tree-ssa-structalias.c (solve_graph): When propagating
5602 to successors update the graphs succ edges and avoid duplicate work.
5604 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5607 * config/i386/i386.c (ix86_option_override_internal): Save target
5608 specific options after ix86_stack_protector_guard_reg was changed.
5610 2017-08-17 Richard Biener <rguenther@suse.de>
5612 PR tree-optimization/81827
5613 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5614 (new_var_info): Initialize it conservatively.
5615 (get_call_vi): Mark register vars.
5616 (new_scalar_tmp_constraint_exp): Likewise.
5617 (handle_rhs_call): Likewise.
5618 (handle_const_call): Likewise.
5619 (create_function_info_for): Likewise.
5620 (solve_constraints): Sort varinfos to separate register from
5621 non-register vars to pack points-to solution bitmaps during
5624 2017-08-17 Marek Polacek <polacek@redhat.com>
5626 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5628 2017-08-17 Richard Biener <rguenther@suse.de>
5630 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5631 to true when overflow is undefined and we saturated the result.
5633 2017-08-17 Alan Modra <amodra@gmail.com>
5636 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5637 Don't use store multiple if only one reg needs saving.
5638 (interesting_frame_related_regno): New function.
5639 (rs6000_frame_related): Don't emit frame info for regs that
5641 (rs6000_emit_epilogue): Likewise.
5643 2017-08-16 Nathan Sidwell <nathan@acm.org>
5645 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5646 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5647 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5648 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5649 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5650 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5651 (verify_type): Adjust for TYPE_BINFO move.
5652 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5653 process TYPE_BINFO directly.
5654 (hash_tree): Likewise.
5655 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5657 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5660 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5662 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5664 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5667 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5668 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5669 Add ix86_binary_operator_ok to insn constraint.
5670 (*iordi_1_bts): Ditto.
5671 (*xordi_1_btc): Ditto.
5672 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5673 Update corresponding peephole2 pattern.
5677 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5679 PR tree-optimization/81832
5680 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5681 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5683 2017-08-16 Marek Polacek <polacek@redhat.com>
5686 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5687 regarding size_zero_node.
5689 2017-08-16 Martin Liska <mliska@suse.cz>
5692 * config.gcc: Respect previously set extra_objs in case
5695 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5697 PR tree-optimization/81835
5698 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5699 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5700 not depend on the phi.
5702 2017-08-16 Alan Modra <amodra@gmail.com>
5704 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5707 2017-08-16 Alan Modra <amodra@gmail.com>
5709 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5710 (save_reg_p): ..into this. Update all callers.
5711 (first_reg_to_save): Simplify.
5713 2017-08-16 Alan Modra <amodra@gmail.com>
5715 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5718 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5722 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5723 constant count if that count is less than 32.
5725 2017-08-15 Nathan Sidwell <nathan@acm.org>
5727 * gcc.c (execute): Emit friendlier message if inferior is killed
5728 by an external cause.
5730 2017-08-15 Richard Biener <rguenther@suse.de>
5732 PR tree-optimization/81790
5733 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5734 CONSTRUCTORs from simplifying and VN.
5736 2017-08-14 Martin Sebor <msebor@redhat.com>
5738 * builtin-attrs.def: Add comments.
5740 2017-08-14 Martin Sebor <msebor@redhat.com>
5743 * doc/extend.texi (attribute nonstring): Document new attribute.
5745 2017-08-14 Martin Sebor <msebor@redhat.com>
5748 * tree-diagnostic.c (default_tree_printer): Handle %G.
5749 * gimple-pretty-print.h (percent_G_format): Declare new function.
5750 * gimple-pretty-print.c (percent_G_format): Define.
5751 * tree-pretty-print.c (percent_K_format): Add argument.
5753 2017-08-14 Martin Sebor <msebor@redhat.com>
5755 PR translation/79998
5756 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5757 Remove a stray space.
5759 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5762 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5763 (*iordi_1_bts): Ditto.
5764 (*xordi_1_btc): Ditto.
5766 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5769 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5770 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5772 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5773 quoted strings, and make more translator-friendly.
5774 (darwin_rs6000_override_options): Likewise.
5775 (rs6000_option_override_internal): Likewise.
5776 (rs6000_return_in_memory): Fix overlong line.
5777 (init_cmulative_args): Use quoted strings, and make more
5778 translator-friendly.
5779 (rs6000_pass_by_reference): Fix overlong line.
5780 (def_builtin): Use quoted strings.
5781 (altivec_expand_predicate_builtin): Use quoted strings, and make
5782 more translator-friendly.
5783 (htm_expand_builtin): Use quoted strings.
5784 (cpu_expand_builtin): Use quoted strings, and make more
5785 translator-friendly.
5786 (altivec_expand_builtin): Likewise.
5787 (paired_expand_predicate_builtin): Likewise.
5788 (rs6000_invalid_builtin): Likewise.
5789 (builtin_function_type): Use quoted strings.
5790 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5791 more translator-friendly.
5792 (rs6000_trampoline_init): Likewise.
5793 (rs6000_handle_altivec_attribute): Likewise.
5794 (rs6000_inner_target_options): Use quoted strings.
5795 (rs6000_disable_incompatible_switches): Likewise.
5796 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5797 strings, and make more translator-friendly.
5798 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5800 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5802 PR tree-optimization/81799
5803 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5804 cond_expr to simple gimple operand.
5806 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5809 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5811 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5815 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5816 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5817 TARGET_GASISR_PROLOGUES.
5818 * config/avr/avr.c (avr_option_override): Same.
5819 (avr_pass_pre_proep::execute): Same.
5821 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5824 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5825 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5827 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5829 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5830 define_insn_and_split. Split to a memory load from 0 in
5831 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5832 using PTR mode iterator.
5833 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5834 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5835 (*add_tp_<mode>): Redefine as define_insn_and_split.
5836 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5837 address space. Merge with *add_tp_x32 using PTR mode iterator.
5838 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5839 Split to an add with a memory load from 0 in
5840 DEFAULT_TLS_SEG_REG address space.
5842 2017-08-12 Andrew Pinski <apinski@cavium.com>
5844 * config/aarch64/aarch64-option-extensions.def (rdma):
5845 Fix feature string to what Linux prints out in /proc/cpuinfo.
5847 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5850 * dwarf2out.c (modified_type_die): For C typedef types that have
5851 an ultimate origin, process the ultimate origin instead of the
5853 (gen_typedef_die): Assert that input DECLs have no ultimate
5855 (gen_type_die_with_usage): For typedef variants that have an
5856 ultimate origin, just call gen_decl_die on the original DECL.
5857 (process_scope_var): Avoid creating DIEs for local typedefs and
5858 concrete static variables.
5860 2017-08-12 Alan Modra <amodra@gmail.com>
5864 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5865 match gnu-user.h startfile.
5866 (ENDFILE_LINUX_SPEC): Similarly.
5868 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5871 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5873 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5875 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5876 * config/aarch64/aarch64.md (mov<mode>): Change.
5877 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5878 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5879 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5881 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5883 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5884 for storage order barriers.
5886 2017-08-11 Martin Liska <mliska@suse.cz>
5889 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5890 variables of void type.
5892 2017-08-11 Martin Liska <mliska@suse.cz>
5894 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5895 TARGET_SUPPORTS_ALIASES.
5896 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5897 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5898 (optimize_weakref): Likewise.
5899 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5900 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5901 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5903 2017-08-11 Martin Liska <mliska@suse.cz>
5906 * config/i386/i386.c (make_resolver_func): Do complete
5907 refactoring of the function.
5909 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5912 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5913 * config/i386/i386.c (ix86_stack_protect_guard): Use
5914 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5915 * doc/invoke.texi (x86 Options): Document
5916 -mstack-protector-guard-symbol= option.
5918 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5920 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5921 * config/i386/i386.c (ix86_split_stack_guard): New function.
5922 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5923 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5924 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5925 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5926 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5927 (split_stack_space_check): Call ix86_split_stack_guard.
5929 2017-08-10 Martin Sebor <msebor@redhat.com>
5931 * print-tree.c (print_node): Print location using the established
5933 Replace spaces with colons.
5934 (debug_raw, debug): Ditto.
5936 2017-08-10 Martin Sebor <msebor@redhat.com>
5939 * pretty-print.c (pp_format): Correct the handling of %s precision.
5941 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5944 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5946 (ix86_finalize_stack_frame_flags): This. Also clear
5947 frame_pointer_needed if -fno-omit-frame-pointer is used without
5949 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5950 with ix86_finalize_stack_frame_flags.
5951 (ix86_expand_epilogue): Likewise.
5952 (ix86_expand_split_stack_prologue): Likewise.
5953 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5955 2017-08-10 Martin Liska <mliska@suse.cz>
5958 * c-attribs.c (handle_target_attribute):
5959 Report warning for an empty string argument of target attribute.
5961 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5964 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5966 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5967 or DECL_NONLOCAL labels.
5968 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5969 or DECL_NONLOCAL labels here.
5971 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5973 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5974 to indicate when early gimple folding has been disabled.
5975 (rs6000_gimple_fold_builtin): Add debug content.
5976 (rs6000_invalid_builtin): Fix whitespace.
5977 (rs6000_expand_builtin): Fix whitespace.
5978 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5980 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5983 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5984 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5986 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5988 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5989 (qdf24xx): Likewise.
5990 * config/aarch64/aarch64.md: Include falkor.md.
5991 * config/aarch64/falkor.md: New.
5993 2017-08-09 Marek Polacek <polacek@redhat.com>
5996 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5997 * diagnostic.c (emit_diagnostic): Add a comment.
5998 (emit_diagnostic_valist): New function.
6000 2017-08-09 Marek Polacek <polacek@redhat.com>
6003 * input.c (make_location): New overload.
6004 * input.h (make_location): Declare.
6006 2017-08-08 Alan Modra <amodra@gmail.com>
6007 H.J. Lu <hongjiu.lu@intel.com>
6010 * gcc.c (NO_PIE_SPEC): Delete.
6011 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6013 (LINK_PIE_SPEC): ..to here.
6014 (LINK_COMMAND_SPEC): Support -no-pie.
6015 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6016 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6018 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6019 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6020 (ENDFILE_CRTEND_SPEC): Similarly.
6022 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6025 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6026 (mstack-protector-guard-offset=): Ditto.
6027 * config/i386/i386.c (ix86_option_override): Handle
6028 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6030 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6031 ix86_stack_protect_guard_offset variables.
6032 (TARGET_STACK_PROTECT_GUARD): Always define.
6033 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6034 and -mstack-protector-guard-offset= options.
6036 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6038 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6039 boundary case for the last candidate.
6041 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6043 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6044 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6046 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6049 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6050 * config/aarch64/aarch64-builtins.c
6051 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6052 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6053 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6055 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6056 Andrew Pinski <pinskia@gmail.com>
6059 * internal-fn.def (XORSIGN): New.
6060 * optabs.def (xorsign_optab): New.
6061 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6062 (convert_expand_mult_copysign): New.
6063 (pass_optimize_widening_mul::execute): Call
6064 convert_expand_mult_copysign.
6066 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6068 PR tree-optimization/81354
6069 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6070 Insert on edges rather than explicitly creating landing pads.
6071 (analyze_candidates_and_replace): Commit edge inserts.
6073 2017-08-08 Richard Biener <rguenther@suse.de>
6076 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6077 (expand_simple_operations): Also look through ADDR_EXPRs with
6078 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6080 2017-08-08 Richard Biener <rguenther@suse.de>
6082 PR tree-optimization/81723
6083 * tree-vect-slp.c (struct bst_traits): New hash traits.
6084 (bst_fail): New global.
6085 (vect_build_slp_tree_2): New worker, split out from ...
6086 (vect_build_slp_tree): ... this now wrapping it with using
6087 bst_fail set to cache SLP tree build fails. Properly handle
6089 (vect_analyze_slp_instance): Allocate and free bst_fail.
6091 2017-08-08 Martin Liska <mliska@suse.cz>
6094 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6095 LABEL_DECLs that can be from a different function.
6097 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6099 PR tree-optimization/81744
6100 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6101 loop's number of iterations.
6103 2017-08-08 Martin Liska <mliska@suse.cz>
6105 * asan.c: Include header files.
6106 * attribs.c (build_decl_attribute_variant): New function moved
6108 (build_type_attribute_qual_variant): Likewise.
6109 (cmp_attrib_identifiers): Likewise.
6110 (simple_cst_list_equal): Likewise.
6111 (omp_declare_simd_clauses_equal): Likewise.
6112 (attribute_value_equal): Likewise.
6113 (comp_type_attributes): Likewise.
6114 (build_type_attribute_variant): Likewise.
6115 (lookup_ident_attribute): Likewise.
6116 (remove_attribute): Likewise.
6117 (merge_attributes): Likewise.
6118 (merge_type_attributes): Likewise.
6119 (merge_decl_attributes): Likewise.
6120 (merge_dllimport_decl_attributes): Likewise.
6121 (handle_dll_attribute): Likewise.
6122 (attribute_list_equal): Likewise.
6123 (attribute_list_contained): Likewise.
6124 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6125 (lookup_attribute_by_prefix): Likewise.
6126 * bb-reorder.c: Include header files.
6127 * builtins.c: Likewise.
6128 * calls.c: Likewise.
6129 * cfgexpand.c: Likewise.
6130 * cgraph.c: Likewise.
6131 * cgraphunit.c: Likewise.
6132 * convert.c: Likewise.
6133 * dwarf2out.c: Likewise.
6134 * final.c: Likewise.
6135 * fold-const.c: Likewise.
6136 * function.c: Likewise.
6137 * gimple-expr.c: Likewise.
6138 * gimple-fold.c: Likewise.
6139 * gimple-pretty-print.c: Likewise.
6140 * gimple.c: Likewise.
6141 * gimplify.c: Likewise.
6142 * hsa-common.c: Likewise.
6143 * hsa-gen.c: Likewise.
6144 * internal-fn.c: Likewise.
6145 * ipa-chkp.c: Likewise.
6146 * ipa-cp.c: Likewise.
6147 * ipa-devirt.c: Likewise.
6148 * ipa-fnsummary.c: Likewise.
6149 * ipa-inline.c: Likewise.
6150 * ipa-visibility.c: Likewise.
6152 * lto-cgraph.c: Likewise.
6153 * omp-expand.c: Likewise.
6154 * omp-general.c: Likewise.
6155 * omp-low.c: Likewise.
6156 * omp-offload.c: Likewise.
6157 * omp-simd-clone.c: Likewise.
6158 * opts-global.c: Likewise.
6159 * passes.c: Likewise.
6160 * predict.c: Likewise.
6161 * sancov.c: Likewise.
6162 * sanopt.c: Likewise.
6163 * symtab.c: Likewise.
6164 * toplev.c: Likewise.
6165 * trans-mem.c: Likewise.
6166 * tree-chkp.c: Likewise.
6167 * tree-eh.c: Likewise.
6168 * tree-into-ssa.c: Likewise.
6169 * tree-object-size.c: Likewise.
6170 * tree-parloops.c: Likewise.
6171 * tree-profile.c: Likewise.
6172 * tree-ssa-ccp.c: Likewise.
6173 * tree-ssa-live.c: Likewise.
6174 * tree-ssa-loop.c: Likewise.
6175 * tree-ssa-sccvn.c: Likewise.
6176 * tree-ssa-structalias.c: Likewise.
6177 * tree-ssa.c: Likewise.
6178 * tree-streamer-in.c: Likewise.
6179 * tree-vectorizer.c: Likewise.
6180 * tree-vrp.c: Likewise.
6182 * ubsan.c: Likewise.
6183 * varasm.c: Likewise.
6184 * varpool.c: Likewise.
6185 * tree.c: Remove functions moved to attribs.[ch].
6187 * config/aarch64/aarch64.c: Add attrs.h header file.
6188 * config/alpha/alpha.c: Likewise.
6189 * config/arc/arc.c: Likewise.
6190 * config/arm/arm.c: Likewise.
6191 * config/avr/avr.c: Likewise.
6192 * config/bfin/bfin.c: Likewise.
6193 * config/c6x/c6x.c: Likewise.
6194 * config/cr16/cr16.c: Likewise.
6195 * config/cris/cris.c: Likewise.
6196 * config/darwin.c: Likewise.
6197 * config/epiphany/epiphany.c: Likewise.
6198 * config/fr30/fr30.c: Likewise.
6199 * config/frv/frv.c: Likewise.
6200 * config/ft32/ft32.c: Likewise.
6201 * config/h8300/h8300.c: Likewise.
6202 * config/i386/winnt.c: Likewise.
6203 * config/ia64/ia64.c: Likewise.
6204 * config/iq2000/iq2000.c: Likewise.
6205 * config/lm32/lm32.c: Likewise.
6206 * config/m32c/m32c.c: Likewise.
6207 * config/m32r/m32r.c: Likewise.
6208 * config/m68k/m68k.c: Likewise.
6209 * config/mcore/mcore.c: Likewise.
6210 * config/microblaze/microblaze.c: Likewise.
6211 * config/mips/mips.c: Likewise.
6212 * config/mmix/mmix.c: Likewise.
6213 * config/mn10300/mn10300.c: Likewise.
6214 * config/moxie/moxie.c: Likewise.
6215 * config/msp430/msp430.c: Likewise.
6216 * config/nds32/nds32-isr.c: Likewise.
6217 * config/nds32/nds32.c: Likewise.
6218 * config/nios2/nios2.c: Likewise.
6219 * config/nvptx/nvptx.c: Likewise.
6220 * config/pa/pa.c: Likewise.
6221 * config/pdp11/pdp11.c: Likewise.
6222 * config/powerpcspe/powerpcspe.c: Likewise.
6223 * config/riscv/riscv.c: Likewise.
6224 * config/rl78/rl78.c: Likewise.
6225 * config/rx/rx.c: Likewise.
6226 * config/s390/s390.c: Likewise.
6227 * config/sh/sh.c: Likewise.
6228 * config/sol2.c: Likewise.
6229 * config/sparc/sparc.c: Likewise.
6230 * config/spu/spu.c: Likewise.
6231 * config/stormy16/stormy16.c: Likewise.
6232 * config/tilegx/tilegx.c: Likewise.
6233 * config/tilepro/tilepro.c: Likewise.
6234 * config/v850/v850.c: Likewise.
6235 * config/vax/vax.c: Likewise.
6236 * config/visium/visium.c: Likewise.
6237 * config/xtensa/xtensa.c: Likewise.
6239 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6242 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6243 constraints since the -mupper-regs-* switches have been
6245 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6246 into a vector from a double word element that was extracted from
6247 another vector, and eliminate extra XXPERMDI instructions.
6248 (vsx_concat_<mode>_2): Likewise.
6249 (vsx_concat_<mode>_3): Likewise.
6250 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6251 concat to allow optimizing inserts from previous extracts.
6253 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6255 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6256 memory reference to a SSP offset in TLS address space.
6257 (ix86_print_operand) <case '@'>: Remove.
6258 (ix86_print_operand_punct_valid_p): Remove '@' code.
6259 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6261 (stack_tls_protect_set_<mode>): Remove.
6262 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6263 (stack_tls_protect_test_<mode>): Remove.
6264 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6266 2017-08-07 Olivier Hainque <hainque@adacore.com>
6269 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6271 2017-08-07 Douglas Rupp <rupp@adacore.com>
6273 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6274 variable was referenced as multidir in command.
6276 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6279 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6282 2017-08-07 Martin Liska <mliska@suse.cz>
6284 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6285 * config/rl78/rl78.c: Add include of attribs.h.
6286 * config/sh/sh.c: Likewise.
6287 * config/v850/v850.c: Likewise.
6289 2017-08-07 Tom de Vries <tom@codesourcery.com>
6292 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6294 2017-08-07 Martin Liska <mliska@suse.cz>
6296 * config/mips/mips.c: Include attribs.h.
6298 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6301 * doc/invoke.texi: Document change in behvaior for -Ofast for
6304 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6306 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6308 (aarch64_pop_regs): Likewise.
6309 (aarch64_gen_load_pair): Likewise.
6310 (aarch64_save_callee_saves): Likewise.
6311 (aarch64_restore_callee_saves): Likewise.
6313 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6315 * config/i386/i386.c: Revert the last change.
6317 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6320 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6322 (ix86_finalize_stack_frame_flags): This. Also clear
6323 frame_pointer_needed if -fno-omit-frame-pointer is used without
6325 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6326 with ix86_finalize_stack_frame_flags.
6327 (ix86_expand_epilogue): Likewise.
6328 (ix86_expand_split_stack_prologue): Likewise.
6330 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6333 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6334 to P_AES for Westmere.
6336 2017-08-07 Jonathan Yong <10walls@gmail.com>
6338 * config/i386/mingw.opt (fset-stack-executable): Removed.
6339 * config/i386/cygming.opt (fset-stack-executable): Moved
6341 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6343 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6345 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6347 2017-08-07 Marek Polacek <polacek@redhat.com>
6350 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6352 2017-08-07 Martin Liska <mliska@suse.cz>
6354 * attribs.h (canonicalize_attr_name): New function.
6355 (cmp_attribs): Move from c-format.c and adjusted.
6356 (is_attribute_p): Moved from tree.h.
6357 * tree-inline.c: Add new includes.
6358 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6359 (private_is_attribute_p): Remove.
6360 (private_lookup_attribute): Likewise.
6361 (private_lookup_attribute_by_prefix): Simplify.
6362 (remove_attribute): Use is_attribute_p.
6363 * tree.h: Remove removed declarations.
6365 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6368 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6369 instead of computing it in the function. Formatting fix.
6370 (expand_case): Don't rely on default_edge being the first edge,
6371 clear it if removing it, pass default_edge to
6372 emit_case_dispatch_table.
6373 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6376 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6378 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6379 insn in the function, emit NOP after the insn.
6381 2017-08-06 Tom de Vries <tom@codesourcery.com>
6383 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6386 2017-08-06 Tom de Vries <tom@codesourcery.com>
6388 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6391 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6393 PR tree-optimization/57371
6394 * match.pd: New pattern.
6396 2017-08-04 Marek Polacek <polacek@redhat.com>
6399 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6400 perform the computation in offset_int.
6402 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6404 PR tree-optimization/81136
6405 * tree-vectorizer.h: Include tree-hash-traits.h.
6406 (vec_base_alignments): New typedef.
6407 (vec_info): Add a base_alignments field.
6408 (vect_record_base_alignments): Declare.
6409 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6411 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6412 (create_data_ref): Add an is_conditional_in_stmt argument.
6413 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6414 the is_conditional_in_stmt field.
6415 (data_ref_loc): Add an is_conditional_in_stmt field.
6416 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6417 (find_data_references_in_stmt): Update call to create_data_ref.
6418 (graphite_find_data_references_in_stmt): Likewise.
6419 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6420 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6421 (vect_record_base_alignment): New function.
6422 (vect_record_base_alignments): Likewise.
6423 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6424 for nested statements even if we fail to compute a misalignment.
6425 Use pooled base alignments for unconditional references.
6426 (vect_find_same_alignment_drs): Compare base addresses instead
6428 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6429 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6431 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6433 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6434 Add an explicit name for the enum. Use auto_vec for slp_instances
6436 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6438 (_bb_vec_info): Add a constructor and destructor.
6439 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6440 (destroy_loop_vec_info): Delete.
6441 (vect_destroy_datarefs): Likewise.
6442 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6443 (vec_info::vec_info): New function.
6444 (vec_info::~vec_info): Likewise.
6445 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6446 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6447 destroy_loop_vec_info.
6448 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6449 (_loop_vec_info::_loop_vec_info): ...this.
6450 (destroy_loop_vec_info): Replace with...
6451 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6452 the stmt_vec_infos. Leave handling of vec_info information to its
6453 destructor. Remove explicit vector releases.
6454 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6455 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6456 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6457 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6458 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6459 (destroy_bb_vec_info): Replace with...
6460 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6461 information to its destructor.
6462 (vect_slp_analyze_bb_1): Use new and delete instead of
6463 new_bb_vec_info and destroy_bb_vec_info.
6464 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6467 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6469 * tree-data-ref.h (subscript): Add access_fn field.
6470 (data_dependence_relation): Add could_be_independent_p.
6471 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6472 (same_access_functions): Move to tree-data-ref.c.
6473 * tree-data-ref.c (ref_contains_union_access_p): New function.
6474 (access_fn_component_p): Likewise.
6475 (access_fn_components_comparable_p): Likewise.
6476 (dr_analyze_indices): Add a reference to access_fn_component_p.
6477 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6479 (constant_access_functions): Likewise.
6480 (add_other_self_distances): Likewise.
6481 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6482 (initialize_data_dependence_relation): Use XCNEW and remove
6483 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6484 of access functions that have the same type. Allow the
6485 subsequence to end with different bases in some circumstances.
6486 Record the chosen access functions in SUB_ACCESS_FN.
6487 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6488 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6489 (subscript_dependence_tester_1): Likewise dra and drb.
6490 (build_classic_dist_vector): Update calls accordingly.
6491 (subscript_dependence_tester): Likewise.
6492 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6493 DDR_COULD_BE_INDEPENDENT_P.
6494 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6495 comp_alias_ddrs instead of may_alias_ddrs.
6496 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6498 (vect_analyze_data_ref_dependence): Use it if
6499 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6500 distance vectors if that fails.
6501 (dependence_distance_ge_vf): New function.
6502 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6503 LOOP_VINFO_MAY_ALIAS_DDRS.
6505 2017-08-04 Richard Biener <rguenther@suse.de>
6508 * fold-const.c (fold_binary_loc): Properly restrict
6509 minus_var0 && minus_var1 case when associating undefined overflow
6512 2017-08-04 Tom de Vries <tom@codesourcery.com>
6514 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6516 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6518 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6519 Don't start diagnostic messages with a capital letter.
6520 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6522 (rs6000_invalid_builtin): Likewise.
6523 (rs6000_trampoline_init): Likewise.
6525 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6528 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6529 after setting changeable df flags.
6531 2017-08-03 Richard Biener <rguenther@suse.de>
6533 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6534 up if the use is in USE - X.
6536 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6538 * toplev.c (dumpfile.h): New include.
6539 (internal_error_reentered): New static function. Use it...
6540 (internal_error_function): ...here to handle reentered internal_error.
6542 2017-08-03 Richard Biener <rguenther@suse.de>
6545 * fold-const.c (split_tree): Add minus_var and minus_con
6546 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6547 here but always use minus_*.
6548 (associate_trees): Assert we never associate with MINUS_EXPR
6549 and NULL first operand. Do not recurse for PLUS_EXPR operands
6550 when associating as MINUS_EXPR either.
6551 (fold_binary_loc): Track minus_var and minus_con.
6553 2017-08-03 Tom de Vries <tom@codesourcery.com>
6556 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6557 ACCEL_COMPILER, apply finish_options on
6558 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6560 2017-08-03 Tom de Vries <tom@codesourcery.com>
6563 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6564 function_entry_patch_area_size > 0.
6566 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6569 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6570 instead of 10??LU, perform unit multiplication in wide_int,
6571 don't change alloc_object_size_limit if the limit is larger
6574 PR tree-optimization/81655
6575 PR tree-optimization/81588
6576 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6577 the case when ranges[i].low and high are 1 for unsigned type with
6581 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6582 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6584 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6586 * tree-vrp.h: Add include guard.
6588 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6591 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6592 (ud2): New insn pattern.
6593 * config/i386/i386.c (ix86_expand_epilogue):
6594 For naked functions, generate ud2 instead of trap insn.
6596 2017-08-02 Marek Polacek <polacek@redhat.com>
6599 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6601 2017-08-02 Tom de Vries <tom@codesourcery.com>
6602 Cesar Philippidis <cesar@codesourcery.com>
6604 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6605 Add missing edge probabilities.
6607 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6609 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6612 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6615 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6616 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6617 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6618 prologue_seq sequences - if any.
6620 2017-08-02 Richard Biener <rguenther@suse.de>
6622 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6623 via vectors if supported, integer extracts via punning if supported
6624 or otherwise vector extracts.
6626 2017-08-02 Richard Biener <rguenther@suse.de>
6628 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6630 (bitmap_insert_into_set): ... this.
6632 2017-08-02 Richard Biener <rguenther@suse.de>
6634 PR tree-optimization/81633
6636 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6638 PR tree-optimization/71752
6639 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6641 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6643 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6644 (machine_function::call_ms2sysv_pad_out): Remove field.
6645 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6646 (ix86_compute_frame_layout): Likewise.
6648 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6651 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6652 attribute with interrupt attribute.
6654 2017-08-01 Andrew Pinski <apinski@cavium.com>
6656 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6657 BIT_INSERT_EXPR's operand 1
6658 to see if the types precision matches.
6660 2017-08-01 Martin Liska <mliska@suse.cz>
6663 * builtins.c (expand_builtin_memcpy_args): Remove.
6664 (expand_builtin_memcpy): Call newly added function
6665 expand_builtin_memory_copy_args.
6666 (expand_builtin_memcpy_with_bounds): Likewise.
6667 (expand_builtin_mempcpy): Remove last argument.
6668 (expand_builtin_mempcpy_with_bounds): Likewise.
6669 (expand_builtin_memory_copy_args): New function created from
6670 expand_builtin_mempcpy_args with small modifications.
6671 (expand_builtin_mempcpy_args): Remove.
6672 (expand_builtin_stpcpy): Remove unused argument.
6673 (expand_builtin): Likewise.
6674 (expand_builtin_with_bounds): Likewise.
6676 2017-08-01 Martin Liska <mliska@suse.cz>
6679 Make mempcpy more optimal (PR middle-end/70140).
6681 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6684 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6685 __builtin_vec_cmpne verify both arguments are compatible vectors
6686 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6687 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6688 move computation of aligned to after checking the argument types.
6692 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6693 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6696 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6697 Doug Rupp <rupp@adacore.com>
6698 Olivier Hainque <hainque@adacore.com>
6700 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6701 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6703 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6704 for TARGET_OS_CPP_BUILTIN.
6705 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6706 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6708 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6709 passing required abi options to the assembler for EABI configurations.
6710 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6711 of .text.hot and .text.unlikely sections for kernel modules when
6712 using ARM style exceptions.
6713 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6714 options. Add EXTRA_CC1_SPEC.
6715 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6717 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6719 (ARM_TARGET2_DWARF_FORMAT): Define.
6720 * config/arm/t-vxworks: Adjust multilib control to removal of the
6721 Diab command line options.
6723 2017-08-01 Martin Liska <mliska@suse.cz>
6725 PR gcov-profile/81561
6726 * gcov.c (unblock): Make unblocking safe as we need to preserve
6727 index correspondence of blocks and block_lists.
6729 2017-08-01 Richard Biener <rguenther@suse.de>
6731 PR tree-optimization/81181
6732 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6733 (compute_antic): ... end of iteration here.
6735 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6737 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6738 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6739 (ftree-slp-vectorize): Likewise.
6740 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6741 can no longer be set independent of flag_tree_loop_vectorize.
6742 * omp-general.c (emp_max_vf): Likewise.
6743 * opts.c (enable_fdo_optimizations): Remove references to
6744 flag_tree_vectorize, these are now implicit.
6745 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6746 and leave it for the options machinery.
6748 2017-08-01 Martin Liska <mliska@suse.cz>
6751 * builtins.c (expand_builtin_memcpy_args): Remove.
6752 (expand_builtin_memcpy): Call newly added function
6753 expand_builtin_memory_copy_args.
6754 (expand_builtin_memcpy_with_bounds): Likewise.
6755 (expand_builtin_mempcpy): Remove last argument.
6756 (expand_builtin_mempcpy_with_bounds): Likewise.
6757 (expand_builtin_memory_copy_args): New function created from
6758 expand_builtin_mempcpy_args with small modifications.
6759 (expand_builtin_mempcpy_args): Remove.
6760 (expand_builtin_stpcpy): Remove unused argument.
6761 (expand_builtin): Likewise.
6762 (expand_builtin_with_bounds): Likewise.
6764 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6767 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6768 print "ds:" only for immediates in generic address space.
6770 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6773 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6774 (ix86_function_ok_for_sibcall): Return false for naked functions.
6776 2017-08-01 Richard Biener <rguenther@suse.de>
6778 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6779 (compute_antic): Seed worklist with exit block predecessors.
6780 * cfganal.c (dfs_find_deadend): For a cycle return the source
6781 of the edge closing it.
6783 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6785 * config/aarch64/aarch64.c
6786 (aarch64_can_const_movi_rtx_p): Move 0 check.
6788 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6790 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6791 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6793 * match.pd: Ditto in address comparison pattern.
6795 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6797 PR tree-optimization/81627
6798 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6799 closed ssa form for store-store chain.
6801 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6803 PR tree-optimization/81620
6804 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6805 for store-store chain.
6807 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6809 PR tree-optimization/81588
6810 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6811 ranges[i].in_p, invert comparison code ccode. For >/>=,
6812 swap rhs1 and rhs2 and comparison code unconditionally,
6813 for </<= don't do that. Don't swap rhs1/rhs2 again if
6814 ranges[i].in_p, instead invert comparison code ccode if
6815 opcode or oe->rank is BIT_IOR_EXPR.
6818 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6819 a direct optab to conversion optab.
6820 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6821 with GET_MODE_INNER as last argument instead of optab_handler.
6822 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6823 vector extraction if possible and optab is available.
6824 * expr.c (store_constructor): Use convert_optab_handler instead
6825 of optab_handler. Use vector initialization from smaller
6826 vectors if possible and optab is available.
6827 * tree-vect-stmts.c (vectorizable_load): Likewise.
6828 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6830 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6831 of vec_init from half-sized vectors with the same element mode.
6832 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6833 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6834 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6835 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6836 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6837 after mode in gen_vec_extract* calls.
6838 (vec_extract<mode>): Renamed to ...
6839 (vec_extract<mode><ssescalarmodelower>): ... this.
6840 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6841 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6842 element mode after mode in gen_vec_init* calls.
6843 (VEC_INIT_HALF_MODE): New mode iterator.
6844 (vec_init<mode>): Renamed to ...
6845 (vec_init<mode><ssescalarmodelower>): ... this.
6846 (vec_init<mode><ssehalfvecmodelower>): New expander.
6847 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6848 (vec_extractv2sfsf): ... this.
6849 (vec_initv2sf): Renamed to ...
6850 (vec_initv2sfsf): ... this.
6851 (vec_extractv2si): Renamed to ...
6852 (vec_extractv2sisi): ... this.
6853 (vec_initv2si): Renamed to ...
6854 (vec_initv2sisi): ... this.
6855 (vec_extractv4hi): Renamed to ...
6856 (vec_extractv4hihi): ... this.
6857 (vec_initv4hi): Renamed to ...
6858 (vec_initv4hihi): ... this.
6859 (vec_extractv8qi): Renamed to ...
6860 (vec_extractv8qiqi): ... this.
6861 (vec_initv8qi): Renamed to ...
6862 (vec_initv8qiqi): ... this.
6863 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6864 (vec_init<mode>): Renamed to ...
6865 (vec_init<mode><VEC_base_l>): ... this.
6866 (vec_extract<mode>): Renamed to ...
6867 (vec_extract<mode><VEC_base_l>): ... this.
6868 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6869 (vec_initv2sfsf): ... this.
6870 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6871 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6872 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6873 element mode after mode in gen_vec_init* calls.
6874 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6875 (vec_init<mode><Vel>): ... this.
6876 (vec_extract<mode>): Renamed to ...
6877 (vec_extract<mode><Vel>): ... this.
6878 * config/aarch64/iterators.md (Vel): New mode attribute.
6879 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6880 Add element mode after mode in gen_vec_extract* calls.
6881 * config/s390/vector.md (non_vec_l): New mode attribute.
6882 (vec_extract<mode>): Renamed to ...
6883 (vec_extract<mode><non_vec_l>): ... this.
6884 (vec_init<mode>): Renamed to ...
6885 (vec_init<mode><non_vec_l>): ... this.
6886 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6887 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6889 * config/arm/iterators.md (V_elem_l): New mode attribute.
6890 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6891 (vec_extract<mode><V_elem_l>): ... this.
6892 (vec_extractv2di): Renamed to ...
6893 (vec_extractv2didi): ... this.
6894 (vec_init<mode>): Renamed to ...
6895 (vec_init<mode><V_elem_l>): ... this.
6896 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6897 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6898 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6899 Add element mode after gen_vec_extract* calls.
6900 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6901 (vec_init<mode><unitmode>): ... this.
6902 (vec_extract<mode>): Renamed to ...
6903 (vec_extract<mode><unitmode>): ... this.
6904 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6905 (vec_init<mode><unitmode>): ... this.
6906 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6907 (vec_initv2sfsf): ... this.
6908 (vec_extractv2sf): Renamed to ...
6909 (vec_extractv2sfsf): ... this.
6910 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6911 Add element mode after gen_vec_extract* calls.
6912 * config/mips/mips.md (unitmode): New mode iterator.
6913 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6914 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6915 * config/spu/spu.md (inner_l): New mode attribute.
6916 (vec_init<mode>): Renamed to ...
6917 (vec_init<mode><inner_l>): ... this.
6918 (vec_extract<mode>): Renamed to ...
6919 (vec_extract<mode><inner_l>): ... this.
6920 * config/sparc/sparc.md (veltmode): New mode iterator.
6921 (vec_init<VMALL:mode>): Renamed to ...
6922 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6923 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6924 (vec_initv2sisi): ... this.
6925 (vec_initv2sf): Renamed to ...
6926 (vec_initv2sfsf): ... this.
6927 (vec_extractv2sf): Renamed to ...
6928 (vec_extractv2sfsf): ... this.
6929 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6930 (vec_init<mode>): Renamed to ...
6931 (vec_init<mode><VEC_base_l>): ... this.
6932 (vec_extract<mode>): Renamed to ...
6933 (vec_extract<mode><VEC_base_l>): ... this.
6934 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6935 (vec_initv2sfsf): ... this.
6936 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6937 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6938 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6939 gen_vec_init* calls.
6941 2017-08-01 Richard Biener <rguenther@suse.de>
6943 PR tree-optimization/81297
6944 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6945 TREE_OVERFLOW from INTEGER_CSTs.
6947 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6949 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6951 2017-07-31 Carl Love <cel@us.ibm.com>
6953 * config/rs6000/rs6000-c: Add support for built-in functions
6954 vector signed char vec_xl_be (signed long long, signed char *);
6955 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6956 vector signed int vec_xl_be (signed long long, signed int *);
6957 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6958 vector signed long long vec_xl_be (signed long long, signed long long *);
6959 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6960 vector signed short vec_xl_be (signed long long, signed short *);
6961 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6962 vector double vec_xl_be (signed long long, double *);
6963 vector float vec_xl_be (signed long long, float *);
6964 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6965 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6966 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6968 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6969 (altivec_expand_builtin): Add switch statement to call
6970 altivec_expand_xl_be for each builtin.
6971 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6972 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6973 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6974 __builtin_vsx_le_be_v16qi.
6975 * doc/extend.texi: Update the built-in documentation file for the
6976 new built-in functions.
6978 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6981 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6983 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6985 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6987 * config.gcc: Add z14.
6988 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6989 CPU model numbers for z13s and z14.
6990 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6992 * config/s390/s390-opts.h (enum processor_type): Rename
6993 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6994 * config/s390/s390.c (processor_table): Add field for CPU name to
6995 be passed to Binutils.
6996 (s390_asm_output_machine_for_arch): Use the new field in
6997 processor_table for Binutils.
6998 (s390_expand_builtin): Replace arch12 with z14.
6999 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7000 (s390_get_sched_attrmask): Likewise.
7001 (s390_get_unit_mask): Likewise.
7002 * config/s390/s390.opt: Add z14 to processor_type enum.
7004 2017-07-31 Martin Jambor <mjambor@suse.cz>
7007 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7008 regardless of optimization level.
7010 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7011 Martin Liska <mliska@suse.cz>
7013 * predict.def: Remove old comment and adjust probability.
7014 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7017 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7020 * config/i386/i386.c (ix86_function_naked): New function.
7021 (ix86_can_use_return_insn_p): Return false for naked functions.
7022 (ix86_expand_prologue): Skip prologue for naked functions.
7023 (ix86_expand_epilogue): Skip epilogue for naked functions
7024 and emit trap instruction.
7025 (ix86_warn_func_return): New function.
7026 (ix86_attribute_table): Add "naked" attribute specification.
7027 (TARGET_WARN_FUNC_RETURN): Define.
7028 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7030 2017-07-31 Martin Liska <mliska@suse.cz>
7032 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7033 (dump_gimple_bb_header): Always dump BB info.
7034 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7036 2017-07-31 Martin Liska <mliska@suse.cz>
7039 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7040 also with current_function_decl non-null equality.
7042 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7045 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7046 change type to the element type, instead add eltype variable and
7047 use it where we are interested in the element type.
7049 PR tree-optimization/81603
7050 * ipa-polymorphic-call.c
7051 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7052 offset arithmetic in offset_int, bail out if the resulting bit offset
7053 doesn't fit into shwi.
7055 2017-07-31 Martin Liska <mliska@suse.cz>
7057 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7058 (gimplify_save_expr): Fix comment.
7060 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7063 * config/i386/i386.c (ix86_function_arg): Update arguments for
7065 (ix86_compute_frame_layout): Set the initial stack offset to
7066 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7067 INCOMING_FRAME_SP_OFFSET.
7068 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7069 stack before exception handler returns.
7070 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7071 the 'ERROR_CODE' for exception handler.
7073 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7075 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7076 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7077 (ASM_OUTPUT_REG_POP): Ditto.
7078 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7079 instead of asm_fprintf to output pure string.
7081 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7083 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7084 to imported_module_or_decl hook.
7085 (debug_nothing_tree_tree_tree_bool): Remove.
7086 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7087 * debug.c (do_nothing_debug_hooks): Use
7088 debug_nothing_tree_tree_tree_bool_bool instead of
7089 debug_nothing_tree_tree_tree_bool.
7090 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7091 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7092 * sdbout.c (sdb_debug_hooks): Likewise.
7093 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7094 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7096 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7097 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7098 attribute, don't add anything.
7100 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7102 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7103 (fold_build2_stat_loc): Likewise.
7104 (fold_build3_stat_loc): Likewise.
7105 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7106 (fold_build1_loc): Remove macro.
7107 (fold_build2_loc): Likewise.
7108 (fold_build3_loc): Likewise.
7110 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7112 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7113 (gimple_build_debug_bind_source_stat): Likewise.
7114 * gimple.h (gimple_build_debug_bind): Remove macro.
7115 (gimple_build_debug_bind_source): Likewise.
7117 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7119 * bitmap.c (bitmap_alloc): Adjust.
7120 (bitmap_gc_alloc): Likewise.
7121 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7123 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7125 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7126 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7127 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7128 (bitmap_gc_alloc_stat): Likewise.
7129 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7131 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7133 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7134 * rtl.h (shallow_copy_rtx): Remove macro.
7136 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7138 * emit-rtl.c (gen_raw_REG): Adjust.
7139 * gengenrtl.c (gendef): Likewise.
7140 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7141 * rtl.h (rtx_alloc): Remove macro.
7143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7145 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7146 (build_tree_list_stat): Likewise.
7147 * tree.h (build_tree_list): Remove macro.
7148 (build_tree_list_vec): Likewise.
7150 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7152 * tree.c (make_vector_stat): Remove _stat from name.
7153 (build_vector_stat): Likewise.
7154 * tree.h (make_vector_stat): Remove macro.
7155 (build_vector_stat): Likewise.
7157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7159 * tree.h (build_var_debug_value): Remove prototype.
7161 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7163 * tree.c (tree_cons_stat): Remove _stat from name.
7164 * tree.h (tree_cons): Remove macro.
7166 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7168 * tree.c (build_vl_exp_stat): Remove _stat from name.
7169 * tree.h (build_vl_exp): Remove macro.
7171 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7173 * tree.c (build_decl_stat): Remove _stat from name.
7174 * tree.h (build_decl): Remove macro.
7176 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7178 * gimple.c (gimple_build_with_ops_stat): Adjust.
7179 (gimple_alloc_stat): Remove _stat from name.
7180 * gimple.h (gimple_alloc): Remove macro.
7182 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7184 * tree.c (make_tree_vec_stat): Remove _stat from name.
7185 (grow_tree_vec_stat): Likewise.
7186 * tree.h (make_tree_vec_stat): Adjust prototype.
7187 (grow_tree_vec_stat): Likewise.
7188 (make_tree_vec): Remove macro.
7189 (grow_tree_vec): Likewise.
7191 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7193 * fold-const.c (fold_build1_stat_loc): Adjust.
7194 (fold_build2_stat_loc): Likewise.
7195 (fold_build3_stat_loc): Likewise.
7196 * tree.c (build0_stat): Remove _stat from name.
7197 (build1_stat): Likewise.
7198 (build2_stat): Likewise.
7199 (build3_stat): Likewise.
7200 (build4_stat): Likewise.
7201 (build5_stat): Likewise.
7202 * tree.h (build1_loc): Remove macro, and rename _stat function
7204 (build2_loc): Likewise.
7205 (build3_loc): Likewise.
7206 (build4_loc): Likewise.
7207 (build5_loc): Likewise.
7209 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7211 * tree.c (make_int_cst_stat): Remove _stat from name.
7212 * tree.h (make_int_cst_stat): Adjust prototype.
7213 (make_int_cst): Remove macro.
7215 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7217 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7218 * tree.h (make_tree_binfo_stat): Adjust prototype.
7219 (make_tree_binfo): Remove.
7221 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7223 * tree.c (copy_node_stat): Rename to copy_node.
7224 (build_distinct_type_copy): Adjust.
7225 * tree.h (copy_node_stat): Adjust prototype.
7226 (copy_node): Remove macro.
7228 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7230 * tree.c (make_node_stat): rename to make_node.
7231 (build_tree_list_stat): Adjust.
7232 (build0_stat): Likewise.
7233 (build2_stat): Likewise.
7234 (build3_stat): Likewise.
7235 (build4_stat): Likewise.
7236 (build5_stat): Likewise.
7237 (build_decl_stat): Likewise.
7238 * tree.h (make_node_stat): Adjust prototype.
7239 (make_node): remove macro.
7241 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7243 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7244 (PPC_FEATURE2_SCV): Likewise.
7245 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7247 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7249 * config/aarch64/aarch64.c
7250 (aarch64_internal_mov_immediate): Add new special pattern.
7251 * config/aarch64/aarch64.md (*movdi_aarch64):
7252 Add reg/32bit const mov case.
7254 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7255 Richard Sandiford <richard.sandiford@linaro.org>
7257 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7258 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7259 Add integer and movi cases.
7260 (movi-split-hf-df-sf split, fp16): New.
7261 (enabled): Added TARGET_FP_F16INST.
7262 * config/aarch64/iterators.md (GPF_HF): New.
7263 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7265 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7267 * config/aarch64/aarch64.c
7268 (aarch64_simd_container_mode): Add prototype.
7269 (aarch64_expand_mov_immediate): Add HI support.
7270 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7271 (aarch64_can_const_movi_rtx_p): New.
7272 (aarch64_preferred_reload_class):
7273 Remove restrictions of using FP registers for certain SIMD operations.
7274 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7275 (aarch64_valid_floating_const): Add integer move validation.
7276 (aarch64_simd_imm_scalar_p): Remove.
7277 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7278 (aarch64_legitimate_constant_p): Expand list of supported cases.
7279 * config/aarch64/aarch64-protos.h
7280 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7281 (aarch64_reinterpret_float_as_int): New.
7282 (aarch64_simd_imm_scalar_p): Remove.
7283 * config/aarch64/constraints.md (Uvi): New.
7284 (Dd): Split into Ds and new Dd.
7285 * config/aarch64/aarch64.md (*movsi_aarch64):
7287 (*movdi_aarch64): Add SIMD mov case.
7289 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7291 * tree-predcom.c: (struct chain): Handle store-store chain in which
7292 stores for elimination only store loop invariant values.
7293 (execute_pred_commoning_chain): Ditto.
7294 (prepare_initializers_chain_store_elim): Ditto.
7295 (prepare_finalizers): Ditto.
7296 (is_inv_store_elimination_chain): New function.
7297 (initialize_root_vars_store_elim_1): New function.
7299 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7301 * tree-predcom.c: Revise general description of the pass.
7302 (enum chain_type): New enum type for store elimination.
7303 (struct chain): New field supporting store elimination.
7304 (struct component): Ditto.
7305 (dump_chain): Dump store-stores chain.
7306 (release_chain): Release resources.
7307 (split_data_refs_to_components): Compute and create component
7308 contains only stores for elimination.
7309 (get_chain_last_ref_at): New function.
7310 (make_invariant_chain): Initialization.
7311 (make_rooted_chain): Specify chain type in parameter and record it.
7312 (add_looparound_copies): Skip for store-stores chain.
7313 (determine_roots_comp): Compute type of chain and pass it to
7315 (initialize_root_vars_store_elim_2): New function.
7316 (finalize_eliminated_stores): New function.
7317 (remove_stmt): Handle store for elimination.
7318 (execute_pred_commoning_chain): Execute predictive commoning on
7320 (determine_unroll_factor): Skip unroll for store-stores chain.
7321 (prepare_initializers_chain_store_elim): New function.
7322 (prepare_initializers_chain): Hanlde store-store chain.
7323 (prepare_finalizers_chain, prepare_finalizers): New function.
7324 (tree_predictive_commoning_loop): Return integer value indicating
7325 if loop is unrolled or lcssa form is corrupted.
7326 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7328 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7330 * tree-predcom.c (initialize_root): Delete.
7331 (execute_pred_commoning_chain): Initialize root vars and replace
7332 reference of non-combined chain directly, rather than call above
7335 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7337 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7338 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7340 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7342 * tree-predcom.c (struct chain): New field init_seq.
7343 (release_chain): Release init_seq.
7344 (prepare_initializers_chain): Record intialization stmts in above
7346 (insert_init_seqs): New function.
7347 (tree_predictive_commoning_loop): Call insert_init_seqs.
7349 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7351 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7353 2017-07-28 Richard Biener <rguenther@suse.de>
7355 * match.pd: Remove superfluous :c.
7356 * genmatch.c (simplify::id): Add member.
7357 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7359 (current_id): New global.
7360 (dt_node::parent): Move from ...
7361 (dt_operand::parent): ... here. Add for_id member.
7362 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7363 (decision_tree::find_node): Relax order requirement when
7364 merging DT_TRUE nodes to ones inbetween the current simplify
7365 and the one we try to merge with. Add diagnostic whenever
7366 we need to enforce pattern order by not merging.
7367 (decision_tree::insert): Set current_id.
7368 (decision_tree::print_node): Dump parent node and for_id.
7369 (parser::last_id): Add member.
7370 (parser::push_simplify): Assign unique id.
7371 (parser::parser): Initialize last_id.
7373 2017-07-28 Martin Liska <mliska@suse.cz>
7376 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7377 gimple_build_debug_bind.
7379 2017-07-28 Richard Biener <rguenther@suse.de>
7381 PR tree-optimization/81502
7382 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7384 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7386 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7387 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7388 for BIT_FIELD_REF args.
7389 * fold-const.c (make_bit_field_ref): Likewise.
7390 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7392 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7395 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7396 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7397 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7398 Or it into SANITIZER_UNDEFINED.
7399 * ubsan.c: Include gimple-fold.h and varasm.h.
7400 (ubsan_expand_ptr_ifn): New function.
7401 (instrument_pointer_overflow): New function.
7402 (maybe_instrument_pointer_overflow): New function.
7403 (instrument_object_size): Formatting fix.
7404 (pass_ubsan::execute): Call instrument_pointer_overflow
7405 and maybe_instrument_pointer_overflow.
7406 * internal-fn.c (expand_UBSAN_PTR): New function.
7407 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7408 * sanitizer.def (__ubsan_handle_pointer_overflow,
7409 __ubsan_handle_pointer_overflow_abort): New builtins.
7410 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7411 * internal-fn.def (UBSAN_PTR): New internal function.
7412 * opts.c (sanitizer_opts): Add pointer-overflow.
7413 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7414 * fold-const.c (build_range_check): Compute pointer range check in
7415 integral type if pointer arithmetics would be needed. Formatting
7418 2017-07-28 Martin Liska <mliska@suse.cz>
7421 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7422 parameters that are of a variable-length.
7424 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7426 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7428 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7429 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7430 (CRT_CALL_STATIC_FUNCTION): Likewise.
7431 (ASM_DEFAULT_SPEC): New define.
7432 (ASM_SPEC32): Likewise.
7433 (ASM_SPEC64): Likewise.
7434 (ASM_SPEC_COMMON): Likewise.
7435 (ASM_SPEC): Likewise.
7436 (INVALID_64BIT): Likewise.
7437 (LINK_OS_DEFAULT_SPEC): Likewise.
7438 (LINK_OS_SPEC32): Likewise.
7439 (LINK_OS_SPEC64): Likewise.
7440 (POWERPC_LINUX): Likewise.
7441 (PTRDIFF_TYPE): Likewise.
7442 (RESTORE_FP_PREFIX): Likewise.
7443 (RESTORE_FP_SUFFIX): Likewise.
7444 (SAVE_FP_PREFIX): Likewise.
7445 (SAVE_FP_SUFFIX): Likewise.
7446 (SIZE_TYPE): Likewise.
7447 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7448 (TARGET_64BIT): Likewise.
7449 (TARGET_64BIT): Likewise.
7450 (TARGET_AIX): Likewise.
7451 (WCHAR_TYPE_SIZE): Likewise.
7452 (WCHAR_TYPE): Undefine.
7453 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7454 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7455 (CPP_OS_RTEMS_SPEC): Delete.
7456 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7457 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7459 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7461 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7463 PR tree-optimization/81578
7464 * tree-parloops.c (build_new_reduction): Bail out if
7465 reduction_code isn't one of the standard OpenMP reductions.
7466 Move the details printing after that decision.
7468 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7470 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7471 related to reload_in_progress.
7472 (splat_input_operand): Likewise.
7473 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7475 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7477 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7478 (TARGET_INSTANTIATE_DECLS): Likewise.
7479 (legitimate_indexed_address_p): Delete reload_in_progress code.
7480 (rs6000_debug_legitimate_address_p): Likewise.
7481 (rs6000_eliminate_indexed_memrefs): Likewise.
7482 (rs6000_emit_le_vsx_store): Likewise.
7483 (rs6000_emit_move_si_sf_subreg): Likewise.
7484 (rs6000_emit_move): Likewise.
7485 (register_to_reg_type): Likewise.
7486 (rs6000_pre_atomic_barrier): Likewise.
7487 (rs6000_machopic_legitimize_pic_address): Likewise.
7488 (rs6000_allocate_stack_temp): Likewise.
7489 (rs6000_address_for_fpconvert): Likewise.
7490 (rs6000_address_for_altivec): Likewise.
7491 (rs6000_secondary_memory_needed_rtx): Delete function.
7492 (rs6000_check_sdmode): Likewise.
7493 (rs6000_alloc_sdmode_stack_slot): Likewise.
7494 (rs6000_instantiate_decls): Likewise.
7495 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7496 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7497 Delete reload_in_progress.
7498 (*vec_reload_and_plus_<mptrsize>): Likewise.
7499 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7500 (vsx_div_v2di): Likewise.
7501 (vsx_udiv_v2di): Likewise.
7503 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7505 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7506 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7507 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7508 (rs6000_debug_reg_global): Delete print of LRA status.
7509 (rs6000_option_override_internal): Delete dead LRA related code.
7510 (rs6000_lra_p): Delete function.
7511 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7513 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7515 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7516 * config/riscv/rtems.h: New file.
7518 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7519 Sudakshina Das <sudi.das@arm.com>
7521 * config/aarch64/aarch64.md
7522 (define_split for and<mode>3nr_compare): Move
7523 non aarch64_logical_operand to a register.
7524 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7525 register immediate operand to a register.
7526 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7528 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7531 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7533 2017-07-27 Richard Biener <rguenther@suse.de>
7535 PR tree-optimization/81573
7536 PR tree-optimization/81494
7537 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7538 multi defuse cycle case.
7540 2017-07-27 Richard Biener <rguenther@suse.de>
7542 PR tree-optimization/81571
7543 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7546 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7548 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7549 earlier and only if MASK_FPU is set. Adjust formatting.
7551 2017-07-27 Martin Liska <mliska@suse.cz>
7553 * opt-functions.awk: Add validation of value of Init.
7554 * optc-gen.awk: Pass new argument.
7556 2017-07-27 Martin Liska <mliska@suse.cz>
7558 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7559 Fix wrong condition.
7561 2017-07-27 Martin Liska <mliska@suse.cz>
7563 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7564 BBs and edges seen by autoFDO.
7566 2017-07-27 Richard Biener <rguenther@suse.de>
7568 PR tree-optimization/81502
7569 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7570 with incompatible but same sized type.
7571 (execute_update_addresses_taken): Likewise.
7573 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7575 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7576 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7578 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7581 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7582 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7583 Change s_operand to memory_operand.
7585 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7587 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7588 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7589 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7590 Emit instructions rather than returning an expression. Handle TFmode
7591 and KFmode by casting to TImode.
7592 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7593 (rs6000_emit_le_vsx_store): Likewise.
7594 * config/rs6000/vsx.md (VSX_TI): New iterator.
7595 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7596 (*vsx_le_undo_permute_<mode>): Likewise.
7597 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7598 emit the split sequence.
7599 (*vsx_le_perm_store_<mode>): Likewise.
7601 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7603 PR tree-optimization/81555
7604 PR tree-optimization/81556
7605 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7606 if true, force CHANGED for the recursive invocation.
7607 (reassociate_bb): Remember original length of ops array, pass
7608 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7610 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7611 attributes for noipa attribute. For naked attribute use
7612 lookup_attribute first before lookup_attribute_spec.
7613 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7615 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7616 for functions with noipa attribute.
7617 (cgraph_externally_visible_p): Return true for functions with noipa
7619 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7620 for functions with noipa attribute.
7621 * doc/extend.texi: Document noipa function attribute.
7622 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7623 also for functions with noipa attribute.
7624 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7626 2017-07-26 Andrew Pinski <apinski@cavium.com>
7628 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7629 vec_unalign_load_cost and vec_unalign_store_cost.
7631 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7633 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7634 -mvsx-small-integer option.
7635 (ISA_3_0_MASKS_IEEE): Likewise.
7636 (OTHER_VSX_VECTOR_MASKS): Likewise.
7637 (POWERPC_MASKS): Likewise.
7638 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7639 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7640 code, only testing for DImode being allowed in non-VSX floating
7642 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7643 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7645 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7646 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7647 TARGET_P8_VECTOR test.
7648 (rs6000_secondary_reload_simple_move): Likewise.
7649 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7650 since TARGET_P9_VECTOR was already tested.
7651 (rs6000_opt_masks): Remove -mvsx-small-integer.
7652 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7653 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7655 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7656 test for TARGET_VEXTRACTUB was used, and that uses
7658 (p9 extract splitter): Likewise.
7659 (vsx_extract_<mode>_di_p9): Likewise.
7660 (vsx_extract_<mode>_store_p9): Likewise.
7661 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7662 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7663 the elimination of TARGET_VSX_SMALL_INTEGER.
7664 (vsx_extract_<mode>_p8): Likewise.
7665 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7666 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7667 (vsx_set_<mode>_p9): Likewise.
7668 (vsx_set_v4sf_p9): Likewise.
7669 (vsx_set_v4sf_p9_zero): Likewise.
7670 (vsx_insert_extract_v4sf_p9): Likewise.
7671 (vsx_insert_extract_v4sf_p9_2): Likewise.
7672 * config/rs6000/rs6000.md (sign extend splitter): Change
7673 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7674 (floatsi<mode>2_lfiwax_mem): Likewise.
7675 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7676 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7677 since a test for TARGET_P9_VECTOR was used.
7678 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7679 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7680 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7681 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7682 TARGET_P8_VECTOR test.
7683 (fix_trunc<mode>si2_stfiwx): Likewise.
7684 (fix_trunc<mode>si2_internal): Likewise.
7685 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7686 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7688 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7689 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7690 TARGET_P8_VECTOR test.
7691 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7692 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7693 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7695 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7696 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7697 since a test for TARGET_P9_VECTOR was used.
7698 (splitter for loading small constants): Likewise.
7700 2017-07-26 Andrew Pinski <apinski@cavium.com>
7702 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7705 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7708 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7709 (fp_valid_at): Likewise.
7711 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7713 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7714 (qdf24xx_addrcost_table): Likewise.
7715 (cortexa57_tunings): Update to use generic_branch_cost.
7716 (cortexa72_tunings): Likewise.
7717 (cortexa73_tunings): Likewise.
7718 (qdf24xx_tunings): Likewise.
7720 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7722 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7723 (thunderx2t99_branch_cost): Likewise.
7724 (cortexa35_tunings): Update to use generic_branch_cost.
7725 (cortexa53_tunings): Likewise.
7726 (cortexa57_tunings): Likewise.
7727 (cortexa72_tunings): Likewise.
7728 (cortexa73_tunings): Likewise.
7729 (thunderx2t99_tunings): Likewise.
7731 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7733 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7734 (sparc_option_override): Honour MASK_FSMULD.
7735 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7736 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7737 * config/sparc/sparc.opt (mfsmuld): New option.
7738 * doc/invoke.texi (mfsmuld): Document option.
7740 2017-07-26 Marek Polacek <polacek@redhat.com>
7743 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7745 2017-07-26 Richard Biener <rguenther@suse.de>
7747 * gimple-match-head.c (do_valueize): Return OP if valueize
7749 (get_def): New helper to get at the def stmt of a SSA name
7751 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7752 do_valueize to get at the def stmt.
7753 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7755 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7758 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7760 2017-07-26 Martin Liska <mliska@suse.cz>
7763 * function.c (expand_function_start): Make expansion of
7764 nonlocal_goto_save_area after parm_birth_insn.
7766 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7768 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7769 from all CPU target flags enable members.
7771 2017-07-26 Richard Biener <rguenther@suse.de>
7773 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7774 (decision_tree::gen): Make 'type' const.
7775 (write_predicate): Likewise.
7777 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7779 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7780 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7781 (rs6000_option_override_internal): Likewise.
7782 (rs6000_expand_vector_set): Likewise.
7783 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7784 (TARGET_UPPER_REGS_SF): Likewise.
7785 (TARGET_UPPER_REGS_DI): Likewise.
7786 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7787 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7788 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7789 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7790 (Splitters for DI constants in Altivec registers): Likewise.
7791 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7792 (vsx_set_v4sf_p9): Likewise.
7793 (vsx_set_v4sf_p9_zero): Likewise.
7794 (vsx_insert_extract_v4sf_p9): Likewise.
7795 (vsx_insert_extract_v4sf_p9_2): Likewise.
7797 2017-07-25 Carl Love <cel@us.ibm.com>
7799 * doc/extend.texi: Update the built-in documentation file for the
7800 existing built-in functions
7801 vector signed char vec_cnttz (vector signed char);
7802 vector unsigned char vec_cnttz (vector unsigned char);
7803 vector signed short vec_cnttz (vector signed short);
7804 vector unsigned short vec_cnttz (vector unsigned short);
7805 vector signed int vec_cnttz (vector signed int);
7806 vector unsigned int vec_cnttz (vector unsigned int);
7807 vector signed long long vec_cnttz (vector signed long long);
7808 vector unsigned long long vec_cnttz (vector unsigned long long);
7810 2017-07-25 Andrew Pinski <apinski@cavium.com>
7812 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7813 accesses where the use is for the first operand of a BIT_INSERT.
7815 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7818 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7819 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7821 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7823 * config/i386/gstabs.h: Delete.
7824 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7826 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7828 * config/i386/i386.c (ix86_decompose_address): Do not check for
7829 register RTX when looking at index_reg or base_reg.
7830 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7832 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7834 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7835 to update EH info here.
7837 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7839 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7841 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7843 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7845 2017-07-25 Torsten Duwe <duwe@suse.de>
7847 * common.opt: Introduce -fpatchable-function-entry
7848 command line option, and its variables function_entry_patch_area_size
7849 and function_entry_patch_area_start.
7850 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7851 including a two-value parser.
7852 * target.def (print_patchable_function_entry): New target hook.
7853 * targhooks.h (default_print_patchable_function_entry): New function.
7854 * targhooks.c (default_print_patchable_function_entry): Likewise.
7855 * toplev.c (process_options): Switch off IPA-RA if
7856 patchable function entries are being generated.
7857 * varasm.c (assemble_start_function): Look at the
7858 patchable-function-entry command line switch and current
7859 function attributes and maybe generate NOP instructions by
7860 calling the print_patchable_function_entry hook.
7861 * doc/extend.texi: Document patchable_function_entry attribute.
7862 * doc/invoke.texi: Document -fpatchable_function_entry
7863 command line option.
7864 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7866 * doc/tm.texi: Re-generate.
7868 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7871 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7872 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7874 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7876 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7878 2017-07-25 Richard Biener <rguenther@suse.de>
7880 PR tree-optimization/81455
7881 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7882 not walk in cycles when looking for guards.
7884 2017-07-25 Richard Biener <rguenther@suse.de>
7886 PR tree-optimization/81529
7887 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7888 when optimizing backedge uses.
7890 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7892 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7894 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7895 to dl_section_ref. On AIX, append an expression to subtract
7896 the size of the section length to dl_section_ref.
7898 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7900 * configure.ac: If any of the config.* scripts fail, exit 1.
7901 * configure: Regenerate.
7903 2017-07-25 Richard Biener <rguenther@suse.de>
7906 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7907 of immediate uses, be more verbose on errors.
7909 2017-07-25 Richard Biener <rguenther@suse.de>
7911 PR tree-optimization/81510
7912 * tree-vect-loop.c (vect_is_simple_reduction): When the
7913 reduction stmt is not inside the loop bail out.
7915 2017-07-25 Richard Biener <rguenther@suse.de>
7917 PR tree-optimization/81303
7918 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7919 profitability check against LOOP_VINFO_NITERSM1.
7921 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7923 * domwalk.c (cmp_bb_postorder): Simplify.
7924 (sort_bbs_postorder): New function. Use it...
7925 (dom_walker::walk): ...here to optimize common cases.
7927 2017-07-25 Martin Liska <mliska@suse.cz>
7930 * ipa-visibility.c (function_and_variable_visibility): Make the
7931 redirection just on target that supports aliasing.
7932 Fix GNU coding style.
7934 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7937 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7939 * config/arm/rtems.h: Likewise.
7940 * config/bfin/rtems.h: Likewise.
7941 * config/i386/rtemself.h: Likewise.
7942 * config/lm32/rtems.h: Likewise.
7943 * config/m32c/rtems.h: Likewise.
7944 * config/m68k/rtemself.h: Likewise.
7945 * config/microblaze/rtems.h: Likewise.
7946 * config/mips/rtems.h: Likewise.
7947 * config/moxie/rtems.h: Likewise.
7948 * config/nios2/rtems.h: Likewise.
7949 * config/powerpcspe/rtems.h: Likewise.
7950 * config/rs6000/rtems.h: Likewise.
7951 * config/rtems.h: Likewise.
7952 * config/sh/rtems.h: Likewise.
7953 * config/sh/rtemself.h: Likewise.
7954 * config/sparc/rtemself.h: Likewise.
7956 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7959 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7960 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7961 * tree-ssa-structalias.c (alias_get_name): Same.
7963 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7966 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7967 instructions if no du chain is found.
7969 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7971 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7973 2017-07-25 Richard Biener <rguenther@suse.de>
7976 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7979 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7981 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7983 (ISA_2_7_MASKS_SERVER): Likewise.
7984 (ISA_3_0_MASKS_IEEE): Likewise.
7985 (OTHER_P8_VECTOR_MASKS): Likewise.
7986 (OTHER_VSX_VECTOR_MASKS): Likewise.
7987 (POWERPC_MASKS): Likewise.
7988 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7989 duplicate list of options.
7990 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7991 explicit -mupper-regs options.
7992 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7993 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7994 alias for -mupper-regs-df.
7995 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7996 (rs6000_init_hard_regno_mode_ok): Likewise.
7997 (rs6000_option_override_internal): Likewise.
7998 (rs6000_opt_masks): Likewise.
7999 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8000 options in terms of whether -mvsx or -mpower8-vector was used.
8001 (TARGET_UPPER_REGS_DI): Likewise.
8002 (TARGET_UPPER_REGS_SF): Likewise.
8003 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8004 -mupper-regs-* options.
8006 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8008 * passes.c (emergency_dump_function): Print some empty lines and a
8009 header before the RTL dump.
8011 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8013 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8015 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8018 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8019 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8021 2017-07-24 Carl Love <cel@us.ibm.com>
8023 * config/rs6000/rs6000-c.c: Add support for built-in functions
8024 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8025 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8026 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8027 vec_extract_fp_from_shortl): Add defines for the two builtins.
8028 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8029 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8031 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8032 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8033 * doc/extend.texi: Update the built-in documentation file for the
8034 new built-in function.
8036 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8039 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8041 * doc/generic.texi: Likewise.
8042 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8043 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8045 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8047 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8048 (aarch64_mls_elt_merge<mode>): Likewise.
8050 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8052 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8053 having __cxa_atexit.
8055 2017-07-23 Michael Collison <michael.collison@arm.com>
8057 * config/arm/arm.c (arm_option_override): Deprecate
8058 use of -mstructure-size-boundary.
8059 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8060 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8062 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8065 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8066 Reduce cost estimate for direct moves.
8068 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8071 * config/i386/i386.c (ix86_option_override_internal): Disable
8072 BMI, BMI2 and TBM instructions for -m16.
8074 2017-07-21 Carl Love <cel@us.ibm.com>
8076 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8077 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8078 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8079 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8080 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8081 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8082 VMULOSW): New enum "unspec" values.
8083 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8084 altivec_vmulosw): New patterns.
8085 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8086 VMULOSW): Add definitions.
8088 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8090 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8091 (qdf24xx): Likewise.
8092 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8093 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8094 (AARCH64_FL_V8_1): Renumber.
8095 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8096 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8097 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8098 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8099 rdma to feature modifiers list.
8101 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8104 * ipa-visibility (function_and_variable_visibility): Convert
8105 recursive PLT call to direct call if appropriate.
8107 2017-07-21 Andrew Pinski <apinski@cavium.com>
8109 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8110 operand 1 to see if the types precision matches.
8111 * fold-const.c (operand_equal_p): Likewise.
8113 2017-07-21 Richard Biener <rguenther@suse.de>
8115 PR tree-optimization/81303
8116 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8117 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8118 (vect_peeling_hash_get_lowest_cost): Adjust.
8119 (vect_enhance_data_refs_alignment): Likewise. Use
8120 vect_get_peeling_costs_all_drs to compute the penalty for no
8121 peeling to match up costs.
8123 2017-07-21 Richard Biener <rguenther@suse.de>
8125 PR tree-optimization/81500
8126 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8127 we didn't identify a reduction path.
8129 2017-07-21 Tom de Vries <tom@codesourcery.com>
8130 Cesar Philippidis <cesar@codesourcery.com>
8132 PR gcov-profile/81442
8133 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8136 2017-07-21 Tom de Vries <tom@codesourcery.com>
8139 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8141 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8142 nvptx_override_options_after_change.
8144 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8146 * dwarf2out.c (output_file_names): Avoid double testing for
8149 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8151 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8153 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8155 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8157 (try_crossjump_to_edge): Do not punt on partitioned functions.
8159 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8161 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8162 Put all BBs reachable only via paths crossing cold region to cold
8164 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8166 2016-07-21 Richard Biener <rguenther@suse.de>
8168 PR tree-optimization/81303
8169 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8170 into account prologue and epilogue iterations when raising
8171 min_profitable_iters to sth at least covering one vector iteration.
8173 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8175 * config/arm/arm.c (arm_test_cpu_arch_dat):
8178 2017-07-20 Nathan Sidwell <nathan@acm.org>
8180 Remove TYPE_METHODS.
8181 * tree.h (TYPE_METHODS): Delete.
8182 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8183 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8184 (dbxout_type_methods): Scan TYPE_FIELDS.
8185 (dbxout_type): Don't check TYPE_METHODS here.
8186 * function.c (use_register_for_decl): Always ignore register for
8187 class types when not optimizing.
8188 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8189 * tree.c (free_lang_data_in_type): Stitch out member functions and
8190 templates from TYPE_FIELDS.
8191 (build_distinct_type_copy, verify_type_variant,
8192 verify_type): Member fns are on TYPE_FIELDS.
8193 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8194 * tree-pretty-print.c (dump_generic_node): Likewise.
8196 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8199 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8200 V2TImode and V4TImode.
8201 (ix86_expand_vector_extract): Likewise.
8202 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8203 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8204 (ssescalarmode): Handle V4TImode and V2TImode.
8205 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8206 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8207 (VEXTRACTI128_MODE): New mode iterator.
8208 (splitter for *vec_extractv?ti first element): New.
8209 (VEC_INIT_MODE): New mode iterator.
8210 (vec_init<mode>): Consolidate 3 expanders into one using
8211 VEC_INIT_MODE mode iterator.
8213 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8215 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8216 non_spilled_static_chain_regno_p.
8218 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8220 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8222 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8224 * bb-reorder.c (connect_traces): Allow copying of blocks within
8227 2017-07-20 Richard Biener <rguenther@suse.de>
8229 * gimple.h (gimple_phi_result): Add gphi * overload.
8230 (gimple_phi_result_ptr): Likewise.
8231 (gimple_phi_arg): Likewise. Adjust index assert to only
8232 allow actual argument accesses rather than all slots available
8234 (gimple_phi_arg_def): Add gphi * overload.
8235 * tree-phinodes.c (make_phi_node): Initialize only actual
8237 (resize_phi_node): Clear memory not covered by old node,
8238 do not initialize excess argument slots.
8239 (reserve_phi_args_for_new_edge): Initialize new argument slot
8242 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8244 PR tree-optimization/81388
8246 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8248 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8249 by removing computation of may_be_zero.
8251 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8252 Tom de Vries <tom@codesourcery.com>
8255 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8256 when gimple level profile disagrees with what RTL expander did.
8258 2017-07-20 Richard Biener <rguenther@suse.de>
8260 PR tree-optimization/61171
8261 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8262 (vect_analyze_stmt): Add slp instance parameter.
8263 (vectorizable_reduction): Likewise.
8264 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8265 (vect_is_simple_reduction): Deal with chains not detected
8266 as SLP reduction chain, specifically not properly associated
8267 chains containing a mix of plus/minus.
8268 (get_reduction_op): Remove.
8269 (get_initial_defs_for_reduction): Simplify, pass in whether
8270 this is a reduction chain, pass in the SLP node for the PHIs.
8271 (vect_create_epilog_for_reduction): Get the SLP instance as
8273 (vectorizable_reduction): Get the SLP instance as arg.
8274 During analysis remember the SLP node with the PHIs in the
8275 instance. Simplify getting at the vectorized reduction PHIs.
8276 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8277 through SLP instance.
8278 (vect_slp_analyze_operations): Likewise.
8279 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8280 (vect_transform_stmt): Likewise.
8282 2017-07-20 Tom de Vries <tom@codesourcery.com>
8284 PR tree-optimization/81489
8285 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8286 read of phi arg location to before loop that modifies phi.
8288 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8290 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8293 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8296 * except.c (execute): Fix ordering issue.
8298 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8300 PR rtl-optimization/81423
8301 * combine.c (make_compound_operation_int): Don't try to optimize
8302 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8304 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8306 PR rtl-optimization/81423
8307 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8308 with a constant that is -1 in the truncated to mode.
8310 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8312 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8313 (determine_unlikely_bbs): ... here.
8314 * predict.h (propagate_unlikely_bbs_forward): Declare.
8315 * cfgexpand.c (pass_expand::execute): Use it.
8316 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8318 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8319 propagate_unlikely_bbs_forward.
8321 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8324 * except.c (maybe_add_nop_after_section_switch): New function.
8327 2017-07-19 Tom de Vries <tom@codesourcery.com>
8329 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8331 2017-07-19 Tom de Vries <tom@codesourcery.com>
8333 * gimple.h (gimple_phi_arg): Make assert more strict.
8335 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8337 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8338 * config/rs6000/mmintrin.h: New file.
8339 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8341 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8343 PR tree-optimization/81346
8344 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8346 2017-07-19 Tom de Vries <tom@codesourcery.com>
8348 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8350 2017-07-19 Tom de Vries <tom@codesourcery.com>
8352 * config/nvptx/nvptx-modes.def: Add V2DImode.
8353 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8354 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8355 (nvptx_output_mov_insn): Handle lack of mov.b128.
8356 (nvptx_print_operand): Handle 'H' and 'L' codes.
8357 (nvptx_vector_mode_supported): Allow V2DImode.
8358 (nvptx_preferred_simd_mode): New function.
8359 (nvptx_data_alignment): New function.
8360 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8361 nvptx_preferred_simd_mode.
8362 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8364 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8366 2017-07-19 Tom de Vries <tom@codesourcery.com>
8368 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8369 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8370 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8371 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8372 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8373 (mov<VECIM>_insn): New define_insn.
8374 (define_expand "mov<VECIM>): New define_expand.
8376 2017-07-19 Tom de Vries <tom@codesourcery.com>
8378 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8380 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8382 PR tree-optimization/81346
8383 * fold-const.h (fold_div_compare, range_check_type): Declare.
8384 * fold-const.c (range_check_type): New function.
8385 (build_range_check): Use range_check_type.
8386 (fold_div_compare): No longer static, rewritten into
8387 a match.pd helper function.
8388 (fold_comparison): Don't call fold_div_compare here.
8389 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8392 2017-07-19 Nathan Sidwell <nathan@acm.org>
8394 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8395 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8396 * tree.c (find_decls_types_r, verify_type): Use
8397 TYPE_{MIN,MAX}_VALUE_RAW.
8398 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8399 (hash_tree): Likewise.
8400 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8402 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8405 2017-07-18 Tom de Vries <tom@codesourcery.com>
8408 * omp-expand.c (expand_omp_for_static_chunk): Handle
8409 equal-argument loop exit phi.
8411 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8414 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8415 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8416 operand 2 predicate.
8417 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8418 operand 2 predicate.
8419 (ror,rol -> rorx splitters): Use const_int_operand as
8420 operand 2 predicate.
8422 2017-06-18 Richard Biener <rguenther@suse.de>
8424 PR tree-optimization/81410
8425 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8426 the gap in the ! slp_perm SLP case after each group.
8428 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8431 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8434 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8437 * predict.c (set_even_probabilities): Cleanup; do not affect
8438 probabilities that are already known.
8439 (combine_predictions_for_bb): Call even when count is set.
8441 2017-07-18 Nathan Sidwell <nathan@acm.org>
8443 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8446 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8449 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8450 optimization for loop niter analysis.
8452 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8455 * config/avr/avr.c (avr_optimize_casesi): Don't use
8456 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8458 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8460 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8461 body_cost_vec from _vect_peel_extended_info.
8462 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8463 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8466 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8468 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8470 2017-07-18 Richard Biener <rguenther@suse.de>
8472 PR tree-optimization/80620
8473 PR tree-optimization/81403
8474 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8475 info when re-using a VN table entry.
8477 2017-07-18 Richard Biener <rguenther@suse.de>
8479 PR tree-optimization/81418
8480 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8481 vectype_in. Verify that with lane-reducing reduction operations
8482 we have a single def-use cycle.
8484 2017-07-17 Carl Love <cel@us.ibm.com>
8486 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8488 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8489 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8490 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8492 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8493 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8494 VMULOSW): New enum "unspec" values.
8495 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8496 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8497 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8498 altivec_vmulosw): New patterns.
8499 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8500 VMULOSW): Add definitions.
8501 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8503 * config/alpha/alpha.c: Include predict.h.
8505 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8507 * tree-vrp.c (compare_assert_loc): Fix comparison function
8508 to return predictable results.
8510 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8512 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8515 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8516 * doc/invoke.texi (mexpand-adddi): Update text.
8518 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8520 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8521 that also clobbers the CC register. The old expand code is moved
8523 (*arc_clzsi2): ... here.
8524 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8525 the CC register. The old expand code is moved to ...
8526 (arc_ctzsi2): ... here.
8528 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8530 * config/arc/arc.opt (mindexed-loads): Use initial value
8531 TARGET_INDEXED_LOADS_DEFAULT.
8532 (mauto-modify-reg): Use initial value
8533 TARGET_AUTO_MODIFY_REG_DEFAULT.
8534 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8535 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8536 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8537 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8539 2017-07-17 Martin Liska <mliska@suse.cz>
8542 * opts.c (finish_options): Do not allow -fgnu-tm
8543 w/ -fsanitize={kernel-,}address. Say sorry.
8545 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8548 * tree-loop-distribution.c (classify_partition): Only assert on
8549 numer of iterations.
8550 (merge_dep_scc_partitions): Delete prameter. Update function call.
8551 (distribute_loop): Remove code handling loop with unknown niters.
8552 (pass_loop_distribution::execute): Skip loop with unknown niters.
8554 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8557 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8558 function sort_partitions_by_post_order.
8560 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8562 PR tree-optimization/81374
8563 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8564 the max index of basic blocks, rather than number of basic blocks.
8566 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8568 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8570 (arc_legitimate_pic_operand_p): Likewise.
8571 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8573 (arc_needs_pcl_p): Likewise.
8574 (arc_legitimate_pc_offset_p): Likewise.
8575 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8576 function is also used in constrains.md.
8577 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8578 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8579 PLUS. Only return true/false in known cases, otherwise assert.
8580 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8581 is already called in arc_legitimate_constant_p.
8582 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8584 (LEGITIMATE_PIC_OPERAND_P): Use
8585 arc_raw_symbolic_reference_mentioned_p function.
8586 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8591 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8592 Andrew Burgess <andrew.burgess@embecosm.com>
8594 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8595 (arc_return_address_register): New function.
8596 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8597 (arc_handle_fndecl_attribute): Add naked attribute.
8598 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8599 (TARGET_WARN_FUNC_RETURN): Likewise.
8600 (arc_allocate_stack_slots_for_args): New function.
8601 (arc_warn_func_return): Likewise.
8602 (machine_function): Change type fn_type.
8603 (arc_compute_function_type): Consider new naked function type,
8604 change function return type.
8605 (arc_must_save_register): Adapt to handle new
8606 arc_compute_function_type's return type.
8607 (arc_expand_prologue): Likewise.
8608 (arc_expand_epilogue): Likewise.
8609 (arc_return_address_regs): Delete.
8610 (arc_return_address_register): New function.
8611 (arc_epilogue_uses): Use above function.
8612 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8613 (arc_function_type): Change encoding, add naked type.
8614 (ARC_INTERRUPT_P): Change to handle the new encoding.
8615 (ARC_FAST_INTERRUPT_P): Likewise.
8616 (ARC_NORMAL_P): Define.
8617 (ARC_NAKED_P): Likewise.
8618 (arc_compute_function_type): Delete prototype.
8619 * config/arc/arc.md (in_ret_delay_slot): Use
8620 arc_return_address_register function.
8621 (simple_return): Likewise.
8622 (p_return_i): Likewise.
8624 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8626 PR tree-optimization/81428
8627 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8628 can't be built for those types.
8630 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8632 Remove stuff dead since r239246.
8634 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8635 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8636 (avr_inform_devices): Remove dead stuff.
8638 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8640 * config/arm/arm_neon.h: Fix softp typo.
8642 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8644 PR tree-optimization/81365
8645 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8646 aggregate moves onto bb predecessor edges, make sure there are no
8647 loads that could alias the lhs in between the start of bb and the
8650 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8653 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8654 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8655 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8657 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8659 PR tree-optimization/81396
8660 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8661 (init_symbolic_number): Initialize it to 1.
8662 (perform_symbolic_merge): Add n_ops from both operands into the new
8664 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8665 without base_addr as useless if they need more than one operation.
8666 (bswap_replace): Handle !bswap case for NULL base_addr.
8668 2017-07-17 Tom de Vries <tom@codesourcery.com>
8671 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8674 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8676 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8677 conditional builtin define __FIX_LEON3FT_B2BST.
8679 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8681 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8682 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8685 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8687 PR rtl-optimization/81424
8688 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8689 to remove potential trapping from operands if -fnon-call-exceptions.
8691 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8693 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8694 profile_proability for scalling.
8695 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8697 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8699 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8701 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8703 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8704 fixpoint arithmetics.
8706 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8708 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8709 fixpoint arithmetics.
8711 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8713 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8714 fixpoint arithmetics.
8716 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8718 * profile-count.h (profile_probability::from_reg_br_prob_note,
8719 profile_probability::to_reg_br_prob_note): New functions.
8720 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8721 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8722 * predict.c (probability_reliable_p): Update.
8723 (edge_probability_reliable_p): Update.
8724 (br_prob_note_reliable_p): Update.
8725 (invert_br_probabilities): Update.
8726 (add_reg_br_prob_note): New function.
8727 (combine_predictions_for_insn): Update.
8728 * asan.c (asan_clear_shadow): Update.
8729 * cfgbuild.c (compute_outgoing_frequencies): Update.
8730 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8731 (update_br_prob_note): Update.
8732 (rtl_verify_edges): Update.
8733 (purge_dead_edges): Update.
8734 (fixup_reorder_chain): Update.
8735 * emit-rtl.c (try_split): Update.
8736 * ifcvt.c (cond_exec_process_insns): Update.
8737 (cond_exec_process_if_block): Update.
8738 (dead_or_predicable): Update.
8739 * internal-fn.c (expand_addsub_overflow): Update.
8740 (expand_neg_overflow): Update.
8741 (expand_mul_overflow): Update.
8742 * loop-doloop.c (doloop_modify): Update.
8743 * loop-unroll.c (compare_and_jump_seq): Update.
8744 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8745 * predict.h: Update.
8746 * reorg.c (mostly_true_jump): Update.
8748 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8749 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8750 * config/arc/arc.c: (emit_unlikely_jump): Update.
8751 * config/arm/arm.c: (emit_unlikely_jump): Update.
8752 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8753 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8754 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8755 (ix86_print_operand): Update.
8756 (ix86_split_fp_branch): Update.
8757 (predict_jump): Update.
8758 * config/ia64/ia64.c (ia64_print_operand): Update.
8759 * config/mmix/mmix.c (mmix_print_operand): Update.
8760 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8761 (rs6000_expand_split_stack_prologue): Update.
8762 * config/rs6000/rs6000.c: Update.
8763 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8764 (s390_expand_vec_movstr): Update.
8765 (s390_expand_cs_tdsi): Update.
8766 (s390_expand_split_stack_prologue): Update.
8767 * config/sh/sh.c (sh_print_operand): Update.
8768 (expand_cbranchsi4): Update.
8769 (expand_cbranchdi4): Update.
8770 * config/sparc/sparc.c (output_v9branch): Update.
8771 * config/spu/spu.c (get_branch_target): Update.
8772 (ea_load_store_inline): Update.
8773 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8774 * config/tilepro/tilepro.c: Update.
8776 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8778 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8779 (gimplify_save_expr): Likewise.
8781 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8783 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8785 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8787 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8789 * timevar.def (TV_IPA_FNSUMMARY): Define.
8791 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8793 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8794 to back store errata sensitive sequence from being generated.
8795 (sqrtdf2_fix): Likewise.
8797 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8799 * tree-ssa-threadupdate.c (compute_path_counts,
8800 update_joiner_offpath_counts): Use profile_probability.
8802 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8805 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8807 * config/arm/arm-c.c (arm_cpu_builtins): Define
8808 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8810 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8812 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8813 array entries to represent __ieee128 versions of the
8814 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8815 scalar_extract_sig, and scalar_insert_exp built-in functions.
8816 (altivec_resolve_overloaded_builtin): Add special case handling
8817 for the __builtin_scalar_insert_exp function, as represented by
8818 the P9V_BUILTIN_VEC_VSIEDP constant.
8819 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8820 exponent support for __ieee128 argument.
8821 (VSESQP): Add scalar extract signature support for __ieee128
8823 (VSTDCNQP): Add scalar test negative support for __ieee128
8825 (VSIEQP): Add scalar insert exponent support for __int128 argument
8826 with __ieee128 result.
8827 (VSIEQPF): Add scalar insert exponent support for __ieee128
8828 argument with __ieee128 result.
8829 (VSTDCQP): Add scalar test data class support for __ieee128
8831 (VSTDCNQP): Add overload support for scalar test negative with
8833 (VSTDCQP): Add overload support for scalar test data class
8835 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8837 (UNSPEC_VSX_SIEXPQP): New constant.
8838 (xsxexpqp): New insn for VSX scalar extract exponent quad
8840 (xsxsigqp): New insn for VSX scalar extract significand quad
8842 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8843 precision with floating point argument.
8844 (xststdcqp): New expand for VSX scalar test data class quad
8846 (xststdcnegqp): New expand for VSX scalar test negative quad
8848 (xststdcqp): New insn to match expansions for VSX scalar test data
8849 class quad precision and VSX scalar test negative quad precision.
8850 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8851 special case operand checking to enforce that second operand of
8852 VSX scalar test data class with quad precision argument is a 7-bit
8854 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8855 prototypes and descriptions of __ieee128 versions of
8856 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8857 scalar_test_data_class, and scalar_test_neg built-in functions.
8859 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8861 PR tree-optimization/81162
8862 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8863 replace a negate with an add.
8865 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8867 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8869 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8871 * config/arm/arm-c.c (arm_cpu_builtins): Define
8872 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8874 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8876 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8877 (armv8-r): Set ARM Cortex-R52 as default CPU.
8878 * config/arm/arm-tables.opt: Regenerate.
8879 * config/arm/arm-tune.md: Regenerate.
8880 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8882 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8883 extension for -mcpu=cortex-r52.
8885 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8887 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8888 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8889 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8890 (fp-armv8): Define it as FP_ARMv8 only.
8891 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8892 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8894 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8895 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8896 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8897 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8898 than TARGET_FPU_ARMV8.
8899 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8900 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8901 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8903 * config/arm/neon.md (neon_vrint): Likewise.
8904 (neon_vcvt): Likewise.
8905 (neon_<fmaxmin_op><mode>): Likewise.
8906 (<fmaxmin><mode>3): Likewise.
8907 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8908 * config/arm/predicates.md (arm_cond_move_operator): Check against
8909 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8911 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8913 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8916 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8918 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8919 loop in comment with memset.
8921 2017-07-14 Martin Liska <mliska@suse.cz>
8923 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8924 * dwarf2out.c (is_java): Remove the function.
8925 (output_pubname): Remove usage of the function.
8926 (lower_bound_default): Remove usage of DW_LANG_Java.
8927 (gen_compile_unit_die): Likewise.
8928 * gcc.c: Remove compiler defaults for .java and .zip files.
8929 * gimple-expr.c (remove_suffix): Change as there's no longer
8930 extension than 4-letter one.
8931 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8932 (gimplify_save_expr): Likewise.
8933 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8934 as it's possible even for other languages than Java.
8935 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8936 * lto-opts.c (lto_write_options): Remove reference to Java.
8937 * opts.c (strip_off_ending): Update file extension handling.
8938 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8939 * tree-eh.c (lower_resx): Likewise.
8940 * tree.c (free_lang_data_in_type): Remove dead code.
8941 (find_decls_types_r): Likewise.
8942 (build_common_builtin_nodes): Remove Java from a comment.
8943 (verify_type): Remove dead code.
8944 * varasm.c (assemble_external): Remove Java from a comment.
8946 2017-07-14 Martin Liska <mliska@suse.cz>
8948 * opts.c (finish_options): Add quotes.
8949 (common_handle_option): Likewise.
8951 2017-07-14 Martin Liska <mliska@suse.cz>
8953 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8954 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8956 * dwarf2out.c (lower_bound_default): Do not handle
8958 (gen_compile_unit_die): Likewise.
8959 * gcc.c: Remove default extension binding for GNU Pascal.
8960 * stmt.c: Remove Pascal language from a comment.
8961 * xcoffout.c: Likewise.
8963 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8966 * diagnostic-show-locus.c (fixit_cmp): New function.
8967 (layout::layout): Sort m_fixit_hints.
8968 (column_range::column_range): Assert that the values are valid.
8969 (struct char_span): New struct.
8970 (correction::overwrite): New method.
8971 (struct source_line): New struct.
8972 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8973 calls in terms of classes source_line and char_span, and
8974 correction::overwrite.
8975 (selftest::test_overlapped_fixit_printing_2): New function.
8976 (selftest::diagnostic_show_locus_c_tests): Call it.
8978 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8981 early if there is no lhs.
8983 2017-07-13 Martin Liska <mliska@suse.cz>
8985 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8986 (gen_reference_type_die): Likewise.
8987 * stor-layout.c: Remove Pascal-related comment.
8989 2017-07-13 Martin Liska <mliska@suse.cz>
8991 * opts.c (finish_options): Add quotes to error messages.
8992 (parse_sanitizer_options): Likewise.
8994 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8996 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8998 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9000 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9002 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9004 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9006 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9008 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9011 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9012 provides the hardware capability bits, define the macro
9013 __BUILTIN_CPU_SUPPORTS__.
9014 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9015 if GLIBC does not provide the hardware capability bits. Add a
9016 gcc_unreachable call if the built-in cpu function is neither
9017 __builtin_cpu_is nor __builtin_cpu_supports.
9018 (rs6000_get_function_versions_dispatcher): Change the warning
9019 that an old GLIBC is used which does not export the capability
9020 bits to be an error.
9021 * doc/extend.texi (target_clones attribute): Document the
9022 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9023 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9024 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9025 the macros defined by GCC if the newer GLIBC is available.
9027 2017-07-12 Jeff Law <law@redhat.com>
9029 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9030 remaining includes slightly.
9031 * config/riscv/riscv-builtins.c: Include profile-count.h.
9033 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9036 * config/avr/avr.c (avr_set_current_function): In diagnostic
9037 messages: Quote keywords and (parts of) identifiers.
9038 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9041 2017-07-12 Carl Love <cel@us.ibm.com>
9043 * config/rs6000/rs6000-c.c: Add support for built-in functions
9044 vector bool char vec_revb (vector bool char);
9045 vector bool short vec_revb (vector short char);
9046 vector bool int vec_revb (vector bool int);
9047 vector bool long long vec_revb (vector bool long long);
9048 * doc/extend.texi: Update the built-in documentation file for the
9049 new built-in functions.
9051 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9053 * config/s390/s390.md: Remove movcc splitter.
9055 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9057 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9058 load/store on condition.
9060 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9063 * config/avr/avr.c (avr_encode_section_info)
9064 [progmem && !TREE_READONLY]: Error if progmem object needs
9067 2017-07-11 Michael Collison <michael.collison@arm.com>
9069 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9072 2017-07-11 Carl Love <cel@us.ibm.com>
9074 * config/rs6000/rs6000-c.c: Add support for builtins
9075 vector unsigned int vec_parity_lsbb (vector signed int);
9076 vector unsigned int vec_parity_lsbb (vector unsigned int);
9077 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9078 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9079 vector unsigned long long vec_parity_lsbb (vector signed long long);
9080 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9081 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9082 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9083 * doc/extend.texi: Update the built-in documentation file for the
9084 new built-in functions.
9086 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9088 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9089 (layout::m_primary_loc): New field.
9090 (layout::layout): Initialize new field. Move location filtering
9091 logic from here to...
9092 (layout::maybe_add_location_range): ...this new method. Add
9093 support for filtering to just the lines already specified by other
9095 (layout::will_show_line_p): New method.
9096 (gcc_rich_location::add_location_if_nearby): New method.
9097 (selftest::test_add_location_if_nearby): New test function.
9098 (selftest::diagnostic_show_locus_c_tests): Call it.
9099 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9102 2017-07-11 Tom de Vries <tom@codesourcery.com>
9104 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9105 (bb_first_real_insn): New function.
9106 (nvptx_single): Add extra initialization of broadcasted condition
9109 2017-07-11 Nathan Sidwell <nathan@acm.org>
9111 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9113 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9115 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9116 Binutils doc as TEXI will mess them up.
9117 * doc/invoke.texi (AVR Options): Same here.
9119 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9121 * config/sparc/sparc.opt (mfix-ut700): New option.
9122 (mfix-gr712rc): Likewise.
9123 (sparc_fix_b2bst): New variable.
9124 * doc/invoke.texi (SPARC options): Document them.
9125 (ARM options): Fix warnings.
9126 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9127 instructions to prevent sequences that can trigger the store-store
9128 errata for certain LEON3FT processors.
9129 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9130 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9131 * config/sparc/sparc.md (fix_b2bst): New attribute.
9132 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9134 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9137 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9139 (*rsqrtsf2_sse): Ditto.
9141 (div<mode>3): Macroize insn from divdf3 and divsf3
9142 using MODEF mode iterator.
9144 2017-07-10 Martin Sebor <msebor@redhat.com>
9146 PR tree-optimization/80397
9147 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9148 instead of testing for equality to INTEGER_TYPE.
9150 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9152 * config.gcc: Remove uclibc from arc target spec.
9154 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9156 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9158 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9161 * lto-wrapper.c (remove_option): New function.
9162 (merge_and_complain): Merge PIC/PIE options more realistically.
9164 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9166 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9170 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9171 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9172 * config.in: Regenerate.
9173 * configure: Regenerate.
9174 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9175 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9176 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9177 (TARGET_GASISR_PROLOGUES): ...target mask.
9178 * common/config/avr/avr-common.c
9179 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9180 Set -mgas-isr-prologues.
9181 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9182 INSERT_PASS_BEFORE for it.
9183 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9184 * config/avr/avr.c (avr_option_override)
9185 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9186 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9187 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9188 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9189 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9191 (make_avr_pass_pre_proep): New function.
9192 (emit_push_sfr) <treg>: Add argument to function and use it
9194 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9195 and set machine->gasisr.yes.
9196 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9197 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9198 __gcc_isr.n_pushed to .L__stack_usage.
9199 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9200 (avr_asm_final_postscan_insn): ...this new static function.
9201 * config/avr/avr.h (machine_function)
9202 <is_no_gccisr, use_L__stack_usage>: New fields.
9203 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9204 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9205 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9206 (gasisr, *gasisr): New expander and insn.
9207 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9208 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9209 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9211 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9213 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9216 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9218 Move jump-tables out of .text again.
9221 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9222 (ASM_OUTPUT_ADDR_VEC): New function.
9223 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9224 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9225 INSN_ADDRESSes as asm comment.
9226 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9227 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9228 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9229 * config/avr/avr.md (*tablejump): Adjust comment.
9230 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9231 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9233 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9234 (avr_output_addr_vec): New proto.
9235 (avr_log_t) <insn_addresses>: New field.
9237 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9240 * config/i386/i386.c (ix86_function_arg_advance): Set
9241 outgoing_args_on_stack to true if there are outgoing arguments
9243 (ix86_function_arg): Likewise.
9244 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9245 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9246 * config/i386/i386.h (machine_function): Add
9247 outgoing_args_on_stack.
9249 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9251 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9253 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9255 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9257 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9258 (REAL_H): Remove $(MACHMODE_H).
9259 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9261 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9262 $(MACHMODE_H) and double-int.h.
9263 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9265 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9268 2017-07-07 Andrew Pinski <apinski@cavium.com>
9270 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9271 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9273 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9275 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9276 Add warning if GCC was not configured to link against a GLIBC that
9277 exports the hardware capability bits.
9278 (make_resolver_func): Make resolver function private and not a
9279 COMDAT function. Create the name with clone_function_name instead
9280 of make_unique_name.
9283 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9284 correct operand in doing the split.
9286 2017-07-07 Carl Love <cel@us.ibm.com>
9288 * config/rs6000/rs6000-c: Add support for built-in function
9289 vector unsigned short vec_pack_to_short_fp32 (vector float,
9291 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9292 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9293 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9294 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9295 (convert_4f32_8i16): Add define_expand.
9296 * doc/extend.texi: Update the built-in documentation file for the
9297 new built-in function.
9299 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9301 * config/sparc/m8.md: New file.
9302 * config/sparc/sparc.md: Include m8.md.
9304 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9306 * config/sparc/sparc.opt: New option -mvis4b.
9307 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9308 (sparc_option_override): Handle VIS4B.
9309 (enum sparc_builtins): Define
9310 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9311 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9312 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9313 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9314 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9315 (check_constant_argument): New function.
9316 (sparc_vis_init_builtins): Define builtins
9317 __builtin_vis_dictunpack{8,16,32},
9318 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9319 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9320 __builtin_vis_fpcmpde{8,16,32}shl and
9321 __builtin_vis_fpcmpur{8,16,32}shl.
9322 (sparc_expand_builtin): Check that the constant operands to
9323 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9324 constant and in range.
9325 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9327 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9328 (SPARC_IMM5_P): Likewise.
9329 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9330 (enabled): Handle vis4b.
9331 (UNSPEC_DICTUNPACK): New unspec.
9332 (UNSPEC_FPCMPSHL): Likewise.
9333 (UNSPEC_FPUCMPSHL): Likewise.
9334 (UNSPEC_FPCMPDESHL): Likewise.
9335 (UNSPEC_FPCMPURSHL): Likewise.
9336 (cpu_feature): New CPU feature `vis4b'.
9337 (dictunpack{8,16,32}): New insns.
9338 (FPCSMODE): New mode iterator.
9339 (fpcscond): New code iterator.
9340 (fpcsucond): Likewise.
9341 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9342 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9343 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9344 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9345 * config/sparc/constraints.md: Define constraints `q' for unsigned
9346 2-bit integer constants and `t' for unsigned 5-bit integer
9348 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9350 (imm5_operand_dictunpack16): Likewise.
9351 (imm5_operand_dictunpack32): Likewise.
9352 (imm2_operand): Likewise.
9353 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9354 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9355 ditunpack* and fpcmp*shl builtins.
9357 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9359 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9360 * config.in: Add HAVE_AS_SPARC6 define.
9361 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9363 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9365 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9366 (CPP_CPU_SPEC): Handle m8.
9367 (ASM_CPU_SPEC): Likewise.
9368 * config/sparc/sparc-opts.h (enum processor_type): Add
9370 * config/sparc/sparc.c (m8_costs): New struct.
9371 (sparc_option_override): Handle TARGET_CPU_m8.
9372 (sparc32_initialize_trampoline): Likewise.
9373 (sparc64_initialize_trampoline): Likewise.
9374 (sparc_issue_rate): Likewise.
9375 (sparc_register_move_cost): Likewise.
9376 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9377 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9378 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9379 (CPP_CPU_SPEC): Handle M8.
9380 (ASM_CPU_SPEC): Likewise.
9381 (AS_M8_FLAG): Define.
9382 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9383 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9384 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9386 * configure: Regenerate.
9387 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9390 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9392 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9394 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9395 ("*movdi_insn_sp32"): Do not set v3pipe.
9396 ("*movsi_insn"): Likewise.
9397 ("*movdi_insn_sp64"): Likewise.
9398 ("*movsf_insn"): Likewise.
9399 ("*movdf_insn_sp32"): Likewise.
9400 ("*movdf_insn_sp64"): Likewise.
9401 ("*zero_extendsidi2_insn_sp64"): Likewise.
9402 ("*sign_extendsidi2_insn"): Likewise.
9403 ("*mov<VM32:mode>_insn"): Likewise.
9404 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9405 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9406 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9407 ("<vlop:code><VL:mode>3"): Likewise.
9408 ("*not_<vlop:code><VL:mode>3"): Likewise.
9409 ("*nand<VL:mode>_vis"): Likewise.
9410 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9411 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9412 ("one_cmpl<VL:mode>2"): Likewise.
9413 ("faligndata<VM64:mode>_vis"): Likewise.
9414 ("alignaddrsi_vis"): Likewise.
9415 ("alignaddrdi_vis"): Likweise.
9416 ("alignaddrlsi_vis"): Likewise.
9417 ("alignaddrldi_vis"): Likewise.
9418 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9419 ("bmaskdi_vis"): Likewise.
9420 ("bmasksi_vis"): Likewise.
9421 ("bshuffle<VM64:mode>_vis"): Likewise.
9422 ("cmask8<P:mode>_vis"): Likewise.
9423 ("cmask16<P:mode>_vis"): Likewise.
9424 ("cmask32<P:mode>_vis"): Likewise.
9425 ("pdistn<P:mode>_vis"): Likewise.
9426 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9428 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9430 * config/sparc/sparc.md ("subtype"): New insn attribute.
9431 ("*wrgsr_sp64"): Set insn subtype.
9432 ("*rdgsr_sp64"): Likewise.
9433 ("alignaddrsi_vis"): Likewise.
9434 ("alignaddrdi_vis"): Likewise.
9435 ("alignaddrlsi_vis"): Likewise.
9436 ("alignaddrldi_vis"): Likewise.
9437 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9438 ("fexpand_vis"): Likewise.
9439 ("fpmerge_vis"): Likewise.
9440 ("faligndata<VM64:mode>_vis"): Likewise.
9441 ("bshuffle<VM64:mode>_vis"): Likewise.
9442 ("cmask8<P:mode>_vis"): Likewise.
9443 ("cmask16<P:mode>_vis"): Likewise.
9444 ("cmask32<P:mode>_vis"): Likewise.
9445 ("fchksm16_vis"): Likewise.
9446 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9447 ("fmean16_vis"): Likewise.
9448 ("fp<plusminus_insn>64_vis"): Likewise.
9449 ("<plusminus_insn>v8qi3"): Likewise.
9450 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9451 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9452 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9453 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9454 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9455 ("*movqi_insn"): Likewise.
9456 ("*movhi_insn"): Likewise.
9457 ("*movsi_insn"): Likewise.
9458 ("movsi_pic_gotdata_op"): Likewise.
9459 ("*movdi_insn_sp32"): Likewise.
9460 ("*movdi_insn_sp64"): Likewise.
9461 ("movdi_pic_gotdata_op"): Likewise.
9462 ("*movsf_insn"): Likewise.
9463 ("*movdf_insn_sp32"): Likewise.
9464 ("*movdf_insn_sp64"): Likewise.
9465 ("*zero_extendhisi2_insn"): Likewise.
9466 ("*zero_extendqihi2_insn"): Likewise.
9467 ("*zero_extendqisi2_insn"): Likewise.
9468 ("*zero_extendqidi2_insn"): Likewise.
9469 ("*zero_extendhidi2_insn"): Likewise.
9470 ("*zero_extendsidi2_insn_sp64"): Likewise.
9471 ("ldfsr"): Likewise.
9472 ("prefetch_64"): Likewise.
9473 ("prefetch_32"): Likewise.
9474 ("tie_ld32"): Likewise.
9475 ("tie_ld64"): Likewise.
9476 ("*tldo_ldub_sp32"): Likewise.
9477 ("*tldo_ldub1_sp32"): Likewise.
9478 ("*tldo_ldub2_sp32"): Likewise.
9479 ("*tldo_ldub_sp64"): Likewise.
9480 ("*tldo_ldub1_sp64"): Likewise.
9481 ("*tldo_ldub2_sp64"): Likewise.
9482 ("*tldo_ldub3_sp64"): Likewise.
9483 ("*tldo_lduh_sp32"): Likewise.
9484 ("*tldo_lduh1_sp32"): Likewise.
9485 ("*tldo_lduh_sp64"): Likewise.
9486 ("*tldo_lduh1_sp64"): Likewise.
9487 ("*tldo_lduh2_sp64"): Likewise.
9488 ("*tldo_lduw_sp32"): Likewise.
9489 ("*tldo_lduw_sp64"): Likewise.
9490 ("*tldo_lduw1_sp64"): Likewise.
9491 ("*tldo_ldx_sp64"): Likewise.
9492 ("*mov<VM32:mode>_insn"): Likewise.
9493 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9494 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9496 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9498 * config/sparc/sparc.md ("type"): New insn type viscmp.
9499 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9501 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9502 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9503 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9504 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9506 ("n7_vis_logical_11cycle"): Likewise.
9507 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9508 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9509 * config/sparc/niagara.md ("niag_vis"): Likewise.
9510 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9511 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9513 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9515 * config/sparc/sparc.md: New instruction type `bmask'.
9516 (bmaskdi_vis): Use the `bmask' type.
9517 (bmasksi_vis): Likewise.
9518 * config/sparc/ultra3.md (us3_array): Likewise.
9519 * config/sparc/niagara7.md (n7_array): Likewise.
9520 * config/sparc/niagara4.md (n4_array): Likewise.
9521 * config/sparc/niagara2.md (niag2_vis): Likewise.
9522 (niag3_vis): Likewise.
9523 * config/sparc/niagara.md (niag_vis): Likewise.
9525 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9527 * ipa-comdats.c: Remove optimize check from gate.
9528 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9529 for functions not optimized.
9530 (ipa_fn_summary_read): Skip optimize check.
9531 (ipa_fn_summary_write): Likewise.
9532 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9534 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9536 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9537 (check_callers): Give up on caller which is not optimized.
9538 (inline_small_functions): Likewise.
9539 (ipa_inline): Do not give up when not optimizing.
9540 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9541 away unoptimizes cdtors.
9542 (whole_program_function_and_variable_visibility): Do
9543 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9544 * ipa.c (process_references): Do not check optimize.
9545 (symbol_table::remove_unreachable_nodes): Update optimize check.
9546 (set_writeonly_bit): Update optimize check.
9547 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9548 (pass_ipa_single_use::gate): Remove.
9550 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9552 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9553 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9554 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9555 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9556 permute_load, permute_store, adjust_extract, adjust_splat,
9557 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9558 replace_swap_with_copy, dump_swap_insn_table,
9559 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9560 recombine_lvx_pattern, recombine_stvx_pattern,
9561 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9562 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9563 to file rs6000-p8swap.c.
9564 * config/rs6000/rs6000-p8swap.c: New file.
9565 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9566 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9567 and rs6000*-*-* targets.
9569 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9571 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9573 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9575 * lto-wrapper.c (merge_and_complain): Do not merge
9576 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9577 fsigned_zeros, ftrapping_math, fwrapv.
9578 (append_compiler_options): Do not track these options.
9579 (append_linker_options): Likewie
9581 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9583 * cgraphunit.c (cgraph_node::finalize_function): When
9584 !flag_toplevel_reorde set no_reorder flag.
9585 (varpool_node::finalize_decl): Likewise.
9586 (symbol_table::compile): Drop no toplevel reorder path.
9588 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9590 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9591 edges; zero probability is not better than uninitialized.
9593 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9595 * asan.h (asan_sanitize_allocas_p): Declare.
9596 * asan.c (asan_sanitize_allocas_p): New function.
9597 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9598 (handle_builtin_alloca): Likewise.
9599 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9600 if !asan_sanitize_allocas_p.
9601 * params.def (asan-instrument-allocas): Add new option.
9602 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9603 * opts.c (common_handle_option): Disable allocas sanitization for
9606 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9608 * asan.c: Include gimple-fold.h.
9609 (get_last_alloca_addr): New function.
9610 (handle_builtin_stackrestore): Likewise.
9611 (handle_builtin_alloca): Likewise.
9612 (asan_emit_allocas_unpoison): Likewise.
9613 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9614 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9615 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9616 (instrument_builtin_call): Pass gimple iterator to
9617 get_mem_refs_of_builtin_call.
9618 (last_alloca_addr): New global.
9619 * asan.h (asan_emit_allocas_unpoison): Declare.
9620 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9621 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9622 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9623 if function calls alloca.
9624 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9625 * gimple-fold.h (replace_call_with_value): Declare.
9626 * internal-fn.c: Include asan.h.
9627 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9628 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9630 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9632 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9633 (C_SELFTEST_FLAGS): New.
9634 (CPP_SELFTEST_FLAGS): New.
9635 (SELFTEST_DEPS): New, from deps of s-selftest.
9636 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9637 (CPP_SELFTEST_DEPS): New.
9638 (selftest): Add dependency on s-selftest-c++.
9639 (s-selftest): Rename to...
9640 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9641 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9642 than SELFTEST_FLAGS.
9643 (selftest-gdb): Rename to...
9644 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9646 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9647 (selftest-valgrind): Rename to...
9648 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9650 (selftest-valgrind): Reintroduce as an alias for
9651 selftest-c-valgrind.
9652 (s-selftest-c++): New.
9653 (selftest-c++-gdb): New.
9654 (selftest-c++-valgrind): New.
9656 2017-07-06 Olivier Hainque <hainque@adacore.com>
9658 * gcc.c (process_command): When deciding if undefined variables
9659 should be ignored when processing specs, accept "gcc -v" as well.
9661 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9663 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9664 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9666 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9668 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9669 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9670 * config/arm/arm-tables.opt: Regenerate.
9671 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9673 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9675 2017-07-06 Carl Love <cel@us.ibm.com>
9677 * ChangeLog: Clean up from mid air collision
9679 2017-07-06 Carl Love <cel@us.ibm.com>
9681 * config/rs6000/rs6000-c.c: Add support for built-in functions
9682 vector signed int vec_subc (vector signed int, vector signed int);
9683 vector signed __int128 vec_subc (vector signed __int128,
9684 vector signed __int128);
9685 vector unsigned __int128 vec_subc (vector unsigned __int128,
9686 vector unsigned __int128);
9687 vector signed int vec_sube (vector signed int, vector signed int,
9689 vector unsigned int vec_sube (vector unsigned int,
9690 vector unsigned int,
9691 vector unsigned int);
9692 vector signed __int128 vec_sube (vector signed __int128,
9693 vector signed __int128,
9694 vector signed__int128);
9695 vector unsigned __int128 vec_sube (vector unsigned __int128,
9696 vector unsigned __int128,
9697 vector unsigned __int128);
9698 vector signed int vec_subec (vector signed int, vector signed int,
9700 vector unsigned int vec_subec (vector unsigned int,
9701 vector unsigned int,
9702 vector unsigned int);
9703 vector signed __int128 vec_subec (vector signed __int128,
9704 vector signed __int128,
9705 vector signed__int128);
9706 vector unsigned __int128 vec_subec (vector unsigned __int128,
9707 vector unsigned __int128,
9708 vector unsigned __int128);
9709 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9710 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9711 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9712 BU_ALTIVEC_OVERLOAD_X definitions.
9713 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9714 * doc/extend.texi: Update the built-in documentation file for the new
9717 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9720 * diagnostic-show-locus.c (layout::layout): Use start and finish
9721 spelling location for the start and finish of each range.
9722 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9723 Add unused aspect param.
9724 * input.c (expand_location_1): Add "aspect" param, and use it
9725 to access the correct part of the location.
9726 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9728 (expand_location_to_spelling_point): Likewise.
9729 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9730 param, and pass it to expand_location_1.
9732 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9734 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9735 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9736 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9737 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9738 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9739 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9740 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9741 _mm_maskz_getmant_ss): New intrinsics.
9742 (__builtin_ia32_getexpss128_mask): Changed to ...
9743 __builtin_ia32_getexpss128_round ... this.
9744 (__builtin_ia32_getexpsd128_mask): Changed to ...
9745 __builtin_ia32_getexpsd128_round ... this.
9746 * config/i386/i386-builtin-types.def
9747 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9748 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9749 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9750 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9751 __builtin_ia32_getmantss_mask_round): New builtins.
9752 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9753 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9754 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9755 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9756 * config/i386/sse.md
9757 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9758 avx512f_sgetexp<mode><mask_scalar_name>
9759 <round_saeonly_scalar_name> ... this.
9760 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9761 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9762 vgetexp<ssescalarmodesuffix>
9763 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9764 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9765 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9766 avx512f_vgetmant<mode><mask_scalar_name>
9767 <round_saeonly_scalar_name> ... this.
9768 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9769 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9770 vgetmant<ssescalarmodesuffix>
9771 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9772 %0<mask_scalar_operand4>, %1, %2
9773 <round_saeonly_scalar_mask_op4>, %3} ... this.
9774 * config/i386/subst.md (mask_scalar_operand4,
9775 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9776 round_saeonly_scalar_nimm_predicate): New subst attributes.
9778 2017-07-06 Julia Koval <julia.koval@intel.com>
9780 * config/i386/i386.c (ix86_erase_embedded_rounding):
9781 Remove code for old rounding pattern.
9783 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9785 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9787 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9789 * doc/sourcebuild.texi (Test Directives, Variants of
9790 dg-require-support): Add documentation for dg-require-stack-check.
9792 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9794 * config/i386/subst.md (mask_scalar, round_scalar,
9795 round_saeonly_scalar): New meta-templates.
9796 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9797 round_scalar_mask_operand3, round_scalar_mask_op3,
9798 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9799 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9800 round_saeonly_scalar_constraint,
9801 round_saeonly_scalar_prefix): New subst attribute.
9802 * config/i386/sse.md
9803 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9804 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9805 <round_scalar_name> ... this.
9806 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9807 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9808 <round_scalar_name> ... this.
9809 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9810 <sse>_vm<code><mode>3<mask_scalar_name>
9811 <round_saeonly_scalar_name> ... this.
9812 (v<plusminus_mnemonic><ssescalarmodesuffix>
9813 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9814 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9815 v<plusminus_mnemonic><ssescalarmodesuffix>
9816 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9817 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9818 (v<multdiv_mnemonic><ssescalarmodesuffix>
9819 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9820 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9821 v<multdiv_mnemonic><ssescalarmodesuffix>
9822 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9823 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9824 (v<maxmin_float><ssescalarmodesuffix>
9825 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9826 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9827 v<maxmin_float><ssescalarmodesuffix>
9828 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9829 %0<mask_scalar_operand3>, %1, %<iptr>2
9830 <round_saeonly_scalar_mask_op3>} ... this.
9832 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9834 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9835 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9837 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9838 Alan Hayward <alan.hayward@arm.com>
9839 David Sherwood <david.sherwood@arm.com>
9841 * combine.c (simplify_if_then_else): Remove "enum" before
9843 * compare-elim.c (can_eliminate_compare): Likewise.
9844 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9846 (aarch64_lookup_simd_builtin_type): Likewise.
9847 (aarch64_simd_builtin_type): Likewise.
9848 (aarch64_init_simd_builtin_types): Likewise.
9849 (aarch64_simd_expand_args): Likewise.
9850 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9852 (aarch64_reverse_mask): Likewise.
9853 (aarch64_simd_emit_reg_reg_move): Likewise.
9854 (aarch64_gen_adjusted_ldpstp): Likewise.
9855 (aarch64_ccmp_mode_to_code): Likewise.
9856 (aarch64_operands_ok_for_ldpstp): Likewise.
9857 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9858 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9860 (aarch64_min_divisions_for_recip_mul): Likewise.
9861 (aarch64_reassociation_width): Likewise.
9862 (aarch64_get_condition_code_1): Likewise.
9863 (aarch64_simd_emit_reg_reg_move): Likewise.
9864 (aarch64_simd_attr_length_rglist): Likewise.
9865 (aarch64_reverse_mask): Likewise.
9866 (aarch64_operands_ok_for_ldpstp): Likewise.
9867 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9868 (aarch64_gen_adjusted_ldpstp): Likewise.
9869 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9871 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9872 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9873 (arm_lookup_simd_builtin_type): Likewise.
9874 (arm_simd_builtin_type): Likewise.
9875 (arm_init_simd_builtin_types): Likewise.
9876 (arm_expand_builtin_args): Likewise.
9877 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9878 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9879 (ft32_setup_incoming_varargs): Likewise.
9880 (ft32_function_arg): Likewise.
9881 (ft32_function_arg_advance): Likewise.
9882 (ft32_pass_by_reference): Likewise.
9883 (ft32_arg_partial_bytes): Likewise.
9884 (ft32_valid_pointer_mode): Likewise.
9885 (ft32_addr_space_pointer_mode): Likewise.
9886 (ft32_addr_space_legitimate_address_p): Likewise.
9887 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9889 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9890 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9891 (iamcu_alignment): Likewise.
9892 (canonicalize_vector_int_perm): Likewise.
9893 (ix86_noce_conversion_profitable_p): Likewise.
9894 (ix86_mpx_bound_mode): Likewise.
9895 (ix86_operands_ok_for_move_multiple): Likewise.
9896 * config/microblaze/microblaze-protos.h
9897 (microblaze_expand_conditional_branch_reg): Likewise.
9898 * config/microblaze/microblaze.c
9899 (microblaze_expand_conditional_branch_reg): Likewise.
9900 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9902 (rs6000_reassociation_width): Likewise.
9903 (rs6000_invalid_binary_op): Likewise.
9904 (fusion_p9_p): Likewise.
9905 (emit_fusion_p9_load): Likewise.
9906 (emit_fusion_p9_store): Likewise.
9907 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9909 (riscv_hard_regno_mode_ok_p): Likewise.
9910 (riscv_address_insns): Likewise.
9911 (riscv_split_symbol): Likewise.
9912 (riscv_legitimize_move): Likewise.
9913 (riscv_function_value): Likewise.
9914 (riscv_hard_regno_nregs): Likewise.
9915 (riscv_expand_builtin): Likewise.
9916 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9917 (riscv_build_integer): Likewise.
9918 (riscv_split_integer): Likewise.
9919 (riscv_legitimate_constant_p): Likewise.
9920 (riscv_cannot_force_const_mem): Likewise.
9921 (riscv_regno_mode_ok_for_base_p): Likewise.
9922 (riscv_valid_base_register_p): Likewise.
9923 (riscv_valid_offset_p): Likewise.
9924 (riscv_valid_lo_sum_p): Likewise.
9925 (riscv_classify_address): Likewise.
9926 (riscv_legitimate_address_p): Likewise.
9927 (riscv_address_insns): Likewise.
9928 (riscv_load_store_insns): Likewise.
9929 (riscv_force_binary): Likewise.
9930 (riscv_split_symbol): Likewise.
9931 (riscv_force_address): Likewise.
9932 (riscv_legitimize_address): Likewise.
9933 (riscv_move_integer): Likewise.
9934 (riscv_legitimize_const_move): Likewise.
9935 (riscv_legitimize_move): Likewise.
9936 (riscv_address_cost): Likewise.
9937 (riscv_subword): Likewise.
9938 (riscv_output_move): Likewise.
9939 (riscv_canonicalize_int_order_test): Likewise.
9940 (riscv_emit_int_order_test): Likewise.
9941 (riscv_function_arg_boundary): Likewise.
9942 (riscv_pass_mode_in_fpr_p): Likewise.
9943 (riscv_pass_fpr_single): Likewise.
9944 (riscv_pass_fpr_pair): Likewise.
9945 (riscv_get_arg_info): Likewise.
9946 (riscv_function_arg): Likewise.
9947 (riscv_function_arg_advance): Likewise.
9948 (riscv_arg_partial_bytes): Likewise.
9949 (riscv_function_value): Likewise.
9950 (riscv_pass_by_reference): Likewise.
9951 (riscv_setup_incoming_varargs): Likewise.
9952 (riscv_print_operand): Likewise.
9953 (riscv_elf_select_rtx_section): Likewise.
9954 (riscv_save_restore_reg): Likewise.
9955 (riscv_for_each_saved_reg): Likewise.
9956 (riscv_register_move_cost): Likewise.
9957 (riscv_hard_regno_mode_ok_p): Likewise.
9958 (riscv_hard_regno_nregs): Likewise.
9959 (riscv_class_max_nregs): Likewise.
9960 (riscv_memory_move_cost): Likewise.
9961 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9962 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9963 (rl78_addr_space_address_mode): Likewise.
9964 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9966 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9967 (rs6000_reassociation_width): Likewise.
9968 (rs6000_invalid_binary_op): Likewise.
9969 (fusion_p9_p): Likewise.
9970 (emit_fusion_p9_load): Likewise.
9971 (emit_fusion_p9_store): Likewise.
9972 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9973 (ok_for_simple_move_operands): Likewise.
9974 (ok_for_simple_move_strict_operands): Likewise.
9975 (ok_for_simple_arith_logic_operands): Likewise.
9976 (visium_legitimize_reload_address): Likewise.
9977 (visium_select_cc_mode): Likewise.
9978 (output_cbranch): Likewise.
9979 (visium_split_double_move): Likewise.
9980 (visium_expand_copysign): Likewise.
9981 (visium_expand_int_cstore): Likewise.
9982 (visium_expand_fp_cstore): Likewise.
9983 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9984 (visium_function_arg): Likewise.
9985 (visium_function_arg_advance): Likewise.
9986 (visium_libcall_value): Likewise.
9987 (visium_setup_incoming_varargs): Likewise.
9988 (visium_legitimate_constant_p): Likewise.
9989 (visium_legitimate_address_p): Likewise.
9990 (visium_legitimize_address): Likewise.
9991 (visium_secondary_reload): Likewise.
9992 (visium_register_move_cost): Likewise.
9993 (visium_memory_move_cost): Likewise.
9994 (prepare_move_operands): Likewise.
9995 (ok_for_simple_move_operands): Likewise.
9996 (ok_for_simple_move_strict_operands): Likewise.
9997 (ok_for_simple_arith_logic_operands): Likewise.
9998 (visium_function_value_1): Likewise.
9999 (rtx_ok_for_offset_p): Likewise.
10000 (visium_legitimize_reload_address): Likewise.
10001 (visium_split_double_move): Likewise.
10002 (visium_expand_copysign): Likewise.
10003 (visium_expand_int_cstore): Likewise.
10004 (visium_expand_fp_cstore): Likewise.
10005 (visium_split_cstore): Likewise.
10006 (visium_select_cc_mode): Likewise.
10007 (visium_split_cbranch): Likewise.
10008 (output_cbranch): Likewise.
10009 (visium_print_operand_address): Likewise.
10010 * expmed.c (flip_storage_order): Likewise.
10011 * expmed.h (emit_cstore): Likewise.
10012 (flip_storage_order): Likewise.
10013 * genrecog.c (validate_pattern): Likewise.
10014 * hsa-gen.c (gen_hsa_addr): Likewise.
10015 * internal-fn.c (expand_arith_overflow): Likewise.
10016 * ira-color.c (allocno_copy_cost_saving): Likewise.
10017 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10018 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10019 (process_invariant_for_inheritance): Likewise.
10020 * lra-eliminations.c (move_plus_up): Likewise.
10021 * omp-low.c (lower_oacc_reductions): Likewise.
10022 * simplify-rtx.c (simplify_subreg): Likewise.
10023 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10024 (TARGET_CHKP_BOUND_MODE): Likewise..
10025 * targhooks.c (default_chkp_bound_mode): Likewise.
10026 (default_setup_incoming_vararg_bounds): Likewise.
10027 * targhooks.h (default_chkp_bound_mode): Likewise.
10028 (default_setup_incoming_vararg_bounds): Likewise.
10029 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10030 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10031 (have_whole_vector_shift): Likewise.
10032 * tree-vect-stmts.c (vectorizable_load): Likewise.
10033 * doc/tm.texi: Regenerate.
10035 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10037 Graceful degrade if Binutils PR21472 is not available.
10040 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10041 .rodata in flash test fails.
10042 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10043 * confgure: Regenerate.
10044 * config.in: Regenerate.
10045 * config/avr/avr.c (avr_asm_named_section)
10046 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10047 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10048 (avr_asm_init_sections): Same.
10050 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10052 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10053 (fma<VH:mode>4_intrinsic): Likewise.
10054 (*fmsub<VCVTF:mode>4): Likewise.
10055 (*fmsub<VH:mode>4_intrinsic): Likewise.
10057 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10060 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10061 Don't depend on "optimize > 0".
10062 (out_movhi_r_mr, out_movqi_mr_r): Same.
10063 (out_movhi_mr_r, out_movqi_r_mr): Same.
10064 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10065 io_address_operand on "optimize > 0".
10067 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10069 * tree-loop-distribution.c: Add general explanantion on the pass.
10070 (generate_loops_for_partition): Mark distributed loop.
10071 (pg_add_dependence_edges): New parameter. Handle alias data
10072 dependence specially and record it in the parameter if asked.
10073 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10074 (init_partition_graph_vertices, add_partition_graph_edge): New.
10075 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10076 (free_partition_graph_vdata, build_partition_graph): New.
10077 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10078 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10079 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10080 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10081 (version_for_distribution_p, finalize_partitions): New.
10082 (distribute_loop): Handle alias data dependence specially. Factor
10083 out loop fusion code as functions and call these functions.
10085 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10087 * tree-loop-distribution.c (classify_partition): New parameter and
10088 better handle reduction statement.
10089 (rdg_build_partitions): Revise comment.
10090 (distribute_loop): Compute statements in all partitions and pass it
10091 to classify_partition.
10093 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10095 * tree-loop-distribution.c (enum partition_type): New.
10096 (struct partition): New field type.
10097 (partition_merge_into): Add parameter. Update partition type.
10098 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10099 (build_rdg_partition_for_vertex): Compute partition type.
10100 (rdg_build_partitions): Dump partition type.
10101 (distribute_loop): Update calls to partition_merge_into.
10103 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10105 * tree-loop-distribution.c (struct ddr_hasher): New.
10106 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10108 (classify_partition): Call get_data_dependence.
10109 (pg_add_dependence_edges): Ditto.
10110 (distribute_loop): Release data dependence hash table.
10112 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10114 * tree-loop-distribution.c (ref_base_address): Delete.
10115 (similar_memory_accesses): Rename ...
10116 (share_memory_accesses): ... to this. Check if partitions access
10117 the same memory reference.
10118 (distribute_loop): Call share_memory_accesses.
10120 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10122 * tree-loop-distribution.c (struct partition): New field recording
10123 its data reference.
10124 (partition_alloc, partition_free): Init and release data refs.
10125 (partition_merge_into): Merge data refs.
10126 (build_rdg_partition_for_vertex): Collect data refs for partition.
10127 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10129 (distribute_loop): Remve data refs from vertice data of partition
10132 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10134 * tree-loop-distribution.c (params.h): Include header file.
10135 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10136 (datarefs_vec): New global var.
10137 (create_rdg_vertices): Use datarefs_vec directly.
10138 (free_rdg): Don't free data references.
10139 (build_rdg): Update use. Don't free data references.
10140 (distribute_loop): Compute global variable for data references.
10141 Bail out if there are too many data references.
10143 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10145 * tree-loop-distribution.c (loop_nest): New global var.
10146 (build_rdg): Use loop directly, rather than loop nest.
10147 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10149 (distribute_loop): Compute global variable loop nest. Update use.
10151 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10153 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10154 (partition_merge_into): New parameter. Dump reason for fusion.
10155 (distribute_loop): Update use of partition_merge_into.
10157 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10159 * tree-loop-distribution.c (bb_top_order_index): New.
10160 (bb_top_order_index_size, bb_top_order_cmp): New.
10161 (stmts_from_loop): Use topological order.
10162 (pass_loop_distribution::execute): Compute and release topological
10163 order for basic blocks.
10165 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10167 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10170 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10172 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10173 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10174 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10175 * internal-fn.def (LOOP_DIST_ALIAS): New.
10176 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10177 (fold_loop_internal_call): ... this.
10178 (vect_loop_dist_alias_call): New function.
10179 (set_uid_loop_bbs): Call fold_loop_internal_call.
10180 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10183 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10186 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10187 Require dead FLAGS_REG at the beginning of a peephole.
10189 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10192 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10193 arguments in the call to __builtin_ia32_sbb_u32.
10194 (_subborrow_u64): Swap _X and _Y arguments in the call to
10195 __builtin_ia32_sbb_u64.
10197 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10200 * tree-vrp.c (compare_assert_loc): Turn into a function template
10201 with stable template parameter. Only test if a->e is NULL,
10202 !a->e == !b->e has been verified already. Use e == NULL or
10203 e != NULL instead of e or ! e tests. If stable is true, don't use
10204 iterative_hash_expr, on the other side allow a or b or both NULL
10205 and sort the NULLs last.
10206 (process_assert_insertions): Sort using compare_assert_loc<false>
10207 instead of compare_assert_loc, later sort using
10208 compare_assert_loc<true> before calling process_assert_insertions_for
10209 in a loop. Use break instead of continue once seen NULL pointer.
10211 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10213 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10214 Cortex-R7 and Cortex-R8 processors.
10216 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10218 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10219 uninitialized while src is not.
10221 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10223 * common/config/arm/arm-common.c: Adjust include path for
10225 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10226 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10227 (arm-cpu-data.h): Likewise.
10228 (arm-cpu-cdata.h): Likewise.
10229 * config/arm/arm-cpu.h: Delete.
10230 * config/arm/arm-cpu-cdata.h: Delete.
10231 * config/arm/arm-cpu-data.h: Delete.
10233 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10235 * config/arm/arm-cpus.in (cortex-a55): New.
10236 (cortex-a75): Likewise.
10237 (cortex-a75.cortex-a55): Likewise.
10238 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10240 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10241 * config/arm/arm-cpu-cdata.h: Regenerate.
10242 * config/arm/arm-cpu-data.h: Regenerate.
10243 * config/arm/arm-cpu.h: Regenerate.
10244 * config/arm/arm-tables.opt: Regenerate.
10245 * config/arm/arm-tune.md: Regenerate.
10247 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10249 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10251 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10253 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10256 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10258 PR tree-optimization/81292
10259 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10260 full_string_p, also call adjust_related_strinfos if the adjustment
10261 is simple, otherwise invalidate related strinfos.
10263 2017-07-04 Martin Liska <mliska@suse.cz>
10266 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10267 newly created variable as DECL_IGNORED_P.
10269 2017-07-04 Martin Liska <mliska@suse.cz>
10272 * ipa-inline.c (inline_small_functions):
10273 Use xstrdup_for_dump.
10275 2017-07-04 Tom de Vries <tom@codesourcery.com>
10277 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10279 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10282 * config/darwin.c (darwin_function_switched_text_sections):
10285 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10287 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10289 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10291 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10293 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10295 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10296 min_profitable_iters, and th as inclusive lower bounds.
10297 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10298 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10299 for min_profitable_iters and min_profitable_estimate.
10300 (vect_transform_loop): Treat th as an inclusive lower bound.
10301 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10303 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10306 * config/darwin.c (darwin_function_switched_text_sections):
10307 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10308 in two pieces, and suppress the use of buf.
10310 2017-07-03 Nathan Sidwell <nathan@acm.org>
10312 * hash-table.h (hash_table_mod1): Fix indentation.
10314 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10316 PR middle-end/81290
10317 * predict.c (force_edge_cold): Be more careful about propagation
10319 * profile-count.h (profile_probability::guessed,
10320 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10322 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10324 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10326 * doc/invoke.texi (rcpc architecture extension): Document it.
10328 2017-07-03 Richard Biener <rguenther@suse.de>
10330 PR tree-optimization/60510
10331 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10332 the scalar reduction PHI and use it.
10333 (vectorizable_reduction): Properly guard the single_defuse_cycle
10334 path for non-SLP reduction chains where we cannot use it.
10335 Rework reduc_def/index and vector type deduction. Rework
10336 vector operand gathering during reduction op code-gen.
10337 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10338 chains dissolve the chain and leave it to non-SLP reduction
10341 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10343 * tree-data-ref.h (dr_alignment): Declare.
10344 * tree-data-ref.c (dr_alignment): New function.
10345 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10346 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10348 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10350 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10352 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10353 and base_misalignment fields.
10354 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10355 * tree-data-ref.c: Include builtins.h.
10356 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10357 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10358 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10359 * tree-vect-data-refs.c: Include tree-cfg.h.
10360 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10361 fields instead of calculating an alignment here.
10362 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10363 innermost_loop_behavior fields.
10365 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10367 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10369 (DR_STEP_ALIGNMENT): New macro.
10370 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10371 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10372 (create_data_ref): Print it.
10373 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10374 to tell whether the step preserves vector (mis)alignment.
10375 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10376 Move the check for an integer step and generalise to all INTEGER_CST.
10377 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10378 Print the outer step alignment.
10380 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10382 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10383 with offset_alignment.
10384 (DR_ALIGNED_TO): Delete.
10385 (DR_OFFSET_ALIGNMENT): New macro.
10386 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10387 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10388 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10389 (create_data_ref): Likewise.
10390 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10391 (vect_analyze_data_refs): Likewise.
10392 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10393 creating dummy innermost behavior.
10395 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10397 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10398 with a "innermost_loop_behavior *" and refeence tree.
10399 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10400 (create_data_ref): Update call accordingly.
10401 * tree-predcom.c (find_looparound_phi): Likewise.
10403 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10405 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10406 fields with dr_wrt_vec_loop.
10407 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10408 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10409 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10410 (vect_dr_behavior): New function.
10411 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10412 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10413 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10414 track whether the step preserves the misalignment.
10415 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10416 Use vect_dr_behavior.
10417 (vect_setup_realignment): Update call accordingly.
10418 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10419 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10420 call to vect_create_addr_base_for_vector_ref.
10421 (vect_create_cond_for_align_checks): Likewise.
10422 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10423 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10424 (vect_recog_mask_conversion_pattern): Likewise.
10425 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10426 (new_stmt_vec_info): Remove redundant zeroing.
10428 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10430 * common/config/arm/arm-common.c (arm_be8_option): New function.
10431 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10432 (ISA_ARMv6): Add isa_bit_be8.
10433 * config/arm/arm.h (arm_be8_option): Add prototype.
10434 (BE8_SPEC_FUNCTION): New define.
10435 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10436 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10437 (mlittle-endian): Similarly.
10438 (mbe8, mbe32): New options.
10439 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10440 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10442 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10444 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10446 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10448 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10449 (cleanup_tree_cfg_bb): Use it.
10450 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10452 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10454 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10457 * loop-doloop.c (add_test): Update profile.
10459 2017-07-03 Martin Liska <mliska@suse.cz>
10462 * sanopt.c (rewrite_usage_of_param): New function.
10463 (sanitize_rewrite_addressable_params): Likewise.
10464 (pass_sanopt::execute): Call rewrite_usage_of_param.
10466 2017-07-03 Richard Biener <rguenther@suse.de>
10468 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10469 back to using VIEW_CONVERT_EXPR.
10471 2017-07-03 Martin Liska <mliska@suse.cz>
10474 * doc/extend.texi: Document when a resolver function is
10475 generated for target_clones.
10477 2017-07-03 Martin Liska <mliska@suse.cz>
10479 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10480 and shadow memory of auto variables which are subject of
10481 use-after-scope sanitization.
10482 (asan_expand_mark_ifn): Add do set only when is_poison.
10484 2016-07-03 Richard Biener <rguenther@suse.de>
10486 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10488 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10489 (vectorizable_reduction): Perform reduction PHI creation when
10490 visiting a reduction PHI and adjust and simplify code generation
10491 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10492 (vect_transform_loop): Visit reduction PHIs.
10493 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10494 defs into the SLP tree.
10495 (vect_build_slp_tree): Reduction defs terminate the recursion.
10496 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10498 (vect_get_vec_defs_for_stmt_copy): Export.
10499 (vect_get_vec_defs): Likewise.
10500 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10502 (vect_get_vec_defs_for_stmt_copy): Declare.
10503 (vect_get_vec_defs): Likewise.
10505 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10507 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10508 parameter with a "loop" parameter and use it instead of the
10509 loop containing DR_STMT. Don't check simple_iv when doing
10510 BB analysis. Describe the two analysis modes in the comment.
10512 2017-07-03 Tom de Vries <tom@codesourcery.com>
10514 PR tree-optimization/69468
10515 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10516 (find_same_succ_bb): Handle ignore_edge_flags.
10518 2017-07-03 Tom de Vries <tom@codesourcery.com>
10520 PR tree-optimization/81192
10521 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10523 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10525 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10527 2017-07-03 Tom de Vries <tom@codesourcery.com>
10529 PR tree-optimization/81192
10530 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10531 BB_SAME_SUCC (bb) == NULL.
10533 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10535 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10538 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10540 * dumpfile.c: Include profile-count.h
10541 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10543 (insert_cond_bb): Update profile.
10544 * tree-cfg.h (insert_cond_bb): Update prototype.
10545 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10546 * tree-dump.c: Do not include tree-cfg.
10548 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10550 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10552 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10554 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10557 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10559 * tree-complex.c (expand_complex_div_wide): update profile.
10561 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10562 Alan Hayward <alan.hayward@arm.com>
10563 David Sherwood <david.sherwood@arm.com>
10565 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10566 (CORETYPES_H): New define.
10567 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10568 (insn-modes-inline.h, s-modes-inline-h): New rules.
10569 (generated_files): Add insn-modes-inline.h.
10570 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10571 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10572 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10573 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10574 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10575 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10576 (build/gencodes.o, build/genconditions.o): Likewise.
10577 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10578 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10579 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10580 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10581 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10582 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10583 * coretypes.h: Include everything up to real.h for generators.
10584 Include insn-modes.h first. Include wide-int-print.h after
10585 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10586 * machmode.h: Don't include insn-modes.h here.
10587 * function-tests.c: Remove includes of signop.h, machmode.h,
10588 double-int.h and wide-int.h.
10590 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10592 * optc-save-gen.awk: Likewise.
10593 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10594 * godump.c: Remove include of wide-int-print.h.
10595 * pretty-print.h: Likewise.
10596 * wide-int-print.cc: Likewise.
10597 * wide-int.cc: Likewise.
10598 * hash-map-tests.c: Remove include of signop.h.
10599 * hash-set-tests.c: Likewise.
10600 * rtl-tests.c: Likewise.
10601 * mkconfig.sh: Remove include of machmode.h.
10602 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10604 (emit_insn_modes_inline_h): ...this new function. Emit the code
10605 into an insn-modes-inline.h header file, adding appropriate
10606 include guards and end comments.
10607 (emit_insn_modes_c_header): Remove include of machmode.h.
10608 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10610 (main): Handle -i flag and call emit_insn_modes_inline_h when
10613 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10615 * tree-ssa-strlen.c (strinfo): Rename the length field to
10616 nonzero_chars. Add a full_string_p field.
10617 (compare_nonzero_chars, zero_length_string_p): New functions.
10618 (get_addr_stridx): Add an offset_out parameter.
10619 Use compare_nonzero_chars.
10620 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10621 (new_strinfo): Update after above changes to strinfo.
10622 (set_endptr_and_length): Set full_string_p.
10623 (get_string_length): Update after above changes to strinfo.
10624 (unshare_strinfo): Update call to new_strinfo.
10625 (maybe_invalidate): Likewise.
10626 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10627 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10628 as a uhwi instead of an shwi. Update after above changes to
10629 strinfo and new_strinfo.
10630 (zero_length_string): Assert that chainsi contains full strings.
10631 Use zero_length_string_p. Update call to new_strinfo.
10632 (adjust_related_strinfos): Update after above changes to strinfo.
10633 Copy full_string_p from origsi.
10634 (adjust_last_stmt): Use zero_length_string_p.
10635 (handle_builtin_strlen): Update after above changes to strinfo and
10636 new_strinfo. Install the lhs as the string length if the previous
10637 entry didn't describe a full string.
10638 (handle_builtin_strchr): Update after above changes to strinfo
10640 (handle_builtin_strcpy): Likewise.
10641 (handle_builtin_strcat): Likewise.
10642 (handle_builtin_malloc): Likewise.
10643 (handle_pointer_plus): Likewise.
10644 (handle_builtin_memcpy): Likewise. Track nonzero characters
10645 that aren't necessarily followed by a nul terminator.
10646 (handle_char_store): Likewise.
10648 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10650 PR tree-optimization/80769
10651 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10652 for malloc and calloc. Document the new invariant that all related
10653 strinfos have delayed lengths or none do.
10654 (verify_related_strinfos): Move earlier in file.
10655 (set_endptr_and_length): New function, split out from...
10656 (get_string_length): ...here. Also set the lengths of related
10658 (zero_length_string): Assert that chainsi has known (rather than
10660 (adjust_related_strinfos): Likewise.
10662 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10664 PR tree-optimization/81136
10665 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10666 assert that two references with the same misalignment have the same
10667 compile-time misalignment if those compile-time misalignments
10670 2017-07-01 Andi Kleen <ak@linux.intel.com>
10672 * print-tree.c (print_node): Print all attributes.
10674 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10676 * cfg.c (scale_bbs_frequencies): New function.
10677 * cfg.h (scale_bbs_frequencies): Declare it.
10678 * cfgloopanal.c (single_likely_exit): Cleanup.
10679 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10681 (scale_loop_profile): Likewise.
10682 (loop_version): Likewise.
10683 (create_empty_loop_on_edge): Update.
10684 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10685 scale_loop_frequencies, scale_loop_profile, loopify,
10686 loop_version): Update prototypes.
10687 * modulo-sched.c (sms_schedule): Update.
10688 * predict.c (unlikely_executed_edge_p): Also check probability.
10689 (probably_never_executed_edge_p): Fix typo.
10690 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10691 * tree-parloops.c (gen_parallel_loop): Update.
10692 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10693 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10694 * tree-ssa-loop-split.c (split_loop): Update.
10695 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10696 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10697 (vect_loop_versioning): Update.
10698 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10700 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10702 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10704 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10706 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10707 to keep profile consistent.
10709 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10711 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10712 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10713 * profile-count.h (max_safe_multiplier): Make unsigned.
10714 (profile_count::guessed_zero): New.
10716 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10718 * bb-reorder.c (fix_up_crossing_landing_pad,
10719 fix_crossing_conditional_branches): Use make_single_succ_edge
10720 to keep profile consistent.
10722 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10724 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10727 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10730 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10731 the right scopes, make sure cond_jump isn't preserved between multiple
10732 iterations. Search for fallthru edge whenever there are 3+ edges and
10733 use find_fallthru_edge for it.
10735 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10737 Patch by Alexander Monakov <amonakov@ispras.ru>
10738 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10739 probabilities consistently.
10741 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10743 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10744 emit_cmp_and_jump_insns.
10746 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10749 * tree-inline.c (expand_call_inline): Combine profile statuses.
10751 2017-06-30 Andrew Pinski <apinski@cavium.com>
10753 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10754 fold_stmt returned true.
10756 2017-06-30 Nathan Sidwell <nathan@acm.org>
10758 * ggc.h (empty_string): Delete.
10759 * cfgexpand.c (expand_asm_stmt): Use plain "".
10760 * optabs.c (expand_asm_memory_barrier): Likewise.
10761 * stringpool.c (empty_string): Delete.
10762 (digit_vector, digit_string): Delete.
10763 (ggc_alloc_string): Use plain "", don't optimize single digit
10764 strings. Use ggc_alloc_atomic.
10766 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10768 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10769 comparison set and one other set, use the cost of the non-comparison
10772 2017-06-30 Nathan Sidwell <nathan@acm.org>
10774 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10777 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10779 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10780 loops. Remove now unneeded calls to gimple_switch_set_label() that
10781 just set removed labels to NULL_TREE.
10783 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10785 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10786 (set_range_info): ...here. Only call set_range_info_raw if domain
10788 (set_nonzero_bits): Call set_range_info_raw.
10789 * tree-ssanames.h (set_range_info_raw): New.
10791 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10794 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10795 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10796 of nonimmediate_operand and <store_mask_constraint> instead of m
10797 for the input operand. For V8FI iterator, always split if input
10798 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10799 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10800 <store_mask_predicate> instead of register_operand and
10801 <store_mask_constraint> instead of v for the input operand. Make
10802 sure both operands aren't MEMs for if not <mask_applied>.
10804 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10806 * lto-wrapper.c (copy_file) Close both file descriptors before
10809 2017-06-30 Martin Liska <mliska@suse.cz>
10812 * multiple_target.c (create_dispatcher_calls): Make ifunc
10813 also for function that don't have calls or are not referenced.
10815 2017-06-30 Richard Biener <rguenther@suse.de>
10817 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10818 analyze the first scalar stmt. Move vector type computation
10819 for the BB case here from ...
10820 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10821 live operation processing in the SLP case properly.
10823 2017-06-30 Richard Biener <rguenther@suse.de>
10825 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10827 2017-06-30 Martin Liska <mliska@suse.cz>
10830 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10831 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10833 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10835 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10838 2017-06-30 Richard Biener <rguenther@suse.de>
10840 PR tree-optimization/81249
10841 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10842 condition reduction result to original scalar type.
10844 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10846 * profile-count.h (enum profile_quality): Fix typos and whitespace
10849 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10851 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10852 type for branch probabilities.
10854 2017-06-29 Julian Brown <julian@codesourcery.com>
10855 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10857 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10858 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10859 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10860 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10862 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10864 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10865 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10866 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10867 CC usage from generic code to here.
10868 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10869 CC usage into the target macros.
10871 2017-06-29 Maya Rashish <coypu@sdf.org>
10873 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10876 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10878 * arm/arm-builtins.c: Include profile-count.h
10879 * except.c (sjlj_emit_function_enter): Use
10880 profile_probability::unlikely.
10882 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10884 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10885 and tocrel_offset be pointer args rather than implicitly using
10887 (legitimate_constant_pool_address_p, rs6000_emit_move,
10888 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10889 tocrel_offset and use in toc_relative_expr_p call.
10890 (print_operand, print_operand_address): Use static tocrel_base_oac
10891 and tocrel_offset_oac.
10892 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10895 2017-06-29 Maya Rashish <coypu@sdf.org>
10897 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10899 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10901 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10902 objects, take into account only the alignment of 'op0' and 'mode1' if
10905 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10907 * ccmp.c (ccmp_tree_comparison_p): New function.
10908 (ccmp_candidate_p): Update to use above function.
10909 (get_compare_parts): New function.
10910 (expand_ccmp_next): Update to use new functions.
10911 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10913 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10914 take mode as argument.
10915 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10916 * expr.c (expand_expr_real_1): Pass mode as argument.
10918 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10920 * combine.c (combine_instructions): Print insns to dump_file, together
10923 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10925 * asan.c (asan_emit_stack_protection): Update.
10926 (create_cond_insert_point): Update.
10927 * auto-profile.c (afdo_propagate_circuit): Update.
10928 * basic-block.h (struct edge_def): Turn probability to
10929 profile_probability.
10930 (EDGE_FREQUENCY): Update.
10931 * bb-reorder.c (find_traces_1_round): Update.
10932 (better_edge_p): Update.
10933 (sanitize_hot_paths): Update.
10934 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10935 (make_single_succ_edge): Update.
10936 (check_bb_profile): Update.
10937 (dump_edge_info): Update.
10938 (update_bb_profile_for_threading): Update.
10939 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10940 probabilitycount to 0.
10941 * cfgbuild.c (compute_outgoing_frequencies): Update.
10942 * cfgcleanup.c (try_forward_edges): Update.
10943 (outgoing_edges_match): Update.
10944 (try_crossjump_to_edge): Update.
10945 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10946 (expand_gimple_tailcall): Update.
10947 (construct_init_block): Use make_single_succ_edge.
10948 (construct_exit_block): Use make_single_succ_edge.
10949 * cfghooks.c (verify_flow_info): Update.
10950 (redirect_edge_succ_nodup): Update.
10951 (split_edge): Update.
10952 (account_profile_record): Update.
10953 * cfgloopanal.c (single_likely_exit): Update.
10954 * cfgloopmanip.c (scale_loop_profile): Update.
10955 (set_zero_probability): Remove.
10956 (duplicate_loop_to_header_edge): Update.
10957 * cfgloopmanip.h (loop_version): Update prototype.
10958 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10959 (force_nonfallthru_and_redirect): Update.
10960 (update_br_prob_note): Update.
10961 (rtl_verify_edges): Update.
10962 (purge_dead_edges): Update.
10963 (rtl_lv_add_condition_to_bb): Update.
10964 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10965 * cgraphunit.c (init_lowered_empty_function): Update.
10966 (cgraph_node::expand_thunk): Update.
10967 * cilk-common.c: Include profile-count.h
10968 * dojump.c (inv): Remove.
10969 (jumpifnot): Update.
10970 (jumpifnot_1): Update.
10971 (do_jump_1): Update.
10973 (do_jump_by_parts_greater_rtx): Update.
10974 (do_compare_rtx_and_jump): Update.
10975 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10976 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10977 * dwarf2cfi.c: Include profile-count.h
10978 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10979 (sjlj_emit_dispatch_table): Likewise.
10980 * explow.c: Include profile-count.h
10981 * expmed.c (emit_store_flag_force): Update.
10982 (do_cmp_and_jump): Update.
10983 * expr.c (compare_by_pieces_d::generate): Update.
10984 (compare_by_pieces_d::finish_mode): Update.
10985 (emit_block_move_via_loop): Update.
10986 (store_expr_with_bounds): Update.
10987 (store_constructor): Update.
10988 (expand_expr_real_2): Update.
10989 (expand_expr_real_1): Update.
10990 * expr.h (try_casesi, try_tablejump): Update prototypes.
10991 * gimple-pretty-print.c (dump_probability): Update.
10992 (dump_profile): New.
10993 (dump_gimple_label): Update.
10994 (dump_gimple_bb_header): Update.
10995 * graph.c (draw_cfg_node_succ_edges): Update.
10996 * hsa-gen.c (convert_switch_statements): Update.
10997 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10998 (find_if_case_1): Update.
10999 (find_if_case_2): Update.
11000 * internal-fn.c (expand_arith_overflow_result_store): Update.
11001 (expand_addsub_overflow): Update.
11002 (expand_neg_overflow): Update.
11003 (expand_mul_overflow): Update.
11004 (expand_vector_ubsan_overflow): Update.
11005 * ipa-cp.c (good_cloning_opportunity_p): Update.
11006 * ipa-split.c (split_function): Use make_single_succ_edge.
11007 * ipa-utils.c (ipa_merge_profiles): Update.
11008 * loop-doloop.c (add_test): Update.
11009 (doloop_modify): Update.
11010 * loop-unroll.c (compare_and_jump_seq): Update.
11011 (unroll_loop_runtime_iterations): Update.
11012 * lra-constraints.c (lra_inheritance): Update.
11013 * lto-streamer-in.c (input_cfg): Update.
11014 * lto-streamer-out.c (output_cfg): Update.
11015 * mcf.c (adjust_cfg_counts): Update.
11016 * modulo-sched.c (sms_schedule): Update.
11017 * omp-expand.c (expand_omp_for_init_counts): Update.
11018 (extract_omp_for_update_vars): Update.
11019 (expand_omp_ordered_sink): Update.
11020 (expand_omp_for_ordered_loops): Update.
11021 (expand_omp_for_generic): Update.
11022 (expand_omp_for_static_nochunk): Update.
11023 (expand_omp_for_static_chunk): Update.
11024 (expand_cilk_for): Update.
11025 (expand_omp_simd): Update.
11026 (expand_omp_taskloop_for_outer): Update.
11027 (expand_omp_taskloop_for_inner): Update.
11028 * omp-simd-clone.c (simd_clone_adjust): Update.
11029 * optabs.c (expand_doubleword_shift): Update.
11030 (expand_abs): Update.
11031 (emit_cmp_and_jump_insn_1): Update.
11032 (expand_compare_and_swap_loop): Update.
11033 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11034 * predict.c (predictable_edge_p): Update.
11035 (edge_probability_reliable_p): Update.
11036 (set_even_probabilities): Update.
11037 (combine_predictions_for_insn): Update.
11038 (combine_predictions_for_bb): Update.
11039 (propagate_freq): Update.
11040 (estimate_bb_frequencies): Update.
11041 (force_edge_cold): Update.
11042 * profile-count.c (profile_count::dump): Add missing space into dump.
11043 (profile_count::debug): Add newline.
11044 (profile_count::differs_from_p): Explicitly convert to unsigned.
11045 (profile_count::stream_in): Update.
11046 (profile_probability::dump): New member function.
11047 (profile_probability::debug): New member function.
11048 (profile_probability::differs_from_p): New member function.
11049 (profile_probability::differs_lot_from_p): New member function.
11050 (profile_probability::stream_in): New member function.
11051 (profile_probability::stream_out): New member function.
11052 * profile-count.h (profile_count_quality): Rename to ...
11053 (profile_quality): ... this one.
11054 (profile_probability): New.
11055 (profile_count): Update.
11056 * profile.c (compute_branch_probabilities): Update.
11057 * recog.c (peep2_attempt): Update.
11058 * sched-ebb.c (schedule_ebbs): Update.
11059 * sched-rgn.c (find_single_block_region): Update.
11060 (compute_dom_prob_ps): Update.
11061 (schedule_region): Update.
11062 * sel-sched-ir.c (compute_succs_info): Update.
11063 * stmt.c (struct case_node): Update.
11064 (do_jump_if_equal): Update.
11065 (get_outgoing_edge_probs): Update.
11066 (conditional_probability): Update.
11067 (emit_case_dispatch_table): Update.
11068 (expand_case): Update.
11069 (expand_sjlj_dispatch_table): Update.
11070 (emit_case_nodes): Update.
11071 * targhooks.c: Update.
11072 * tracer.c (better_p): Update.
11073 (find_best_successor): Update.
11074 * trans-mem.c (expand_transaction): Update.
11075 * tree-call-cdce.c: Update.
11076 * tree-cfg.c (gimple_split_edge): Upate.
11077 (move_sese_region_to_fn): Upate.
11078 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11079 * tree-eh.c (lower_resx): Upate.
11080 (cleanup_empty_eh_move_lp): Upate.
11081 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11082 * tree-inline.c (copy_edges_for_bb): Update.
11083 (copy_cfg_body): Update.
11084 * tree-parloops.c (gen_parallel_loop): Update.
11085 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11086 (gimple_gen_time_profiler): Update.
11087 * tree-ssa-dce.c (remove_dead_stmt): Update.
11088 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11089 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11090 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11091 (unloop_loops): Update.
11092 (try_peel_loop): Update.
11093 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11094 * tree-ssa-loop-split.c (connect_loops): Update.
11095 (split_loop): Update.
11096 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11097 (hoist_guard): Update.
11098 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11099 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11100 (value_replacement): Update.
11101 * tree-ssa-reassoc.c (branch_fixup): Update.
11102 * tree-ssa-tail-merge.c (replace_block_by): Update.
11103 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11104 (create_edge_and_update_destination_phis): Update.
11105 (compute_path_counts): Update.
11106 (recompute_probabilities): Update.
11107 (update_joiner_offpath_counts): Update.
11108 (freqs_to_counts_path): Update.
11109 (duplicate_thread_path): Update.
11110 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11111 (struct switch_conv_info): Update.
11112 (gen_inbound_check): Update.
11113 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11114 (vect_do_peeling): Update.
11115 (vect_loop_versioning): Update.
11116 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11117 (optimize_mask_stores): Update.
11118 * ubsan.c (ubsan_expand_null_ifn): Update.
11119 * value-prof.c (gimple_divmod_fixed_value): Update.
11120 (gimple_divmod_fixed_value_transform): Update.
11121 (gimple_mod_pow2): Update.
11122 (gimple_mod_pow2_value_transform): Update.
11123 (gimple_mod_subtract): Update.
11124 (gimple_mod_subtract_transform): Update.
11125 (gimple_ic): Update.
11126 (gimple_stringop_fixed_value): Update.
11127 (gimple_stringops_transform): Update.
11128 * value-prof.h: Update.
11130 2017-06-29 Carl Love <cel@us.ibm.com>
11132 * config/rs6000/rs6000-c.c: Add support for built-in functions
11133 vector signed int vec_signed (vector float);
11134 vector signed long long vec_signed (vector double);
11135 vector signed int vec_signed2 (vector double, vector double);
11136 vector signed int vec_signede (vector double);
11137 vector signed int vec_signedo (vector double);
11138 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11139 instruction generator.
11140 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11141 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11142 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11144 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11145 vunsignede_v2df): Add define_expands.
11146 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11147 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11148 VEC_UNSIGNEDO): Add definitions.
11149 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11150 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11151 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11152 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11153 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11154 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11155 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11156 vec_unsignede, vec_unsignedo): Add builtin defines.
11157 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11159 * doc/extend.texi: Update the built-in documentation file for the
11160 new built-in functions.
11162 2017-06-29 Richard Biener <rguenther@suse.de>
11164 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11165 reduction chains to LOOP_VINFO_REDUCTIONS.
11166 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11167 SLP reductions after processing reduction chains.
11169 2017-06-29 Nathan Sidwell <nathan@acm.org>
11171 * builtins.c (fold_builtin_FUNCTION): Use
11172 lang_hooks.decl_printable_name.
11174 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11176 PR middle-end/81194
11177 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11178 with only one label.
11179 * stmt.c (expand_case): Assert NCASES is greater than one.
11181 2017-06-29 Richard Biener <rguenther@suse.de>
11183 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11185 (group_case_labels): Likewise.
11186 (find_taken_edge): Push sanity checking on val to workers...
11187 (find_taken_edge_cond_expr): ... here
11188 (find_taken_edge_switch_expr): ... and here, handle cases
11189 with just a default label.
11190 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11191 (group_case_labels): Likewise.
11192 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11193 group_case_labels does anything cleanup the CFG again.
11195 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11197 PR tree-optimization/81196
11198 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11199 exit condition comparing two IVs.
11201 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11203 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11204 profile to the dummy entry at the end of the list of architectures.
11205 * config/arm/arm-cpu-cdata.h: Regenerated.
11207 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11208 Michael Collison <michael.collison@arm.com>
11211 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11213 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11214 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11215 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11216 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11217 of shift when the shift amount is masked with constant equal to
11218 the size of the mode.
11219 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11222 2017-06-29 Martin Liska <mliska@suse.cz>
11224 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11226 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11229 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11232 (edge_growth_cache_entry::edge_growth_cache_entry): New
11234 (reset_edge_growth_cache): Update to use constructor.
11236 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11238 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11239 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11240 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11242 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11244 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11245 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11247 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11249 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11250 (*-linux-uclibc*): Add t-uclibc tmake_file.
11251 * config/t-musl: New.
11252 * config/t-uclibc: New.
11254 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11256 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11258 (gen_comm_data): Emit architectural setting of arch_prof.
11259 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11261 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11262 (armv8-m.base, armv8-m.main): Likewise.
11263 * arm-protos.h (arm_build_target): Add profile field.
11264 (arch_option): Likewise.
11265 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11267 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11268 arm_active_target.profile.
11270 2017-06-28 Richard Biener <rguenther@suse.de>
11272 PR middle-end/81227
11273 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11274 TYPE_OVERFLOW_WRAPS.
11275 * match.pd (negate_expr_p): Likewise.
11276 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11277 fold_build2, not fold_binary.
11279 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11281 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11282 Convert memory address to Pmode.
11283 (aarch64_print_operand): Assert MEM operands are always Pmode.
11285 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11288 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11289 Remove redundant if.
11290 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11291 * config/arm/aarch-common-protos.h
11292 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11293 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11295 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11298 * multiple_target.c (create_dispatcher_calls): Set the default
11299 clone to be static, not public.
11301 2017-06-28 Richard Biener <rguenther@suse.de>
11303 * tree-vect-loop.c (vectorizable_reduction): Move special
11304 cond reduction IV var creation ...
11305 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11306 parameter. Use STMT_VINFO_VECTYPE.
11307 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11310 2017-06-28 Martin Liska <mliska@suse.cz>
11313 * ipa-visibility.c (non_local_p): Handle visibility.
11315 2017-06-28 Martin Liska <mliska@suse.cz>
11318 * common.opt: Add IntegerRange to various options.
11319 * opt-functions.awk (integer_range_info): New function.
11320 * optc-gen.awk: Add integer_range_info to cl_options struct.
11321 * opts-common.c (decode_cmdline_option): Handle
11322 CL_ERR_INT_RANGE_ARG.
11323 (cmdline_handle_error): Likewise.
11324 * opts.c (print_filtered_help): Show valid interval in
11325 when --help is provided.
11326 * opts.h (struct cl_option): Add range_min and range_max fields.
11327 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11329 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11331 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11332 (x * C EQ/NE y * C): New transformation.
11334 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11336 * genmultilib (combination_space): Accept '+' in option names.
11338 2017-06-28 Martin Liska <mliska@suse.cz>
11341 * asan.c (instrument_derefs): Bail out inner references
11342 that are hard register variables.
11344 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11347 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11348 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11350 2017-06-28 Richard Biener <rguenther@suse.de>
11352 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11353 (vect_get_slp_defs): Adjust.
11354 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11356 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11358 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11359 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11360 (vectorizable_reduction): Adjust.
11361 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11363 (vect_get_slp_defs): Likewise.
11364 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11365 (vectorizable_bswap): Adjust.
11366 (vectorizable_call): Likewise.
11367 (vectorizable_conversion): Likewise.
11368 (vectorizable_assignment): Likewise.
11369 (vectorizable_shift): Likewise.
11370 (vectorizable_operation): Likewise.
11371 (vectorizable_store): Likewise.
11372 (vectorizable_condition): Likewise.
11373 (vectorizable_comparison): Likewise.
11375 2017-06-28 Michael Collison <michael.collison@arm.com>
11378 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11380 (arm_gen_movmemqi): Removed unused variable 'i'.
11381 Convert 'for' loop into 'while' loop.
11382 (arm_expand_prologue): Remove last unnecessary set of insn.
11383 (thumb_pop): Remove unused variable 'pushed_words'.
11384 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11386 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11388 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11389 * config/s390/s390-protos.h: Add prototype of
11390 s390_rel_address_ok_p.
11391 * config/s390/s390.c (s390_got_symbol): New function.
11392 (s390_rel_address_ok_p): New function.
11393 (legitimize_pic_address): Use s390_rel_address_ok_p.
11394 (s390_load_got): Use s390_got_symbol.
11395 (s390_option_override): Issue error if
11396 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11397 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11399 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11401 2017-06-27 Andrew Pinski <apinski@cavium.com>
11403 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11404 (X * copysign (1.0, X)): New pattern.
11405 (X * copysign (1.0, -X)): New pattern.
11406 (copysign (-1.0, CST)): New pattern.
11408 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11410 * genmultilib (combination_space): Remove variable.
11411 Validate reuse rules against regular expression for any sequence
11412 of multilib options in any order.
11414 2017-06-27 Michael Collison <michael.collison@arm.com>
11416 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11417 call aarch64_split_simd_combine.
11418 * (aarch64_combine_internal<mode>): Delete pattern.
11419 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11420 Allow register and subreg operands.
11422 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11424 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11425 specific need, just fallback on defaults.
11426 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11428 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11429 Olivier Hainque <hainque@adacore.com>
11431 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11433 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11434 targets. Pick a default if no particular attempt applied.
11435 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11438 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11440 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11441 (x86_64-wrs-vxworks7): Likewise.
11443 2017-06-27 Marek Polacek <polacek@redhat.com>
11446 * ubsan.c (instrument_null): Check get_base_address's result for null.
11448 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11450 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11452 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11454 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11455 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11456 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11457 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11458 New function types.
11459 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11460 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11461 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11462 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11463 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11464 BUILT_IN_FEUPDATEENV): New builtins.
11465 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11466 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11467 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11468 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11470 (builtin_structptr_types): Adjust size.
11471 * tree.c (builtin_structptr_types): Add four entries.
11473 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11474 Olivier Hainque <hainque@adacore.com>
11476 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11477 (TLS_SYM): New local macro, forcing reference to __tls__ on
11478 link command lines for VxWorks 7 RTPs, triggering initialization
11480 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11481 OS features TLS support, true for RTPs on VxWorks 7.
11482 * config/vxworks.c (vxworks_override_options): Setup emutls
11485 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11487 * predict.c (test_prediction_value_range): Use -1U instead of -1
11488 to avoid narrowing conversion warning.
11489 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11490 to avoid narrowing conversion warning.
11491 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11493 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11495 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11497 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11498 64bit configurations.
11499 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11500 (SIZE_TYPE): Likewise.
11501 * config/vxworks.c (vxworks_emutls_var_fields): Use
11502 long_unsigned_type_node instead of unsigned_type_node as the offset
11503 field type, which is "pointer" mode in emutls.c.
11505 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11508 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11510 PR middle-end/81207
11511 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11512 gimple_vuse copying separately from gimple_vdef copying.
11514 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11516 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11519 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11520 Olivier Hainque <hainque@adacore.com>
11522 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11523 for all vxworks7 targets.
11524 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11525 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11526 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11527 variations for VX6/VX7 and 32/64bits later on in ...
11528 (VXWORKS_LIB_SPEC): Leverage new macros.
11529 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11530 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11532 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11534 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11535 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11537 2017-06-26 Carl Love <cel@us.ibm.com>
11539 * config/rs6000/rs6000-c.c: Add support for built-in functions
11540 vector bool char vec_reve (vector bool char);
11541 vector signed char vec_reve (vector signed char);
11542 vector unsigned char vec_reve (vector unsigned char);
11543 vector bool int vec_reve (vector bool int);
11544 vector signed int vec_reve (vector signed int);
11545 vector unsigned int vec_reve (vector unsigned int);
11546 vector bool long long vec_reve (vector bool long long);
11547 vector signed long long vec_reve (vector signed long long);
11548 vector unsigned long long vec_reve (vector unsigned long long);
11549 vector bool short vec_reve (vector bool short);
11550 vector signed short vec_reve (vector signed short);
11551 vector double vec_reve (vector double);
11552 vector float vec_reve (vector float);
11553 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11554 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11555 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11556 (altivec_vreve): New pattern.
11557 * config/rs6000/altivec.h (vec_reve): New define.
11558 * doc/extend.texi (vec_rev): Update the built-in documentation file
11559 for the new built-in functions.
11561 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11563 PR tree-optimization/71815
11564 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11566 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11568 (slsr_process_phi): Likewise.
11569 (replace_uncond_cands_and_profitable_phis): Don't replace a
11570 multiply candidate with a stride of 1 (copy or cast).
11571 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11573 (lowest_cost_path): Likewise.
11574 (total_savings): Likewise.
11576 2017-06-26 Richard Biener <rguenther@suse.de>
11579 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11580 Use def_builtin_pure for all gather builtins.
11582 2017-06-26 Richard Biener <rguenther@suse.de>
11584 PR tree-optimization/81203
11585 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11586 non-dominating BBs.
11588 2017-06-26 Marek Polacek <polacek@redhat.com>
11591 * doc/invoke.texi: Document -Wmultistatement-macros.
11593 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11595 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11596 arm_neon_ok_no_float_abi effective target.
11598 2017-06-26 Richard Biener <rguenther@suse.de>
11600 PR tree-optimization/80928
11601 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11602 (copy_bbs): Set BB_DUPLICATED flag early.
11603 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11605 (execute_on_shrinking_pred): Likewise.
11606 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11607 BB_DUPLICATED blocks.
11608 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11609 iterate over all PHIs considering removal of *gsi.
11611 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11613 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11616 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11618 * config/rs6000/rs6000-string.c: (expand_block_clear,
11619 do_load_for_compare, select_block_compare_mode,
11620 compute_current_alignment, expand_block_compare,
11621 expand_strncmp_align_check, expand_strn_compare,
11622 expand_block_move, rs6000_output_load_multiple)
11623 Move functions related to string/block move/compare
11624 to a separate file.
11625 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11626 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11627 for this function which is now used in two files.
11628 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11629 * config.gcc: Add rs6000-string.o to extra_objs for
11630 targets powerpc*-*-* and rs6000*-*-*.
11632 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11635 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11636 32-bit, since indexed is not valid for DImode.
11637 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11638 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11639 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11640 (define_peephole2 for Altivec d-form store): Likewise.
11643 * multiple_target.c (create_dispatcher_calls): Only create the
11644 dispatcher call if the function is the default clone of a
11645 versioned function.
11647 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11649 PR middle-end/80902
11650 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11651 a call, force the call to not be a tail call.
11653 2017-06-23 Jeff Law <law@redhat.com>
11655 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11658 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11660 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11661 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11662 handling for early expansion of vector shifts (sl,sr,sra,rl).
11663 (builtin_function_type): Add vector shift right instructions
11664 to the unsigned argument list.
11666 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11668 rtl-optimizatoin/79286
11669 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11670 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11671 trap. PIC register plus a const unspec without offset can never trap.
11673 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11675 * tree.h (builtin_structptr_type): New type.
11676 (builtin_structptr_types): Declare new array.
11677 * tree.c (builtin_structptr_types): New array.
11678 (free_lang_data, build_common_tree_nodes): Use it.
11680 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11683 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11686 2017-06-22 Matt Turner <mattst88@gmail.com>
11688 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11689 Lake models to skylake case. Assume skylake for unknown
11690 models with clflushopt.
11692 2017-06-22 Jeff Law <law@redhat.com>
11694 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11695 frame sizes that do not satisfy aarch64_uimm12_shift.
11697 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11699 * profile-count.h (apply_probability,
11700 apply_scale, probability_in): Fix checks for zero.
11702 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11704 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11705 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11707 2016-06-22 Richard Biener <rguenther@suse.de>
11709 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11710 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11711 REDUC_MAX_EXPR support.
11712 (vectorizable_reduction): Likewise.
11713 (vect_create_epilog_for_reduction): Likewise.
11715 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11717 * match.pd (A / (1 << B) -> A >> B): New.
11718 * generic-match-head.c: Include optabs-tree.h.
11719 * gimple-match-head.c: Likewise.
11720 * optabs-tree.h (target_supports_op_p): New.
11721 * optabs-tree.c (target_supports_op_p): New.
11723 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11725 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11726 $gcc_cv_ld --help output.
11727 (gcc_cv_ld_demangle): Likewise.
11728 (gcc_cv_ld_eh_frame_hdr): Likewise.
11729 (gcc_cv_ld_pie): Likewise.
11730 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11731 (gcc_cv_ld_buildid): Likewise.
11732 (gcc_cv_ld_sysroot): Likewise.
11733 (ld_bndplt_support): Likewise.
11734 (ld_pushpopstate_support): Likewise.
11735 * configure: Regenerate.
11736 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11738 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11741 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11742 operands indexes to avoid gap between operands and match_dups.
11744 2017-06-21 Andrew Pinski <apinski@cavium.com>
11746 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11747 Increment Arith_shift and Arith_shift_reg by 1.
11748 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11750 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11751 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11752 (aarch64_strip_extend): Add new argument and test for it.
11753 (aarch64_cheap_mult_shift_p): New function.
11754 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11755 add a cost if it is true.
11756 Update calls to aarch64_strip_extend.
11757 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11759 2017-06-21 Andrew Pinski <apinski@cavium.com>
11761 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11763 (thunderxt88): Likewise.
11764 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11765 (thunderx_prefetch_tune): New variable.
11766 (thunderx2t99_prefetch_tune): Update for the correct values.
11767 (thunderxt88_tunings): New variable.
11768 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11769 generic_prefetch_tune.
11770 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11772 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11774 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11775 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11776 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11777 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11778 (aarch64_atomic_cas<mode>, GPI): Likewise.
11780 2017-06-21 Martin Liska <mliska@suse.cz>
11782 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11783 statements on cold and hot labels.
11784 * predict.c (tree_estimate_probability_bb): Remove the
11785 prediction from this place.
11787 2017-06-21 Martin Liska <mliska@suse.cz>
11789 PR tree-optimization/79489
11790 * gimplify.c (maybe_add_early_return_predict_stmt): New
11792 (gimplify_return_expr): Call the function.
11793 * predict.c (tree_estimate_probability_bb): Remove handling
11795 * predict.def: Update comment about early return predictor.
11796 * gimple-predict.h (is_gimple_predict): New function.
11797 * predict.def: Change default value of early return to 66.
11798 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11800 * passes.def: Put pass_strip_predict_hints to the beginning of
11803 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11805 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11806 FUNCTION_DECL declarations.
11807 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11809 (dwaf2out_decl): Likewise.
11810 * godump.c (go_early_global_decl): Skip call to the real debug hook
11811 for FUNCTION_DECL declarations.
11812 * passes.c (rest_of_decl_compilation): Skip call to the
11813 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11814 -fdump-go-spec is passed.
11816 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11818 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11819 Reorder for compactness.
11820 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11821 (def_builtin_pure, def_builtin_pure2): New functions.
11822 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11824 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11826 * match.pd (nop_convert): New predicate.
11827 ((A +- CST1) +- CST2): Allow some NOP conversions.
11829 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11832 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11833 with ctors/dtors if GOVD_SHARED is set.
11835 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11837 * config/aarch64/aarch64.md (movti_aarch64):
11838 Emit mov rather than orr.
11839 (movtf_aarch64): Likewise.
11840 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11841 Emit mov rather than orr.
11843 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11845 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11846 Swap alternatives, make integer dup more expensive.
11848 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11850 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11851 Return true for non-tls symbols.
11853 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11855 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11856 (cortex-a75): Likewise.
11857 (cortex-a75.cortex-a55): Likewise.
11858 * config/aarch64/aarch64-tune.md: Regenerate.
11859 * doc/invoke.texi (-mtune): Document new values for -mtune.
11861 2017-06-21 Tom de Vries <tom@codesourcery.com>
11863 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11864 stack_size feature.
11865 (Effective-Target Keywords, Other attributes): Suggest using
11866 dg-add-options stack_size feature to get stack limit in stack_size
11867 effective target documentation.
11869 2017-06-21 Julian Brown <julian@codesourcery.com>
11870 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11872 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11873 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11874 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11876 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11877 attribute type list for neon_multiply.
11878 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11879 attribute type list for neon_multiply.
11880 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11881 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11882 attribute type list for neon_multiply.
11883 * config/arm/types.md (crypto_pmull): Add.
11884 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11885 attribute type list.
11887 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11889 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11892 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11894 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11895 to make sure not to dereference a NULL cost_classes_ptr pointer.
11897 2017-06-20 Carl Love <cel@us.ibm.com>
11899 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11900 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11901 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11902 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11903 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11904 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11905 VMULOSW): New enum "unspec" values.
11906 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11907 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11908 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11909 altivec_vmulosw): New patterns.
11910 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11911 VMULOSW): Add definitions.
11913 2017-06-20 Julia Koval <julia.koval@intel.com>
11915 * config/i386/i386.c: Fix rounding expand for new pattern.
11916 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11918 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11920 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11921 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11923 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11925 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11928 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11930 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11931 architecture, then by implementer ID.
11932 * config/aarch64/aarch64-tune.md: Regenerate.
11934 2017-06-20 Richard Biener <rguenther@suse.de>
11936 PR middle-end/81097
11937 * fold-const.c (split_tree): Fold to type before negating.
11939 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11941 * diagnostic-show-locus.c
11942 (selftest::test_fixit_deletion_affecting_newline): New function.
11943 (selftest::diagnostic_show_locus_c_tests): Call it.
11945 2017-06-20 Andreas Schwab <schwab@suse.de>
11948 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11951 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11953 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11954 __ARM_FEATURE_COPROC according to support.
11956 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11958 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11959 Rewritten to avoid overflow for > 32-bit pointers.
11962 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11963 by removing enum keyword.
11964 (ubsan_type_descriptor): Likewise. Formatting fix.
11967 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11968 splitter): Require TARGET_SSE2 in the condition.
11970 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11973 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11974 for doing vector set of SFmode on ISA 3.0.
11975 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11976 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11978 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11979 SFmode value into a V4SF variable that was extracted from another
11980 V4SF variable without converting the element to double precision
11981 and back to single precision vector format.
11982 (vsx_insert_extract_v4sf_p9_2): Likewise.
11984 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11986 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11987 in UWHI to avoid undefined overflow.
11990 * ubsan.h (enum ubsan_encode_value_phase): New.
11991 (ubsan_encode_value): Change second argument to
11992 enum ubsan_encode_value_phase with default value of
11993 UBSAN_ENCODE_VALUE_GENERIC.
11994 * ubsan.c (ubsan_encode_value): Change second argument to
11995 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11996 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11997 create_tmp_var_raw instead of create_tmp_var and use a
11999 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12000 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12001 ubsan_encode_value callers.
12004 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12005 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12006 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12008 2017-06-19 Richard Biener <rguenther@suse.de>
12010 PR middle-end/81118
12011 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12012 estimates if we changed anything.
12014 2017-06-19 Richard Biener <rguenther@suse.de>
12016 PR tree-optimization/80887
12017 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12018 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12019 simplified lookups, then reset mprts_hook.
12020 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12022 (try_to_simplify): Likewise.
12024 2017-06-19 Martin Liska <mliska@suse.cz>
12027 * gimplify.c (gimplify_switch_expr):
12028 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12030 2017-06-19 Martin Liska <mliska@suse.cz>
12032 * doc/install.texi: Document that PGO runs in 4 stages.
12034 2017-06-19 Martin Liska <mliska@suse.cz>
12037 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12038 to dispatcher function name.
12039 * multiple_target.c (replace_function_decl): New function.
12040 (create_dispatcher_calls): Redirect both edges and references.
12042 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12044 * profile-count.c (profile_count::dump): Dump quality.
12045 (profile_count::differs_from_p): Update for unsigned val.
12046 * profile-count.h (profile_count_quality): New enum.
12047 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12049 2017-06-19 Richard Biener <rguenther@suse.de>
12051 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12052 struct function as arg.
12053 (estimate_numbers_of_iterations): Export overload with loop arg.
12054 (free_numbers_of_iterations_estimates_loop): Use an overload of
12055 free_numbers_of_iterations_estimates instead.
12056 * tree-cfg.c (remove_bb): Adjust.
12057 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12058 * tree-parloops.c (gen_parallel_loop): Likewise.
12059 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12061 (tree_unroll_loops_completely): Likewise.
12062 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12063 Use an overload instead and export.
12064 (estimated_loop_iterations): Adjust.
12065 (max_loop_iterations): Likewise.
12066 (likely_max_loop_iterations): Likewise.
12067 (estimate_numbers_of_iterations): Take struct function as arg
12069 (loop_exits_before_overflow): Adjust.
12070 (free_numbers_of_iterations_estimates_loop): Use an overload.
12071 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12072 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12074 2017-06-19 Richard Biener <rguenther@suse.de>
12077 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12078 RANGE_EXPR conservatively.
12080 2017-06-16 Carl Love <cel@us.ibm.com>
12082 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12083 definitions for vec_float, vec_float2, vec_floato,
12084 vec_floate built-ins.
12085 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12086 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12088 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12089 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12090 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12091 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12092 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12093 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12094 vec_floato): Add builtin defines.
12095 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12096 Update the built-in documentation file for the new built-in
12099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12101 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12102 (mthumb): Mark as the negative of -marm.
12104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12106 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12108 (ARM Options, -mtune): Document that this accepts the same
12109 extension options as -mcpu.
12110 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12114 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12115 permitted extensions.
12117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12119 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12120 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12121 (armv8-m.main): Add option +nodsp.
12122 * config/arm/arm-cpu-cdata.h: Regenerated.
12124 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12126 * config/arm/t-fuchsia: New file.
12127 * config.gcc (arm*-*-fuchsia*): Use it.
12129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12131 * config/arm/t-symbian: Rewrite for new option infrastructure.
12133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12135 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12136 (MULTILIB_REQUIRED): Likewise.
12138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12140 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12141 (MULTILIB_RESUE): Likewise.
12142 (MULTILIB_MATCHES): Likewise.
12143 (MULTLIB_REQUIRED): Likewise.
12145 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12147 * config/arm/t-rtems: Rewrite for new option framework.
12149 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12151 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12152 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12153 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12154 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12155 * config/arm/t-multilib: ... here.
12156 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12157 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12158 armv7-a and armv8*-a when A-profile libraries have not been built.
12159 * config/arm/t-rmprofile: Rewrite.
12161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12163 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12164 with a backslash. Remove the backslash after substituting unescaped
12166 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12170 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12171 the driver to pass through the default setting of -mfloat-abi.
12172 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12174 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12175 (all_feat_combs): New rule.
12176 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12178 * config/arm/t-aprofile: Rewrite.
12180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12182 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12183 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12184 fpu is not specified by the user/command-line.
12185 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12186 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12187 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12188 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12189 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12190 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12192 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12194 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12195 * config/arm/t-arm-elf: Rewritten.
12197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12199 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12200 have some floating-point instructions.
12201 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12202 (TARGET_MAYBE_HARD_FLOAT): New macro.
12203 * config/arm/arm-builtins.c (arm_init_builtins): Use
12204 TARGET_MAYBE_HARD_FLOAT.
12205 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12209 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12210 (configargs.h): Include it.
12211 (arm_print_hint_for_fpu_option): New function.
12212 (arm_parse_fpu_option): New function.
12213 (candidate_extension): New class.
12214 (arm_canon_for_multilib): New function.
12215 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12216 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12217 (ARCH_CANONICAL_SPECS): New macro.
12218 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12222 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12223 are set after handling multilib fragments. Set target_cpu_default2
12226 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12228 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12230 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12231 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12233 (check_arch): Likewise.
12234 * config/arm/arm.c (arm_configure_build_target): Handle
12235 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12236 options in the default.
12238 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12240 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12241 when an option is an alias of another.
12242 * config/arm/parsecpu.awk (optalias): New parser token.
12243 (gen_comm_data): Mark non-alias options as such. Emit entries
12244 for extension aliases.
12245 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12246 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12247 (armv6kz, armv6zk, armv6t2): Likewise.
12248 (armv7): Make vfpv3-d16 an alias.
12249 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12251 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12252 Sort in canonical order.
12253 (armv8-a): Sort in canonical order.
12254 (armv8.1-a, armv8.2-a): Likewise.
12255 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12257 (cortex-a9): Sort in canonical order.
12258 * config/arm/arm.c (selftests.h): Include it.
12259 (arm_test_cpu_arch_data): New function.
12260 (arm_run_self_tests): New function.
12261 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12262 (targetm): Move declaration to the end of the file.
12263 * arm-cpu-cdata.h: Regenerated.
12265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12267 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12268 call to target_mode_check describing the type of option passed.
12269 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12270 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12271 arm_parse_cpu_option_name to match parameters against list of
12273 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12274 arm_arch_core_flags data structure.
12275 * config/arm/arm-cpu_cdata.h: Regenerated.
12277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12279 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12281 (arm_print_hint_for_cpu_option): Likewise.
12282 (arm_print_hint_for_arch_option): Likewise.
12283 (arm_parse_cpu_option_name): Likewise.
12284 (arm_parse_arch_option_name): Likewise.
12285 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12286 of entries in the all_fpus list.
12287 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12288 (arm_parse_cpu_option_name): Declare.
12289 (arm_parse_arch_option_name): Declare.
12290 (arm_parse_option_features): Declare.
12291 (arm_intialize_isa): Declare.
12292 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12294 (gen_comm_data): ... here. Make definitions non-static.
12295 * config/arm/arm-cpu-data.h: Regenerated.
12296 * config/arm/arm-cpu-cdata.h: Regenerated.
12298 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12300 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12301 (cpu_arch_extension): New structure.
12302 (cpu_arch_option, arch_option, cpu_option): New structures.
12303 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12304 architecture types.
12305 (gen_data): Generate new format data tables.
12306 * config/arm/arm.c (cpu_tune): New structure.
12307 (cpu_option, processors): Delete.
12308 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12309 (arm_print_hint_for_cpu_option): ... this and ...
12310 (arm_print_hint_for_arch_option): ... this.
12311 (arm_parse_arch_cpu_name): Delete. Replace with ...
12312 (arm_parse_cpu_option_name): ... this and ...
12313 (arm_parse_arch_option_name): ... this.
12314 (arm_unrecognized_feature): Change type of target parameter to
12316 (arm_parse_arch_cpu_features): Delete. Replace with ...
12317 (arm_parse_option_features): ... this.
12318 (arm_configure_build_target): Rework to use new configuration data
12320 (arm_print_tune_info): Rework for new configuration data tables.
12321 * config/arm/arm-cpu-data.h: Regenerated.
12322 * config/arm/arm-cpu.h: Regenerated.
12324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12326 * Makefile.in (OBJS): Move sbitmap.o from here ...
12327 (OBJS-libcommon): ... to here.
12329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12331 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12332 (ISA_ALL_CRYPTO): New macro.
12333 (ISA_ALL_SIMD): New macro
12334 (ISA_ALL_FP): New macro.
12335 * config/arm/arm.c (fpu_bitlist): Update initializer.
12336 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12338 (arm9e): Add fpu. Add option for nofp
12339 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12340 (arm926ej-s, arm1026ej-s): Likewise.
12341 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12342 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12343 neon-fp16, neon-vfpv4, nofp and nosimd.
12344 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12345 (cortex-a8): Add fpu. Add option for nofp.
12346 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12347 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12348 (cortex-r4f): Add fpu.
12349 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12350 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12352 (cortex-r8): Likewise.
12353 (cortex-m4): Add fpu. Add option for nofp.
12354 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12355 (cortex-a17.cortex-a7): Likewise.
12356 (cortex-a32): Add fpu. Add options for crypto and nofp.
12357 (cortex-a35, cortex-a53): Likewise.
12358 (cortex-a57): Add fpu. Add option for crypto.
12359 (cortex-a72, cortex-a73): Likewise.
12360 (exynos-m1): Likewise.
12361 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12362 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12363 (cortex-m33): Add fpu. Add option for nofp.
12364 * config/arm/arm-cpu-cdata.h: Regenerated
12365 * config/arm/arm-cpu-data.h: Regenerated.
12367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12369 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12370 (armv5te, armv5tej): Likewise.
12371 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12372 (armv7): Add options fp and vfpv3-d16.
12373 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12374 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12376 (armv7ve): Likewise.
12377 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12378 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12379 (armv8-a): Add nocrypto option.
12380 (armv8.1-a, armv8.2-a): Likewise.
12381 (armv8-m.main): add options fp, fp.dp and nofp.
12383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12385 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12387 (armv8-a+crc): Delete.
12388 (armv8.1-a): Add options simd, crypto and nofp.
12389 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12390 (armv8.2-a+fp16): Delete.
12391 (armv8-m.main): Add option dsp.
12392 (armv8-m.main+dsp): Delete.
12393 (cortex-a8): Add fpu. Add nofp option.
12394 (cortex-a9): Add fpu. Add nofp and nosimd options.
12395 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12396 link to main cpu and architecture data structures.
12397 (gen_comm_data): Only put isa attributes from the main architecture
12399 (option): New statement for architecture and CPU entries.
12400 * arm.c (struct cpu_option): New structure.
12401 (struct processors): Add entry for options.
12402 (arm_unrecognized_feature): New function.
12403 (arm_parse_arch_cpu_name): Ignore any characters after the first
12405 (arm_parse_arch_cpu_feature): New function.
12406 (arm_configure_build_target): Separate out any CPU and architecture
12407 features and parse separately. Don't error out if -mfpu=auto is
12408 used with only an architecture string.
12409 (arm_print_asm_arch_directives): New function.
12410 (arm_file_start): Call it.
12411 * config/arm/arm-cpu-cdata.h: Regenerated.
12412 * config/arm/arm-cpu-data.h: Likewise.
12413 * config/arm/arm-tables.opt: Likewise.
12415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12417 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12418 assembler when it is not -mfpu=auto.
12420 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12422 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12423 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12424 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12425 (ASM_CPU_SPEC): Rewrite.
12426 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12427 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12428 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12429 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12430 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12431 copied string is NUL-terminated. Also strip any characters prefixed
12433 (arm_rewrite_selected_arch): New function.
12434 (arm_rewrite_march): New function.
12436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12438 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12439 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12440 (march, mcpu, mtune): Convert to string-based options.
12441 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12442 (arm_parse_arch_cpu_name): New function.
12443 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12444 identify selected architecture or CPU.
12445 (arm_option_save): New function.
12446 (TARGET_OPTION_SAVE): Redefine.
12447 (arm_option_restore): Restore string options.
12448 (arm_option_print): Print string options.
12450 2017-06-16 Martin Sebor <msebor@redhat.com>
12452 PR tree-optimization/80933
12453 PR tree-optimization/80934
12454 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12455 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12456 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12457 (gimple_fold_builtin): Call them.
12459 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12461 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12462 as unlikely; update profile.
12464 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12466 * predict.c (force_edge_cold): Handle declaring edges impossible
12469 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12471 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12473 (try_unroll_loop_completely): Fix reporting.
12475 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12477 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12479 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12482 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12483 if given a non-constant argument for an intrinsic which requires a
12486 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12488 * profile.c (compare_freqs): New function.
12489 (branch_prob): Sort edge list.
12490 (find_spanning_tree): Assume that the list is priority sorted.
12492 2017-06-16 Richard Biener <rguenther@suse.de>
12494 PR tree-optimization/81090
12495 * passes.def (pass_record_bounds): Remove.
12496 * tree-pass.h (make_pass_record_bounds): Likewise.
12497 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12498 make_pass_record_bounds): Likewise.
12499 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12500 not free niter estimates at the beginning but at the end.
12501 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12503 2017-06-16 Richard Biener <rguenther@suse.de>
12505 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12506 initializer to workaround ICE in host GCC 4.8.
12508 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12510 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12512 (clone_inlined_nodes): Update.
12514 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12516 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12517 prefetch settings, and enable prefetching by default at -O3.
12519 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12521 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12522 Set flag_prefetch_loop_arrays according to tuning data.
12524 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12526 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12527 New tune structure.
12528 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12529 [Unrelated to main purpose of the patch] Place the pointer field last
12530 to enable type checking errors when tune structure are wrongly merged.
12531 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12532 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12533 (thunderx2t99_prefetch_tune): New tune constants.
12534 (tune_params *_tunings): Update all tunings (no functional change).
12535 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12536 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12537 from tunings structures.
12539 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12542 * ubsan.c (instrument_null): Add T argument, use it instead
12543 of computing it based on IS_LHS.
12544 (instrument_object_size): Likewise.
12545 (pass_ubsan::execute): Adjust instrument_null and
12546 instrument_object_size callers to pass gimple_get_lhs or
12547 gimple_assign_rhs1 result to it. Use instrument_null instead of
12548 calling get_base_address and instrument_mem_ref. Handle
12549 aggregate call arguments for object-size sanitization.
12551 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12553 PR tree-optimization/81089
12554 * tree-vrp.c (is_masked_range_test): Validate operands of
12557 2017-06-15 Martin Sebor <msebor@redhat.com>
12560 * dumpfile.c (dump_register): Avoid calling memset to initialize
12561 a class with a default ctor.
12562 * gcc.c (struct compiler): Remove const qualification.
12563 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12564 * hash-table.h: Ditto.
12565 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12567 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12568 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12570 * params.h (struct param_info): Make struct members non-const.
12571 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12572 with default initialization.
12573 * vec.h (vec_copy_construct, vec_default_construct): New helper
12575 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12576 with vec_copy_construct.
12577 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12578 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12579 * doc/invoke.texi (-Wclass-memaccess): Document.
12581 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12583 * emit-rtl.h (is_leaf): Update comment about local
12584 register allocator.
12586 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12589 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12590 for a variable to have a section before checking if the section has a
12592 Set section to.persistent if persistent attribute is set.
12593 Warn if .persistent attribute is used on an automatic variable.
12595 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12597 PR rtl-optimization/80474
12598 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12600 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12602 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12605 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12607 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12608 (casesi): Emit bounds checking as RTL.
12609 (casesi_internal_mips16_<mode>): Remove bounds checking.
12611 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12613 * config/xtensa/xtensa.c (xtensa_option_override): Append
12614 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12615 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12616 xtensa_doloop_hooks): Define unconditionally.
12617 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12619 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12620 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12621 for it in xtensa_option_override.
12622 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12623 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12625 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12627 * doc/cppopts.texi: Document '-' special value to -MF.
12629 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12631 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12632 (cortex_a53_fconst): Likewise.
12633 (cortex_a53_fpmul): Likewise.
12634 (cortex_a53_f_load_64): Likewise.
12635 (cortex_a53_f_load_many): Likewise.
12636 (cortex_a53_advsimd_alu): Likewise.
12637 (cortex_a53_advsimd_alu_q): Likewise.
12638 (cortex_a53_advsimd_mul): Likewise.
12639 (cortex_a53_advsimd_mul_q): Likewise.
12640 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12641 Add missing fmul, r2f_cvt and fconst cases.
12643 2017-06-14 Richard Biener <rguenther@suse.de>
12645 PR middle-end/81088
12646 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12648 (fold_binary_loc): When associating do not treat pre-existing
12649 TREE_OVERFLOW on literal constants as a reason to allow
12650 TREE_OVERFLOW on associated literal constants.
12652 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12654 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12655 (MASK_FEATURES): New macro.
12656 * config/sparc/sparc.c (sparc_option_override): Remove the special
12657 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12659 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12661 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12662 a division of 0 if non-call exceptions are enabled.
12664 2017-06-14 Andrew Pinski <apinski@cavium.com>
12665 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12668 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12669 Improve vector initialization code gen for only variable case.
12671 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12673 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12675 2017-06-14 Richard Biener <rguenther@suse.de>
12677 PR tree-optimization/81083
12678 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12681 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12683 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12684 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12685 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12686 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12687 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12688 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12690 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12692 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12693 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12695 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12697 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12699 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12701 * config/rs6000/t-rtems: Don't handle SPE.
12703 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12705 * config/rs6000/t-linux: Don't handle SPE.
12707 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12709 * config/rs6000/eabispe.h: Delete file.
12711 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12713 * config/rs6000/t-spe: Delete file.
12715 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12717 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12718 (rs6000_legitimate_offset_address_p): Return false for anything in
12719 V2SImode or V2SFmode.
12721 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12723 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12724 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12725 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12727 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12728 (rs6000_legitimate_offset_address_p): Ditto.
12729 (rs6000_emit_move): Ditto.
12730 (rs6000_init_builtins): Remove V4HI_type_node.
12732 2017-06-13 Martin Liska <mliska@suse.cz>
12735 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12736 (gate_asan): Likewise.
12737 * asan.h (asan_no_sanitize_address_p): Remove the function.
12738 (sanitize_flags_p): New function.
12739 * builtins.def: Fix coding style.
12740 * common.opt: Use renamed enum value.
12741 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12742 * doc/extend.texi: Document no_sanitize attribute.
12743 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12744 to SANITIZE_UNDEFINED_NONDEFAULT.
12745 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12746 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12747 Use sanitize_flags_p.
12748 * gimplify.c (gimplify_function_tree): Likewise.
12749 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12750 * opts.c (parse_no_sanitize_attribute): New function.
12751 (common_handle_option): Use renamed enum value.
12752 * opts.h (parse_no_sanitize_attribute): Declare.
12753 * tree.c (sanitize_flags_p): New function.
12754 * tree.h: Declared here.
12755 * tsan.c: Use sanitize_flags_p.
12756 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12757 (instrument_mem_ref): Likewise.
12758 (instrument_bool_enum_load): Likewise.
12759 (do_ubsan_in_current_function): Remove the function.
12760 (pass_ubsan::execute): Use sanitize_flags_p.
12761 * ubsan.h: Remove do_ubsan_in_current_function
12762 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12763 (dump_function_to_file): Use it here.
12765 2017-06-13 Martin Jambor <mjambor@suse.cz>
12767 PR tree-optimization/80803
12768 PR tree-optimization/81063
12769 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12770 (propagate_subaccesses_across_link): Enqueue subtree whenever
12771 necessary instead of relying on the caller.
12773 2017-06-13 Martin Jambor <mjambor@suse.cz>
12775 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12776 that have a first_link.
12777 (sort_and_splice_var_accesses): Do not check first_link before
12779 (subtree_mark_written_and_enqueue): Likewise.
12780 (propagate_all_subaccesses): Likewise and do not stop at first
12781 parent with a first_link.
12783 2017-06-13 Martin Jambor <mjambor@suse.cz>
12785 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12788 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12790 * match.pd: New pattern.
12792 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12794 * tree-vrp.c (is_masked_range_test): New function.
12795 (register_edge_assert_for): Determine ranges for
12798 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12800 PR tree-optimization/67328
12801 * fold-const.c (maskable_range_p): New function.
12802 (build_range_check): Generate bittests if possible.
12804 2017-06-13 Martin Liska <mliska@suse.cz>
12806 * gimple-pretty-print.c (dump_probability): Add new argument.
12807 (dump_edge_probability): Dump both probability and count.
12808 (dump_gimple_label): Likewise.
12809 (dump_gimple_bb_header): Likewise.
12811 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12814 * config/avr/avr-devices.c: Fix indentation.
12815 * config/avr/gen-avr-mmcu-specs.c: Dito.
12817 2017-06-13 Richard Biener <rguenther@suse.de>
12819 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12820 instead get vector type from stmt_info.
12821 (vectorizable_reduction): Adjust. Remove dead code.
12823 2017-06-13 Richard Biener <rguenther@suse.de>
12825 PR middle-end/81065
12826 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12827 case of C * (x * C2 + C3).
12828 (fold_addr_of_array_ref_difference): Properly fold index difference.
12830 2017-06-12 David S. Miller <davem@davemloft.net>
12833 * config/sparc/sparc.md (return expander): Emit frame blockage if
12834 function uses alloca.
12836 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12838 * combine.c (make_field_assignment): Check len rather than the mode
12839 precision when calling force_to_mode.
12841 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12843 Support multilibs and devices that see flash in RAM address range.
12846 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12847 (avr_mcu_t) <flash_pm_offset>: New field.
12848 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12849 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12850 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12851 (AVR_TINY_PM_OFFSET): Remove macro.
12852 * config/avr/avr.opt (-mshort-calls): New option.
12853 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12854 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12855 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12856 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12857 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12858 instead of avr_arch->have_jmp_call.
12859 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12860 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12861 avr_arch->flash_pm_offset to define.
12862 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12863 new field flash_pm_offset. Add entry for avrxmega3.
12864 (avr_texinfo): Add entry for avrxmega3.
12865 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12866 attiny212, attiny214,
12867 attiny412, attiny414, attiny416, attiny417,
12868 attiny814, attiny816, attiny817,
12869 attiny1614, attiny1616, attiny1617,
12870 attiny3214, attiny3216, attiny3217.
12871 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12872 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12873 (avr_print_operand_address) [AVR_TINY]: Same.
12874 (avr_asm_init_sections) <readonly_data_section>: Only patch
12875 callback if avr_arch->flash_pm_offset = 0.
12876 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12877 for rodata if avr_arch->flash_pm_offset != 0.
12878 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12879 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12880 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12881 (m_options): Append opt_rcall.
12882 (m_dirnames): Append dir_rcall.
12883 * config/avr/t-multilib: Regenerate.
12885 * configure.ac [target=avr]: Check whether avrxmega3 default
12886 linker description file works as needed.
12887 * configure: Regenerate.
12888 * doc/avr-mmcu.texi: Regenerate.
12889 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12890 <__AVR_ARCH__>: Document avrxmega3 and 103.
12891 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12892 <__AVR_SHORT_CALLS__>: Document it.
12893 <__AVR_PM_BASE_ADDRESS__>: Document it.
12894 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12895 (AVR Variable Attributes) <progmem>: Document this is
12896 not needed for avrxmega3.
12897 (AVR Named Address Spaces) <__flash>: Dito.
12899 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12901 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12903 2017-06-12 Doug Rupp <rupp@adacore.com>
12905 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12906 Append vxworks-stdint.h to the tm_file list.
12907 * config/vxworks-stdint.h: New file.
12909 2017-06-12 Martin Liska <mliska@suse.cz>
12911 PR tree-optimization/81041
12912 * tree-profile.c (gimple_gen_ic_func_profiler):
12913 Create an extra BB in profile-generate
12914 (gimple_gen_time_profiler): Likewise.
12916 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12918 PR tree-optimization/81003
12919 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12920 (update_range_test): Use it instead of force_gimple_operand_gsi.
12922 2017-06-12 Richard Biener <rguenther@suse.de>
12924 PR tree-optimization/81053
12925 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12926 with backedge value not defined in loop. Simplify def stmt
12929 2017-06-11 Tom de Vries <tom@codesourcery.com>
12932 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12934 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12935 nvptx_cannot_force_const_mem.
12937 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12939 * opts.c (finish_options): Move test for flag_split_stack after
12940 it has been initialized.
12942 2017-06-11 Jason Merrill <jason@redhat.com>
12944 * tree.h (id_equal): New.
12945 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12946 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12947 instead of strcmp of IDENTIFIER_POINTER.
12949 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12951 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12952 (mark_all_inlined_calls_cdtor): Fix formating.
12953 (inline_transform): Rescale profile before inlining.
12955 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12957 * cgraph.h (cgraph_edge::clone): Update prototype.
12958 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12959 (cgraph_node::create_clone): Update.
12960 (cgraph_node::create_version_clone): Update.
12961 * tree-inline.c (copy_bb): Update.
12962 (expand_call_inline): Update.
12964 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12966 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12967 factored out from ...
12968 (rs6000_emit_prologue): ... here.
12970 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12972 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12973 factored out from ...
12974 (rs6000_emit_prologue): ... here.
12976 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12978 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12980 (handle_missing_profiles): Fix computation of tp_first_run.
12981 (counts_to_freqs): Do not touch freqs when count is 0.
12983 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12985 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12988 2017-06-10 Tom de Vries <tom@codesourcery.com>
12990 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12991 attributes): Document signal effective target.
12993 2017-06-10 Tom de Vries <tom@codesourcery.com>
12995 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12996 Document effective target stack_size.
12998 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13000 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13001 to the edit_context if they can be auto-applied.
13003 2017-06-9 Ian Lance Taylor <iant@golang.org>
13005 * opts.c (finish_options): If -fsplit-stack, disable implicit
13006 -forder-blocks-and-partition.
13007 * doc/invoke.texi (Optimize Options): Document that when using
13008 -fsplit-stack -forder-blocks-and-partition is not implicitly
13011 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13013 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13014 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13015 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13016 * builtins.def (abort, trap, unreachable): Declare cold.
13017 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13018 * tree-core.h (ECF_COLD): New.
13019 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13020 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13022 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13024 * predict.c (unlikely_executed_stmt_p): Cleanup.
13026 2017-06-09 Richard Biener <rguenther@suse.de>
13028 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13029 model if the ref is always written to.
13031 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13033 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13035 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13037 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13040 2017-06-09 Tom de Vries <tom@codesourcery.com>
13043 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13044 "target cannot support label values" when encountering LABEL_REF.
13046 2017-06-09 Martin Liska <mliska@suse.cz>
13048 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13049 (gimple_gen_ic_func_profiler): Emit direct comparison
13050 of __gcov_indirect_call_callee with NULL.
13051 (gimple_gen_time_profiler): Change probability from
13052 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13054 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13056 * profile.c (edge_gcov_counts): Turn to pointer.
13057 (compute_branch_probabilities, compute_branch_probabilities): Update.
13058 (branch_prob): Do not clear edge_gcov_count.
13059 * profile.h (edge_gcov_counts): Turn to pointer.
13060 (edge_gcov_count): Update.
13062 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13064 * gimple.h (gimple_check_failed): Mark cold.
13066 2017-06-09 Richard Biener <rguenther@suse.de>
13068 PR tree-optimization/66623
13069 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13070 refactor check_reduction into two parts, properly computing
13071 whether we have to check reduction validity for outer loop
13074 2017-06-09 Richard Biener <rguenther@suse.de>
13076 PR tree-optimization/79483
13077 * graphite-scop-detection.c (order): New global.
13078 (get_order): Compute bb to order mapping that satisfies code
13079 generation constraints.
13080 (cmp_pbbs): New helper.
13081 (build_scops): Start domwalk at entry block, sort generated
13084 2017-06-09 Richard Biener <rguenther@suse.de>
13086 PR middle-end/81007
13087 * ipa-polymorphic-call.c
13088 (ipa_polymorphic_call_context::restrict_to_inner_class):
13089 Skip FIELD_DECLs with error_mark_node type.
13090 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13093 2017-06-09 Martin Liska <mliska@suse.cz>
13095 * predict.c (struct branch_predictor): New struct.
13096 (test_prediction_value_range): New test.
13097 (predict_c_tests): New function.
13098 * selftest-run-tests.c (selftest::run_tests): Run the function.
13099 * selftest.h: Declare new tests.
13101 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13104 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13105 gen_add3_insn did not fail.
13106 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13107 r0, construct that number in a temporary reg and add that reg to r0.
13108 If asked to put the result in r0 as well, fail.
13110 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13112 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13113 for early expansion of vec_eqv.
13115 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13117 PR middle-end/81005
13118 * ubsan.c (instrument_null): Avoid pointless code temporary.
13119 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13121 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13125 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13128 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13131 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13133 * predict.c (unlikely_executed_edge_p): Move ahead.
13134 (probably_never_executed_edge_p): Use it.
13136 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13138 PR middle-end/79988
13139 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13140 gimple_call_builtin_p call.
13142 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13144 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13145 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13146 rtl_check_failed_type2, rtl_check_failed_code1,
13147 rtl_check_failed_code2, rtl_check_failed_code_mode,
13148 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13149 rtvec_check_failed_bounds, rtl_check_failed_flag,
13150 _fatal_insn_not_found, _fatal_insn): Likewise.
13151 * tree.h (tree_contains_struct_check_failed,
13152 tree_check_failed, tree_not_check_failed,
13153 tree_class_check_failed, tree_range_check_failed,
13154 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13155 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13156 tree_operand_check_failed, omp_clause_check_failed,
13157 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13160 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13162 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13163 flag_branch_probabilities.
13164 * ipa-inline.c (edge_badness): Likewise.
13165 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13166 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13167 * predict.c (maybe_hot_frequency_p): Likewise.
13168 (probably_never_executed): Likewise.
13169 * sched-ebb.c (schedule_ebbs): Likewise.
13170 * sched-rgn.c (find_single_block_region): Likewise.
13171 * tracer.c (tail_duplicate): Likewise.
13173 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13175 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13176 longer requires x_flag_profile_use.
13178 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13180 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13181 instead of flag_reorder_blocks_and_partition.
13182 * dbxout.c (dbxout_function_end): Likewise.
13183 * dwarf2out.c (gen_subprogram_die): Likewise.
13184 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13185 * hw-doloop.c (reorg_loops): Likewise.
13186 * varasm.c (assemble_start_function,
13187 assemble_end_function): Likewise.
13188 (decide_function_section): Do not check for
13189 flag_reorder_blocks_and_partition.
13191 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13193 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13195 (chkp_get_hard_register_fake_addr_expr): Ditto.
13196 (chkp_build_addr_expr): Add check for hard reg case.
13197 (chkp_parse_array_and_component_ref): Ditto.
13198 (chkp_find_bounds_1): Ditto.
13199 (chkp_process_stmt): Don't generate bounds store for
13202 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13204 * predict.c (maybe_hot_bb_p): Do not check profile status.
13205 (maybe_hot_edge_p): Likewise.
13206 (probably_never_executed): Check for zero counts even if profile
13208 (unlikely_executed_edge_p): New function.
13209 (unlikely_executed_stmt_p): New function.
13210 (unlikely_executed_bb_p): New function.
13211 (set_even_probabilities): Use unlikely predicates.
13212 (combine_predictions_for_bb): Likewise.
13213 (predict_paths_for_bb): Likewise.
13214 (predict_paths_leading_to_edge): Likewise.
13215 (determine_unlikely_bbs): New function.
13216 (estimate_bb_frequencies): Use it.
13217 (compute_function_frequency): Use zero counts even if profile is
13219 * profile-count.h: Fix typo.
13221 2017-08-08 Julia Koval <julia.koval@intel.com>
13223 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13224 _mm512_mask_cvtsepi16_storeu_epi8,
13225 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13226 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13227 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13228 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13229 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13230 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13231 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13232 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13233 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13234 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13235 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13236 __builtin_ia32_pmovuswb256mem_mask,
13237 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13238 __builtin_ia32_pmovwb512mem_mask): New builtins.
13240 2017-08-08 Julia Koval <julia.koval@intel.com>
13242 PR target/73350,80862
13243 * config/i386/subst.md (round): Fix round pattern.
13244 * config/i386/i386.c (ix86_erase_embedded_rounding):
13245 Fix erasing rounding for the fixed pattern.
13247 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13249 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13251 2017-06-08 Martin Liska <mliska@suse.cz>
13253 PR gcov-profile/80911
13254 * gcov.c (block_info::block_info): New constructor.
13256 2017-06-07 Carl Love <cel@us.ibm.com>
13258 * config/rs6000/rs6000-c: The return type of the following
13259 built-in functions was implemented as int not long long. Fix sign
13260 of return value for the unsigned version of vec_mulo and vec_mule.
13261 vector unsigned long long vec_bperm (vector unsigned long long,
13262 vector unsigned char)
13263 vector signed long long vec_mule (vector signed int,
13265 vector unsigned long long vec_mule (vector unsigned int,
13266 vector unsigned int)
13267 vector signed long long vec_mulo (vector signed int,
13269 vector unsigned long long vec_mulo (vector unsigned int,
13270 vector unsigned int)
13271 * doc/extend.texi: Fix the documentation for the built-in
13274 2017-06-07 Carl Love <cel@us.ibm.com>
13277 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13280 2017-06-07 Carl Love <cel@us.ibm.com>
13282 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13283 support, Generate doublehv for signed int/float for BE case only.
13285 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13287 * doc/invoke.texi (mcx16): Rewrite.
13289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13291 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13292 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13293 *mov<mode>_softfloat, and an anonymous splitter): Use
13294 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13296 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13298 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13300 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13301 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13302 (rs6000_debug_reg_global): Adjust.
13303 (rs6000_init_hard_regno_mode_ok): Adjust.
13304 (rs6000_dbx_register_number): Adjust.
13305 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13306 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13307 Remove SPE_ACC and SPEFSCR.
13308 (REG_ALLOC_ORDER): Ditto.
13309 (FRAME_POINTER_REGNUM): Change to 111.
13310 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13311 (REG_CLASS_NAMES): Ditto.
13312 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13313 (REGISTER_NAMES): Ditto.
13314 (ADDITIONAL_REG_NAMES): Ditto.
13315 (rs6000_reg_names): Ditto.
13316 * config/rs6000/rs6000.md: Renumber some register number
13319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13321 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13323 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13324 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13326 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13327 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13328 Delete the SPE high registers.
13329 (REG_ALLOC_ORDER): Ditto.
13330 (enum reg_class): Remove SPE_HIGH_REGS.
13331 (REG_CLASS_NAMES): Ditto.
13332 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13333 (REGISTER_NAMES): Ditto.
13334 (rs6000_reg_names): Ditto.
13335 * doc/tm.texi.in: Remove SPE as example.
13336 * doc/tm.texi: Regenerate.
13338 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13340 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13341 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13342 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13343 * config/rs6000/rs6000.md (type): Remove "brinc".
13345 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13347 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13348 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13349 * config/rs6000/linuxspe.h: Delete file.
13350 * config/rs6000/rs6000.md: Don't include spe.md.
13351 * config/rs6000/spe.h: Delete file.
13352 * config/rs6000/spe.md: Delete file.
13353 * config/rs6000/t-rs6000: Remove spe.md.
13355 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13357 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13358 (reg_or_none500mem_operand): Delete.
13359 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13360 instead of reg_or_none500mem_operand.
13362 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13364 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13365 handling of SPE flags.
13366 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13368 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13370 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13372 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13373 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13374 paired_divv2sf3): Similar.
13375 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13376 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13377 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13379 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13380 Rename the paired_* instruction patterns.
13381 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13383 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13384 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13385 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13386 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13387 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13388 PAIRED_VECTOR_MODE.
13389 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13390 (spe_func_has_64bit_regs_p): Delete.
13391 (spe_expand_predicate_builtin): Delete.
13392 (spe_expand_evsel_builtin): Delete.
13393 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13394 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13395 (invalid_e500_subreg): Delete.
13396 (rs6000_legitimize_address): Always force_reg op2 as well, for
13397 paired single memory accesses.
13398 (rs6000_member_type_forces_blk): Delete.
13399 (rs6000_spe_function_arg): Delete.
13400 (rs6000_expand_unop_builtin): Delete SPE handling.
13401 (rs6000_expand_binop_builtin): Ditto.
13402 (spe_expand_stv_builtin): Delete.
13403 (bdesc_2arg_spe): Delete.
13404 (spe_expand_builtin): Delete.
13405 (spe_expand_predicate_builtin): Delete.
13406 (spe_expand_evsel_builtin): Delete.
13407 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13408 (spe_init_builtins): Delete.
13409 (spe_func_has_64bit_regs_p): Delete.
13410 (savres_routine_name): Delete "info" parameter. Adjust callers.
13411 (rs6000_emit_stack_reset): Ditto.
13412 (rs6000_dwarf_register_span): Delete.
13413 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13414 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13415 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13417 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13419 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13420 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13421 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13422 mulv2sf3, divv2sf3): Delete expanders.
13424 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13426 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13428 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13430 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13431 * config/rs6000/rs6000.c: Ditto.
13433 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13435 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13436 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13437 comparison_operator.
13439 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13441 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13442 * config/rs6000/rs6000.opt: Ditto.
13443 * config/rs6000/t-rtems: Ditto.
13445 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13447 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13448 TARGET_E500_SINGLE by 0, simplify.
13449 * config/rs6000/rs6000.c: Ditto.
13450 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13451 (spe_build_register_parallel): Delete.
13452 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13453 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13454 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13455 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13456 (E500_CONVERT): Delete.
13457 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13459 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13461 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13462 * config/rs6000/dfp.md: Ditto.
13463 (negdd2, *negdd2_fpr): Merge.
13464 (absdd2, *absdd2_fpr): Merge.
13465 (negtd2, *negtd2_fpr): Merge.
13466 (abstd2, *abstd2_fpr): Merge.
13467 * config/rs6000/e500.h: Delete file.
13468 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13469 TARGET_FPRS by 1 and simplify.
13470 * config/rs6000/rs6000-c.c: Ditto.
13471 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13472 TARGET_DF_SPE by 0.
13473 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13475 * config/rs6000/rs6000.md: Ditto.
13476 (floatdidf2, *floatdidf2_fpr): Merge.
13477 (move_from_CR_gt_bit): Delete.
13478 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13479 (E500_CR_IOR_COMPARE): Delete.
13480 (All patterns that require !TARGET_FPRS): Delete.
13481 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13483 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13485 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13487 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13489 * graphds.c (add_edge): Intitialize edge's attached data.
13490 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13491 pointer parameter. Call pointed function on each edge during
13492 graph traversing. Skip traversing the edge when the function
13494 (graphds_dfs, graphds_scc): Ditto.
13495 (for_each_edge): New parameter. Pass the new parameter to callback
13497 * graphds.h (skip_edge_callback): New function pointer type.
13498 (graphds_dfs, graphds_scc): New function pointer parameter.
13499 (graphds_edge_callback, for_each_edge): New parameter.
13501 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13503 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13504 out code checking if runtime alias check is possible to below ...
13505 Call the new function.
13506 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13507 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13509 2017-06-07 Marek Polacek <polacek@redhat.com>
13512 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13513 TYPE_OVERFLOW_WRAPS check.
13515 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13517 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13518 if versioning is required.
13519 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13520 peeling with the check for versioning.
13522 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13524 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13525 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13526 Set true to new parameter if new ssa variable is defined.
13527 (vect_gen_vector_loop_niters): Refactor. Set range information
13528 for the new vector loop bound variable.
13529 (vect_do_peeling): Ditto.
13531 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13533 * tree-affine.c (ssa.h): Include header file.
13534 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13535 has wrapping overflow behavior.
13537 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13539 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13541 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13543 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13544 (tree_to_aff_combination): ... here.
13546 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13548 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13549 reg_pressure model function.
13550 (ivopts_global_cost_for_size): Delete.
13551 (determine_set_costs, iv_ca_recount_cost): Call new model function
13552 ivopts_estimate_reg_pressure.
13554 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13556 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13557 expensive than udiv. Remove floating point cases from mod.
13559 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13561 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13562 Increase idiv cost.
13564 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13566 * config/aarch64/aarch64.md
13567 (copysignsf3): Fix mask generation.
13569 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13571 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13573 (class dump_manager): Add register_dumps method.
13574 * dumpfile.c: Include langhooks.h.
13575 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13576 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13577 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13578 (dump_manager::dump_register): Start with 512 entries instead of 32.
13579 (dump_manager::register_dumps): New method.
13580 * toplev.c (general_init): Instead of invoking register_dumps
13581 langhook, invoke register_dumps method on the dump manager.
13582 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13585 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13587 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13588 Say that # requires an associated define_split to exist, and that
13589 the define_split must be suitable for use after register allocation.
13591 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13593 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13594 (compute_outgoing_frequencies): Also initialize zero counts.
13595 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13596 around loops; preserve more of profile when nothing changes.
13598 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13600 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13602 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13603 * config/arm/arm-cpu-cdata.h: Regenerate.
13604 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13605 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13606 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13607 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13608 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13610 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13612 * config/arm/t-rmprofile: Likewise.
13613 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13615 2017-06-06 David S. Miller <davem@davemloft.net>
13618 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13619 blockage if function uses alloca.
13621 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13623 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13624 New "uid" fields to hold pretty-print IDs of group and ref.
13625 Memory references are now identified as <group_id>:<ref_id>
13626 instead of using [random] addresses.
13627 (dump_mem_details): Simplify, no functional change.
13628 (dump_mem_ref): Simplify and make output more concise.
13629 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13630 (find_or_create_group): Initialize group uid.
13631 (record_ref): Initialize ref uid. Improve debug output.
13632 (prune_group_by_reuse, should_issue_prefetch_p,)
13633 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13634 (mark_nontemporal_store, determine_loop_nest_reuse):
13635 Improve debug output.
13637 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13639 * dbgcnt.def (prefetch): New debug counter.
13640 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13641 (schedule_prefetches): Stop issueing prefetches if debug counter
13644 2017-06-06 Tom de Vries <tom@codesourcery.com>
13646 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13647 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13649 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13651 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13652 Use aarch64_reg_or_zero predicate for operand 4.
13653 (aarch64_compare_and_swap<mode> define_insn_and_split):
13654 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13655 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13657 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13659 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13660 (arm_compute_save_core_reg_mask): This.
13661 (thumb1_compute_save_reg_mask): Rename into ...
13662 (thumb1_compute_save_core_reg_mask): This.
13663 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13664 (arm_compute_frame_layout): Likewise.
13666 2017-06-06 Richard Biener <rguenther@suse.de>
13668 PR tree-optimization/80974
13669 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13670 keep or clear leaders SSA info.
13672 2017-06-06 Tom de Vries <tom@codesourcery.com>
13674 * config/nvptx/nvptx.c (split_mode_p): New function.
13675 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13677 2017-06-06 Tom de Vries <tom@codesourcery.com>
13679 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13681 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13684 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13687 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13689 * shrink-wrap.c (handle_simple_exit): Update profile.
13690 (try_shrink_wrapping): Upate profile.
13692 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13694 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13695 (tree_guess_outgoing_edge_probabilities): New.
13696 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13697 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13699 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13701 * ipa-split.c (split_function): Initialize return bb profile.
13703 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13705 * profile.c (compute_branch_probabilities): Also initialize
13706 EXIT_BLOCK profile.
13708 2017-06-06 Richard Biener <rguenther@suse.de>
13710 PR tree-optimization/80928
13711 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13712 (vect_analyze_loop_operations): Properly guard analysis for
13714 (vect_transform_loop): Likewise.
13715 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13716 (vect_model_induction_cost): Do not cost for pure SLP.
13717 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13718 of induction in inner loop vectorization.
13719 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13720 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13721 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13723 (vect_analyze_slp_cost_1): Cost induction.
13724 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13725 (vect_get_slp_vect_defs): Likewise.
13726 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13727 (vect_transform_stmt): Handle SLP reductions.
13728 * tree-vectorizer.h (vectorizable_induction): Adjust.
13730 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13732 * config/rs6000/rs6000.c (make_resolver_func): Update
13733 init_lowered_empty_function call.
13735 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13737 * doc/invoke.texi: Document the -fprofile-abs-path option.
13738 * common.opt (fprofile-abs-path): New option.
13739 * gcov-io.h (gcov_write_filename): Declare.
13740 * gcov-io.c (gcov_write_filename): New function.
13741 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13742 * profile.c (output_location): Likewise.
13744 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13746 * shring-wrap.c: Revert accidental commit.
13748 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13750 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13752 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13754 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13756 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13757 profile in callgraph edge.
13758 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13759 (apply_scale): Likewise.
13760 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13761 Also scale profile when inlining function with zero profile.
13762 (initialize_cfun): Update exit block profile even when it is zero.
13763 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13764 when profile is read.
13766 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13768 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13769 (CLONE_*): New constants to define the processors we can generate
13770 code for with the target_clone attribute.
13771 (rs6000_clone_map): New array to identify which clone processors
13772 the current program is running on.
13773 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13774 target_clone attribute.
13775 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13776 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13777 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13778 (cpu_expand_builtin): Add support for target_clone attribute.
13779 (rs6000_valid_attribute_p): Allow "default" attribute.
13780 (get_decl_name): New debug function to simplify printing the
13781 current function name in debugging statements.
13782 (rs6000_clone_priority): New functions to support the target_clone
13783 attribute, and be able to generate code to switch between ISA 2.05
13784 through ISA 3.0 (power6 through power9).
13785 (rs6000_compare_version_priority): Likewise.
13786 (rs6000_get_function_versions_dispatcher): Likewise.
13787 (make_resolver_func): Likewise.
13788 (add_condition_to_bb): Likewise.
13789 (dispatch_function_versions): Likewise.
13790 (rs6000_generate_version_dispatcher_body): Likewise.
13791 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13792 (fusion_gpr_load_p): Fix a spacing issue.
13793 * doc/extend.texi (Common Function Attributes): Document that the
13794 PowerPC supports the target_clone attribute.
13796 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13798 * config/arm/arm.h: explain F symbol found in description of ARM
13799 register allocation in its legend.
13801 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13803 * config/mips/frame-header-opt.c: Include profile-count.h.
13804 * config/riscv/riscv.c: Include profile-count.h
13806 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13808 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13810 (sm_set_flag_if_changed): Add bbs field.
13811 (execute_sm_if_changed_flag_set): Pass BBS.
13812 (execute_sm): Update.
13814 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13816 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13819 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13821 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13822 (peephole2): New peephole2 to emit the above.
13823 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13825 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13827 * config/aarch64/aarch64.c (define_peephole2 above
13828 *sub_<shift>_<mode>): New peephole.
13830 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13832 * config/i386/i386.c (make_resolver_func): Update.
13833 * Makefile.in: Add profile-count.h and profile-count.o
13834 * auto-profile.c (afdo_indirect_call): Update to new API.
13835 (afdo_set_bb_count): Update.
13836 (afdo_propagate_edge): Update.
13837 (afdo_propagate_circuit): Update.
13838 (afdo_calculate_branch_prob): Update.
13839 (afdo_annotate_cfg): Update.
13840 * basic-block.h: Include profile-count.h
13841 (struct edge_def): Turn count to profile_count.
13842 (struct basic_block_def): Likewie.
13843 (REG_BR_PROB_BASE): Move to profile-count.h
13844 (RDIV): Move to profile-count.h
13845 * bb-reorder.c (max_entry_count): Turn to profile_count.
13846 (find_traces): Update.
13847 (rotate_loop):Update.
13848 (connect_traces):Update.
13849 (sanitize_hot_paths):Update.
13850 * bt-load.c (migrate_btr_defs): Update.
13851 * cfg.c (RDIV): Remove.
13852 (init_flow): Use alloc_block.
13853 (alloc_block): Uninitialize count.
13854 (unchecked_make_edge): Uninitialize count.
13855 (check_bb_profile): Update.
13856 (dump_edge_info): Update.
13857 (dump_bb_info): Update.
13858 (update_bb_profile_for_threading): Update.
13859 (scale_bbs_frequencies_int): Update.
13860 (scale_bbs_frequencies_gcov_type): Update.
13861 (scale_bbs_frequencies_profile_count): New.
13862 * cfg.h (update_bb_profile_for_threading): Update.
13863 (scale_bbs_frequencies_profile_count): Declare.
13864 * cfgbuild.c (compute_outgoing_frequencies): Update.
13865 (find_many_sub_basic_blocks): Update.
13866 * cfgcleanup.c (try_forward_edges): Update.
13867 (try_crossjump_to_edge): Update.
13868 * cfgexpand.c (expand_gimple_tailcall): Update.
13869 (construct_exit_block): Update.
13870 * cfghooks.c (verify_flow_info): Update.
13871 (dump_bb_for_graph): Update.
13872 (split_edge): Update.
13873 (make_forwarder_block): Update.
13874 (duplicate_block): Update.
13875 (account_profile_record): Update.
13876 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13877 (get_estimated_loop_iterations): Update.
13878 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13879 (single_likely_exit): Update.
13880 * cfgloopmanip.c (scale_loop_profile): Update.
13882 (set_zero_probability): Update.
13883 (lv_adjust_loop_entry_edge): Update.
13884 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13885 (purge_dead_edges): Update.
13886 (rtl_account_profile_record): Update.
13887 * cgraph.c (cgraph_node::create): Uninitialize count.
13888 (symbol_table::create_edge): Uninitialize count.
13889 (cgraph_update_edges_for_call_stmt_node): Update.
13890 (cgraph_edge::dump_edge_flags): Update.
13891 (cgraph_node::dump): Update.
13892 (cgraph_edge::maybe_hot_p): Update.
13893 * cgraph.h: Include profile-count.h
13894 (create_clone), create_edge, create_indirect_edge): Update.
13895 (cgraph_node): Turn count to profile_count.
13896 (cgraph_edge0: Likewise.
13897 (make_speculative, clone): Update.
13898 (create_edge): Update.
13899 (init_lowered_empty_function): Update.
13900 * cgraphclones.c (cgraph_edge::clone): Update.
13901 (duplicate_thunk_for_node): Update.
13902 (cgraph_node::create_clone): Update.
13903 * cgraphunit.c (cgraph_node::analyze): Update.
13904 (cgraph_node::expand_thunk): Update.
13905 * final.c (dump_basic_block_info): Update.
13906 * gimple-streamer-in.c (input_bb): Update.
13907 * gimple-streamer-out.c (output_bb): Update.
13908 * graphite.c (print_global_statistics): Update.
13909 (print_graphite_scop_statistics): Update.
13910 * hsa-brig.c: Include basic-block.h.
13911 * hsa-dump.c: Include basic-block.h.
13912 * hsa-gen.c (T sum_slice): Update.
13913 (convert_switch_statements):Update.
13914 * hsa-regalloc.c: Include basic-block.h.
13915 * ipa-chkp.c (chkp_produce_thunks): Update.
13916 * ipa-cp.c (struct caller_statistics): Update.
13917 (init_caller_stats): Update.
13918 (gather_caller_stats): Update.
13919 (ipcp_cloning_candidate_p): Update.
13920 (good_cloning_opportunity_p): Update.
13921 (get_info_about_necessary_edges): Update.
13922 (dump_profile_updates): Update.
13923 (update_profiling_info): Update.
13924 (update_specialized_profile): Update.
13925 (perhaps_add_new_callers): Update.
13926 (decide_about_value): Update.
13927 (ipa_cp_c_finalize): Update.
13928 * ipa-devirt.c (struct odr_type_warn_count): Update.
13929 (struct decl_warn_count): Update.
13930 (struct final_warning_record): Update.
13931 (possible_polymorphic_call_targets): Update.
13932 (ipa_devirt): Update.
13933 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13934 * ipa-icf.c (sem_function::merge): Update.
13935 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13936 * ipa-inline.c (compute_uninlined_call_time): Update.
13937 (compute_inlined_call_time): Update.
13938 (want_inline_small_function_p): Update.
13939 (want_inline_self_recursive_call_p): Update.
13940 (edge_badness): Update.
13941 (lookup_recursive_calls): Update.
13942 (recursive_inlining): Update.
13943 (inline_small_functions): Update.
13944 (dump_overall_stats): Update.
13945 (dump_inline_stats): Update.
13946 * ipa-profile.c (ipa_profile_generate_summary): Update.
13947 (ipa_propagate_frequency): Update.
13948 (ipa_profile): Update.
13949 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13950 * ipa-utils.c (ipa_merge_profiles): Update.
13951 * loop-doloop.c (doloop_modify): Update.
13952 * loop-unroll.c (report_unroll): Update.
13953 (unroll_loop_runtime_iterations): Update.
13954 * lto-cgraph.c (lto_output_edge): Update.
13955 (lto_output_node): Update.
13956 (input_node): Update.
13957 (input_edge): Update.
13958 (merge_profile_summaries): Update.
13959 * lto-streamer-in.c (input_cfg): Update.
13960 * lto-streamer-out.c (output_cfg): Update.
13961 * mcf.c (create_fixup_graph): Update.
13962 (adjust_cfg_counts): Update.
13963 (sum_edge_counts): Update.
13964 * modulo-sched.c (sms_schedule): Update.
13965 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13966 * predict.c (maybe_hot_count_p): Update.
13967 (probably_never_executed): Update.
13968 (dump_prediction): Update.
13969 (combine_predictions_for_bb): Update.
13970 (propagate_freq): Update.
13971 (handle_missing_profiles): Update.
13972 (counts_to_freqs): Update.
13973 (rebuild_frequencies): Update.
13974 (force_edge_cold): Update.
13975 * predict.h: Include profile-count.h
13976 (maybe_hot_count_p, counts_to_freqs): UPdate.
13977 * print-rtl-function.c: Do not include cfg.h
13978 * print-rtl.c: Include basic-block.h
13979 * profile-count.c: New file.
13980 * profile-count.h: New file.
13981 * profile.c (is_edge_inconsistent): Update.
13982 (correct_negative_edge_counts): Update.
13983 (is_inconsistent): Update.
13984 (set_bb_counts): Update.
13985 (read_profile_edge_counts): Update.
13986 (compute_frequency_overlap): Update.
13987 (compute_branch_probabilities): Update; Initialize and deinitialize
13989 (branch_prob): Update.
13990 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13991 (edge_gcov_count): New.
13992 (bb_gcov_count): New.
13993 * shrink-wrap.c (try_shrink_wrapping): Update.
13994 * tracer.c (better_p): Update.
13995 * trans-mem.c (expand_transaction): Update.
13996 (ipa_tm_insert_irr_call): Update.
13997 (ipa_tm_insert_gettmclone_call): Update.
13998 * tree-call-cdce.c: Update.
13999 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14000 (gimple_duplicate_sese_tail): Update.
14001 (gimple_account_profile_record): Update.
14002 (execute_fixup_cfg): Update.
14003 * tree-inline.c (copy_bb): Update.
14004 (copy_edges_for_bb): Update.
14005 (initialize_cfun): Update.
14006 (freqs_to_counts): Update.
14007 (copy_cfg_body): Update.
14008 (expand_call_inline): Update.
14009 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14010 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14011 (try_unroll_loop_completely): Update.
14012 (try_peel_loop): Update.
14013 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14014 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14015 * tree-ssa-loop-split.c (connect_loops): Update.
14016 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14017 * tree-ssa-reassoc.c (branch_fixup): Update.
14018 * tree-ssa-tail-merge.c (replace_block_by): Update.
14019 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14020 (compute_path_counts): Update.
14021 (update_profile): Update.
14022 (recompute_probabilities): Update.
14023 (update_joiner_offpath_counts): Update.
14024 (estimated_freqs_path): Update.
14025 (freqs_to_counts_path): Update.
14026 (clear_counts_path): Update.
14027 (ssa_fix_duplicate_block_edges): Update.
14028 (duplicate_thread_path): Update.
14029 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14030 (struct switch_conv_info): Update.
14031 * tree-tailcall.c (decrease_profile): Update.
14032 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14033 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14034 * value-prof.c (check_counter): Update.
14035 (gimple_divmod_fixed_value): Update.
14036 (gimple_mod_pow2): Update.
14037 (gimple_mod_subtract): Update.
14038 (gimple_ic_transform): Update.
14039 (gimple_stringop_fixed_value): Update.
14040 * value-prof.h (gimple_ic): Update.
14042 2017-06-02 Carl Love <cel@us.ibm.com>
14044 * config/rs6000/rs6000-c: Add support for built-in functions
14045 vector double vec_doublee (vector signed int);
14046 vector double vec_doublee (vector unsigned int);
14047 vector double vec_doublee (vector float);
14048 vector double vec_doubleh (vector signed int);
14049 vector double vec_doubleh (vector unsigned int);
14050 vector double vec_doubleh (vector float);
14051 vector double vec_doublel (vector signed int);
14052 vector double vec_doublel (vector unsigned int);
14053 vector double vec_doublel (vector float);
14054 vector double vec_doubleo (vector signed int);
14055 vector double vec_doubleo (vector unsigned int);
14056 vector double vec_doubleo (vector float);.
14057 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14058 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14060 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14061 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14062 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14064 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14065 vec_doublel, vec_doubleh.
14066 * doc/extend.texi: Update the built-in documentation file for the
14067 new built-in functions.
14069 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14072 * ipa-inline-analysis.c (free_growth_caches): Set
14073 edge_removal_hook_holder to NULL after removing it.
14075 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14077 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14078 comparision with zero.
14080 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14082 for early expansion of vec_min and vec_max builtins.
14083 (builtin_function_type): Add min/max unsigned variants to those
14084 identified as having unsigned arguments.
14086 2017-06-02 Olivier Hainque <hainque@adacore.com>
14088 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14090 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14092 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14093 Use VALL_F16 iterator rather than VALL.
14095 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14097 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14098 Emit CBNZ inside loop when doing a strong exchange and comparing
14099 against zero. Generate the CC flags after the loop.
14101 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14103 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14104 (dl_section_ref): New.
14105 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14106 On AIX, append an expression to subtract the size of the
14107 section length to dl_section_ref.
14109 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14111 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14112 for early expansion of vector absolute builtins.
14114 2017-06-02 Richard Biener <rguenther@suse.de>
14116 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14117 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14119 2017-06-02 Richard Biener <rguenther@suse.de>
14121 PR tree-optimization/80948
14122 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14123 stmt order as well.
14125 2017-06-02 Richard Biener <rguenther@suse.de>
14127 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14129 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14130 for inductions as relevant.
14132 2017-06-02 Richard Biener <rguenther@suse.de>
14134 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14135 (vectorizable_induction): ... this. Remove dead code.
14137 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14139 * builtins. (expand_builtin_alloca): Remove second parameter and
14140 infer its value from the first parameter instead.
14141 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14143 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14145 PR rtl-optimization/80903
14146 * loop-doloop.c (add_test): Unshare sequence.
14148 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14150 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14152 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14154 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14156 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14157 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14158 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14159 (xlogue_layout::get_stub_rtx): Make static.
14160 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14161 (xlogue_layout::compute_stub_managed_regs): Rename to...
14162 (xlogue_layout::count_stub_managed_regs): ...this.
14163 (xlogue_layout::is_stub_managed_reg): New function.
14164 (xlogue_layout::m_stub_names): Rename to...
14165 (xlogue_layout::s_stub_names): ...this, make static.
14166 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14167 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14168 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14169 xlogue_layout::s_stub_names): Instantiate statics.
14170 (stub_managed_regs): Remove.
14171 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14172 (disable_call_ms2sysv_xlogues): Rename to...
14173 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14174 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14176 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14177 change after reload_completed.
14178 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14180 (ix86_expand_prologue): Likewise.
14181 (ix86_expand_epilogue): Likewise.
14182 (ix86_expand_split_stack_prologue): Likewise.
14183 (ix86_compute_frame_layout): Remove frame parameter ...
14184 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14185 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14187 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14188 (ix86_frame): Move from here ...
14189 * config/i386/i386.h (ix86_frame): ... to here.
14190 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14191 complete ix86_frame data structure instead. Remove some_ld_name.
14193 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14195 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14196 symbols that hold a DECL_VALUE_EXPR.
14198 2017-06-01 Martin Jambor <mjambor@suse.cz>
14200 PR tree-optimization/80898
14201 * tree-sra.c (process_subtree_disqualification): Removed.
14202 (disqualify_candidate): Do not acll
14203 process_subtree_disqualification.
14204 (subtree_mark_written_and_enqueue): New function.
14205 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14206 RHS has been disqualified and re-queue LHS if necessary. Apart
14207 from that, ignore disqualified RHS.
14209 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14211 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14212 address fetch for z10 or later.
14214 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14216 * config/arc/arc.md (tst_movb): Add guard when splitting.
14218 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14220 * config/arc/arc.c (arc_can_eliminate): Test against
14221 arc_frame_pointer_needed.
14223 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14225 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14226 to prevent store reordering.
14227 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14228 (type): Add block type.
14229 (stack_tie): Define special instruction to be used in
14232 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14234 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14235 constraint. It is not valid for the pattern.
14236 (noncommutative_binary_comparison): Likewise.
14238 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14240 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14243 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14245 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14246 be used by the reg-alloc.
14248 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14250 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14251 reg-alloc when having mul64 or mul32x16 instructions.
14252 (mulsidi3): Likewise.
14253 (umulsidi3): Likewise.
14254 (mulsi32x16): New pattern.
14255 (mulsi64): Likewise.
14256 (mulsidi64): Likewise.
14257 (umulsidi64): Likewise.
14258 (MUL32x16_REG): Define.
14259 (mul64_600): Use MUL32x16_REG.
14260 (mac64_600): Likewise.
14261 (umul64_600): Likewise.
14262 (umac64_600): Likewise.
14264 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14266 * config/arc/arc.md (mulsi3_700): Make it commutative.
14268 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14270 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14272 (*sign_extendsidi2_insn): Likewise for movstosw.
14274 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14276 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14277 the type of the input discriminant value. Convert the
14278 discriminant value of signedness vary.
14280 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14282 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14283 Add to -Wall section.
14285 2017-06-01 Richard Biener <rguenther@suse.de>
14287 PR middle-end/66313
14288 * fold-const.c (fold_plusminus_mult_expr): If the factored
14289 factor may be zero use a wrapping type for the inner operation.
14290 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14291 and handle moved defs.
14292 (process_assignment): Properly guard the unary op case. Return a
14293 tri-state indicating that moving the stmt before the call may allow
14294 to continue. Pass through to_move.
14295 (find_tail_calls): Handle moving unrelated defs before
14298 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14301 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14302 splitter result in the canonical way.
14304 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14306 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14307 also for 32bit target. Update insn attributes.
14308 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14310 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14312 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14313 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14314 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14315 (_mm_maskz_min_ss): New intrinsics.
14317 2017-05-31 Martin Liska <mliska@suse.cz>
14319 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14320 Change comment style to one we normally use.
14321 (vectorizable_reduction): Likewise.
14322 (vectorizable_induction): Likewise.
14323 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14324 (vectorizable_call): Likewise.
14325 (vectorizable_simd_clone_call): Likewise.
14326 (vectorizable_conversion): Likewise.
14327 (vectorizable_assignment): Likewise.
14328 (vectorizable_shift): Likewise.
14329 (vectorizable_operation): Likewise.
14330 (vectorizable_store): Likewise.
14331 (vectorizable_load): Likewise.
14332 * tree-vectorizer.h: Likewise.
14334 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14336 * passes.c (emergency_dump_function): New.
14337 * tree-pass.h (emergency_dump_function): Declare.
14338 * plugin.c (plugins_internal_error_function): Remove.
14339 * plugin.h (plugins_internal_error_function): Remove declaration.
14340 * toplev.c (internal_error_function): New static function. Use it...
14341 (general_init): ...here.
14343 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14345 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14346 (arc_rtx_costs): Add costs for new patterns.
14347 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14348 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14350 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14352 * tree-ssa-strlen.c (get_next_strinfo): New function.
14353 (get_stridx_plus_constant): Use it.
14354 (zero_length_string): Likewise.
14355 (adjust_related_strinfos): Likewise.
14356 (adjust_last_stmt): Likewise.
14358 2017-05-31 Richard Biener <rguenther@suse.de>
14361 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14362 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14364 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14366 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14367 loop_vinfo argument and use of dependence distance vectors.
14368 Check instead whether the two references differ only in their
14369 initial value and assume that they have the same alignment if the
14370 difference is a multiple of the vector alignment.
14371 (vect_analyze_data_refs_alignment): Update call accordingly.
14373 2017-05-31 Martin Liska <mliska@suse.cz>
14376 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14378 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14380 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14381 (create_intersect_range_checks): Move from ...
14382 * tree-data-ref.c (create_intersect_range_checks_index)
14383 (create_intersect_range_checks): ... to here.
14384 (create_runtime_alias_checks): New function factored from ...
14385 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14386 here. Call above function.
14387 * tree-data-ref.h (create_runtime_alias_checks): New function.
14389 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14391 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14392 segment length for dr_b and compute it in wide_int.
14394 2017-05-31 Richard Biener <rguenther@suse.de>
14396 PR tree-optimization/80906
14397 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14398 and pass through iv_map.
14399 (copy_bb_and_scalar_dependences): Adjust.
14400 (translate_pending_phi_nodes): Likewise.
14401 (copy_loop_close_phi_args): Handle code-generating IVs instead
14404 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14406 * diagnostic-color.c (color_dict): Add "type-diff".
14407 (parse_gcc_colors): Update comment.
14408 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14409 -fdiagnostics-show-template-tree and -fno-elide-type.
14410 (GCC_COLORS): Add type-diff to example.
14412 (-fdiagnostics-show-template-tree): New.
14413 (-fno-elide-type): New.
14414 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14415 the pp_format_decoder callback. Call any m_format_postprocessor's
14417 (pretty_printer::pretty_printer): Initialize
14418 m_format_postprocessor.
14419 (pretty_printer::~pretty_printer): Delete any
14420 m_format_postprocessor.
14421 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14422 (class format_postprocessor): New class.
14423 (struct pretty_printer::format_decoder): Document the new parameters.
14424 (struct pretty_printer::m_format_postprocessor): New field.
14425 * tree-diagnostic.c (default_tree_printer): Update for new
14426 bool and const char ** params.
14427 * tree-diagnostic.h (default_tree_printer): Likewise.
14429 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14431 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14432 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14433 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14434 rs6000_gen_cell_microcode code.
14435 (rs6000_final_prescan_insn): Delete.
14436 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14437 "warn-cell-microcode" entries.
14438 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14439 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14440 throughout. Change cc_reg_not_micro_cr0_operand to
14441 cc_reg_not_cr0_operand throughout.
14442 (*extendhi<mode>2_noload): Delete.
14443 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14444 (mwarn-cell-microcode): Delete.
14445 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14446 -mgen-cell-microcode and -mwarn-cell-microcode.
14448 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14451 * config/i386/constraints.md (Yd): New constraint.
14453 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14454 and (?Yd, r) alternatives. Update insn attributes.
14455 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14456 and (?*Yd, r) alternatives. Update insn attributes.
14457 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14459 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14461 * gimplify.c (gimplify_modify_expr): Don't create a
14462 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14465 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14467 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14469 2017-05-30 Richard Biener <rguenther@suse.de>
14471 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14472 and reduc_def fields.
14473 (STMT_VINFO_REDUC_TYPE): New define.
14474 (STMT_VINFO_REDUC_DEF): Likewise.
14475 (vect_force_simple_reduction): Adjust prototype.
14476 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14477 (vect_is_simple_reduction): Remove check_reduction argument.
14478 (vect_force_simple_reduction): Adjust and set
14479 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14480 (vectorizable_reduction): Do not re-do reduction analysis
14481 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14482 * tree-parloops.c (gather_scalar_reductions): Adjust.
14484 2017-05-30 Richard Biener <rguenther@suse.de>
14486 PR middle-end/80901
14487 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14490 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14492 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14493 Introduce unknown_misalignment parameter and remove vf.
14494 (vect_peeling_hash_get_lowest_cost):
14495 Pass unknown_misalignment parameter.
14496 (vect_enhance_data_refs_alignment):
14497 Fix unsupportable data ref treatment.
14499 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14501 * tree-vect-data-refs.c (vect_get_data_access_cost):
14502 Workaround for SLP handling.
14503 (vect_enhance_data_refs_alignment):
14504 Compute costs for doing no peeling at all, compare to the best
14505 peeling costs so far and avoid peeling if cheaper.
14507 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14509 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14510 Return peeling info and set costs to zero for unlimited cost
14512 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14513 with unknown misalignment. Compute and costs for unknown
14514 misalignment, compare them to the costs for known misalignment
14515 and choose the cheapest for peeling.
14517 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14519 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14520 (vect_get_peeling_costs_all_drs): Create function.
14521 (vect_peeling_hash_get_lowest_cost):
14522 Use vect_get_peeling_costs_all_drs.
14523 (vect_peeling_supportable): Create function.
14524 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14526 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14528 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14529 DR_HAS_NEGATIVE_STEP.
14530 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14531 (vect_enhance_data_refs_alignment): Use.
14532 (vect_duplicate_ssa_name_ptr_info): Use.
14533 * tree-vectorizer.h (dr_misalignment): Use.
14534 (known_alignment_for_access_p): Use.
14536 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14539 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14541 (has_section_name): New function.
14543 2017-05-30 Martin Liska <mliska@suse.cz>
14546 * auto-profile.c (get_function_decl_from_block): Fix
14549 2017-05-30 Richard Biener <rguenther@suse.de>
14551 PR middle-end/80876
14552 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14554 2017-05-30 Martin Liska <mliska@suse.cz>
14556 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14557 * dumpfile.h (struct dump_file_info): Remove ctors.
14559 2017-05-30 Martin Liska <mliska@suse.cz>
14561 * predict.def: Fix GNU coding style.
14563 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14565 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14566 Mark 'to' argument with ATTRIBUTE_UNUSED.
14568 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14570 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14571 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14572 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14575 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14577 * doc/install.texi (Options specification): Restore entry of
14578 --enable-sjlj-exceptions.
14580 2017-05-27 Michael Eager <eager@eagercon.com>
14583 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14585 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14587 * config/microblaze/microblaze.h
14588 (FIXED_REGISTERS): Update in macro.
14589 (CALL_USED_REGISTERS): Update in macro.
14591 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14593 * doc/install.texi: Add links to macOS binary distributions.
14595 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14599 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14601 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14603 2017-05-26 Martin Liska <mliska@suse.cz>
14605 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14607 2017-05-26 Martin Liska <mliska@suse.cz>
14609 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14610 always leading ';; '.
14611 (dump_bb_info): Likewise.
14612 (brief_dump_cfg): Likewise.
14613 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14614 * dumpfile.c: Remove usage of TDF_VERBOSE.
14615 * dumpfile.h (enum dump_kind): Likewise.
14616 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14617 * print-tree.c (debug_verbose): Remove.
14618 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14619 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14620 * tree-diagnostic.c (default_tree_printer): Replace
14621 TDF_DIAGNOSTIC with TDF_SLIM.
14623 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14625 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14626 in parameter loop, rather than loop_vinfo.
14627 (create_intersect_range_checks): Ditto.
14628 (vect_create_cond_for_alias_checks): Update call to above functions.
14630 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14632 PR tree-optimization/80815
14633 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14634 for merging runtime alias checks. Handle negative DR_STEPs.
14636 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14638 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14640 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14641 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14642 out code pruning runtime alias checks.
14643 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14644 factored out from above.
14645 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14647 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14649 (prune_runtime_alias_test_list): New decalaration.
14651 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14653 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14654 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14655 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14656 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14657 (operator==, comp_dr_with_seg_len_pair): Ditto.
14658 (vect_prune_runtime_alias_test_list): Ditto.
14660 2017-05-26 Martin Liska <mliska@suse.cz>
14663 * params.def: Bound partial-inlining-entry-probability param.
14665 2017-05-26 Marek Polacek <polacek@redhat.com>
14668 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14671 2017-05-26 Richard Biener <rguenther@suse.de>
14673 PR tree-optimization/80842
14674 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14677 2017-05-26 Richard Biener <rguenther@suse.de>
14679 PR tree-optimization/80844
14680 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14682 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14684 * doc/md.texi (Machine Constraints): Update x86 family
14685 machine constraints section to match 'config/i386/constraints.md'.
14687 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14689 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14691 2017-05-25 Nathan Sidwell <nathan@acm.org>
14693 * doc/invoke.texi (--enable-languages): Update documentation.
14695 2017-05-25 Martin Liska <mliska@suse.cz>
14697 * dumpfile.c: Add TDF_FOLDING.
14698 * dumpfile.h (enum dump_kind): Likewise.
14699 * genmatch.c (dt_simplify::gen_1): Use it.
14701 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14703 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14705 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14707 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14708 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14710 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14712 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14713 * match.pd (X == C): Rewrite it here.
14714 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14715 with_certain_nonzero_bits2): New predicates.
14716 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14718 2017-05-24 Nathan Sidwell <nathan@acm.org>
14720 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14723 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14726 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14728 * config/powerpcspe: New port. Files are copied from the rs6000
14729 port, with "rs6000" in filenames replaced by "powerpcspe".
14731 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14733 PR rtl-optimization/80754
14734 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14736 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14738 * config/sparc/sparc.md (length): Return the correct value for -mflat
14739 sibcalls to match output_sibcall.
14741 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14745 * config/rs6000/rs6000.c (struct machine_function): Add new field
14747 (rs6000_get_separate_components): Init that field, use it.
14748 (rs6000_components_for_bb): Use the field.
14750 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14752 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14754 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14756 PR middle-end/80823
14757 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14759 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14762 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14763 address against address_operand predicate.
14764 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14766 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14768 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14769 parameters passed indirectly.
14771 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14773 * config/i386/i386.md (*movdi_internal): Remove SSE4
14774 alternative 18 (?r, *v). Update insn attributes.
14775 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14776 Update insn attributes.
14777 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14778 Update insn attributes.
14779 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14780 alternative 1 (r, v). Remove isa attribute.
14781 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14782 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14783 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14785 2017-05-23 Tom de Vries <tom@codesourcery.com>
14787 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14790 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14792 * cgraphunit.c (symbol_table::process_new_functions): Update.
14793 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14794 (inline_generate_summary): Rename to ...
14795 (ipa_fn_summary_generate): ... this one.
14796 (inline_read_summary): Rename to ...
14797 (ipa_fn_summary_read): ... this one.
14798 (inline_write_summary): Rename to ...
14799 (ipa_fn_summary_write): ... this one.
14800 (inline_free_summary): Rename to ...
14801 (ipa_free_fn_summary): ... this one.
14802 (pass_data_local_fn_summary, pass_local_fn_summary,
14803 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14804 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14805 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14806 make_pass_ipa_fn_summary): New.
14807 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14808 inline_write_summary, inline_free_summary): Remove.
14809 (ipa_free_fn_summary) : New.
14810 * ipa-inline.c (ipa_inline): Update.
14811 (pass_ipa_inline): Do not generate summaries.
14812 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14814 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14815 and add pass_ipa_fn_summary.
14816 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14818 (make_pass_inline_parameters): Remove.
14820 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14822 * omp-low.c (struct omp_context): Remove "default_kind" member.
14825 * omp-offload.c (execute_oacc_device_lower): Remove the
14826 parallelism dimensions function attributes for unparallelized
14827 OpenACC kernels constructs.
14829 2017-05-23 Martin Liska <mliska@suse.cz>
14831 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14833 (cgraph_edge::make_speculative): Likewise.
14834 (cgraph_edge::resolve_speculation): Likewise.
14835 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14836 (cgraph_node::dump): Likewise.
14837 * cgraph.h: Likewise.
14838 * cgraphunit.c (analyze_functions): Likewise.
14839 (symbol_table::compile): Likewise.
14840 * ipa-cp.c (print_all_lattices): Likewise.
14841 (determine_versionability): Likewise.
14842 (initialize_node_lattices): Likewise.
14843 (ipcp_verify_propagated_values): Likewise.
14844 (estimate_local_effects): Likewise.
14845 (update_profiling_info): Likewise.
14846 (create_specialized_node): Likewise.
14847 (perhaps_add_new_callers): Likewise.
14848 (decide_about_value): Likewise.
14849 (decide_whether_version_node): Likewise.
14850 (identify_dead_nodes): Likewise.
14851 (ipcp_store_bits_results): Likewise.
14852 * ipa-devirt.c (dump_targets): Likewise.
14853 (ipa_devirt): Likewise.
14854 * ipa-icf.c (sem_item::dump): Likewise.
14855 (sem_function::equals): Likewise.
14856 (sem_variable::equals): Likewise.
14857 (sem_item_optimizer::read_section): Likewise.
14858 (sem_item_optimizer::execute): Likewise.
14859 (congruence_class::dump): Likewise.
14860 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14861 (dump_inline_summary): Likewise.
14862 (estimate_node_size_and_time): Likewise.
14863 (inline_analyze_function): Likewise.
14864 * ipa-inline-transform.c (inline_call): Likewise.
14865 * ipa-inline.c (report_inline_failed_reason): Likewise.
14866 (want_early_inline_function_p): Likewise.
14867 (edge_badness): Likewise.
14868 (update_edge_key): Likewise.
14869 (inline_small_functions): Likewise.
14870 * ipa-profile.c (ipa_profile): Likewise.
14871 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14872 (ipa_make_edge_direct_to_target): Likewise.
14873 (remove_described_reference): Likewise.
14874 (ipa_impossible_devirt_target): Likewise.
14875 (propagate_controlled_uses): Likewise.
14876 (ipa_print_node_params): Likewise.
14877 (ipcp_transform_function): Likewise.
14878 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14879 (propagate_pure_const): Likewise.
14880 * ipa-reference.c (generate_summary): Likewise.
14881 (read_write_all_from_decl): Likewise.
14882 (propagate): Likewise.
14883 (ipa_reference_read_optimization_summary): Likewise.
14884 * ipa-utils.c (ipa_merge_profiles): Likewise.
14885 * ipa.c (walk_polymorphic_call_targets): Likewise.
14886 (symbol_table::remove_unreachable_nodes): Likewise.
14887 (ipa_single_use): Likewise.
14888 * passes.c (execute_todo): Likewise.
14889 * predict.c (drop_profile): Likewise.
14890 * symtab.c (symtab_node::get_dump_name): New function.
14891 (symtab_node::dump_name): Likewise.
14892 (symtab_node::dump_asm_name): Likewise.
14893 (symtab_node::dump_references): Likewise.
14894 (symtab_node::dump_referring): Likewise.
14895 (symtab_node::dump_base): Likewise.
14896 (symtab_node::debug_symtab): Likewise.
14897 * tree-sra.c (convert_callers_for_node): Likewise.
14898 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14899 * value-prof.c (init_node_map): Likewise.
14901 2017-05-23 Martin Liska <mliska@suse.cz>
14903 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14904 and symtab_node::debug_symtab to symbol_table::debug.
14905 * cgraphunit.c (analyze_functions): Use the renamed function.
14906 (symbol_table::compile): Likewise.
14907 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14908 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14909 * passes.c (execute_todo): Likewise.
14910 * symtab.c (symbol_table::dump): New function.
14911 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14913 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14915 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14916 that nonconst implies exec.
14918 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14920 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14921 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14922 (inline_edge_summary_vec): Turn into ...
14923 (ipa_call_summaries): ... this one.
14924 (redirect_to_unreachable, edge_set_predicate,
14925 evaluate_properties_for_edge, inline_summary_alloc,
14926 reset_ipa_call_summary, reset_inline_summary,
14927 inline_summary_t::duplicate): Update.
14928 (inline_edge_duplication_hook): Turn to ...
14929 (ipa_call_summary_t::duplicate): ... this one.
14930 (inline_edge_removal_hook): Turn to ...
14931 (ipa_call_summary_t::remove): ... this one.
14932 (dump_inline_edge_summary): Turn to ...
14933 (dump_ipa_call_summary): ... this one.
14934 (estimate_function_body_sizes): Update.
14935 (inline_update_callee_summaries): Update.
14936 (remap_edge_change_prob): Update.
14937 (remap_edge_summaries): Update.
14938 (inline_merge_summary): Update.
14939 (do_estimate_edge_time): Update.
14940 (inline_generate_summary): Update.
14941 (inline_read_section): Update.
14942 (inline_read_summary): Update.
14943 (inline_free_summary): Update.
14944 * ipa-inline.c (can_inline_edge_p): Update.
14945 (compute_inlined_call_time): Update.
14946 (want_inline_small_function_p): Update.
14947 (edge_badness): Update.
14948 (early_inliner): Update.
14949 * ipa-inline.h (inline_edge_summary): Turn to ...
14950 (ipa_call_summary): ... this one.
14951 (ipa_call_summary_t): New class.
14952 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14953 (ipa_call_summaries): New.
14954 (inline_edge_summary): Remove.
14955 (estimate_edge_growth): Update.
14956 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14957 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14958 * ipa-split.c (execute_split_functions): Update.
14959 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14961 2017-05-23 Tom de Vries <tom@codesourcery.com>
14963 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14964 attributes): Document rdrand effective target.
14966 2017-05-23 Tom de Vries <tom@codesourcery.com>
14968 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14969 attributes): Sort alphabetically.
14971 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14973 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14975 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14978 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14979 V2DF/V2DI splat into two separate patterns, one that handles
14980 registers, and the other that only handles memory. Drop support
14981 for splatting from a GPR on ISA 2.07 and then splitting the
14982 splat into direct move and splat.
14983 (vsx_splat_<mode>_reg): Likewise.
14984 (vsx_splat_<mode>_mem): Likewise.
14986 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14988 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14990 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14992 PR middle-end/80809
14993 * omp-low.c (finish_taskreg_remap): New function.
14994 (finish_taskreg_scan): If unit size of ctx->record_type
14995 is non-constant, unshare the size expression and replace
14996 decls in it with possible outer var refs.
14998 PR middle-end/80809
14999 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15000 GOVD_SHARED rather than GOVD_PRIVATE with it.
15001 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15002 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15004 PR middle-end/80853
15005 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15006 as last argument to build_outer_var_ref for pointer bases of array
15007 section reductions.
15009 2017-05-19 Martin Sebor <msebor@redhat.com>
15011 * print-tree.c (print_node): Print DECL_READ_P flag.
15013 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15015 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15016 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15017 * cgraph.c: Likewise.
15018 * cgraphunit.c: Likewise.
15019 * gengtype.c: Likewise.
15020 * ipa-cp.c: Likewise.
15021 * ipa-devirt.c: Likewise.
15022 * ipa-icf.c: Likewise.
15023 * ipa-predicate.c: Likewise.
15024 * ipa-profile.c: Likewise.
15025 * ipa-prop.c: Likewise.
15026 * ipa-split.c: Likewise.
15028 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15029 edge_predicate_pool, dump_inline_hints,
15030 inline_summary::account_size_time, redirect_to_unreachable,
15031 edge_set_predicate, set_hint_predicate,
15032 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15033 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15034 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15035 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15036 ipa_call_summary_t::remove, initialize_growth_caches,
15037 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15038 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15039 mark_modified, unmodified_parm_1, unmodified_parm,
15040 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15041 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15042 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15043 will_be_nonconstant_predicate, record_modified_bb_info,
15044 get_minimal_bb, record_modified, param_change_prob,
15045 phi_result_unknown_predicate, predicate_for_phi_result,
15046 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15047 estimate_function_body_sizes, compute_inline_parameters,
15048 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15049 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15050 inline_update_callee_summaries, remap_edge_change_prob,
15051 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15052 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15053 inline_analyze_function, inline_summary_t::insert,
15054 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15055 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15056 inline_free_summary): Move to ipa-fnsummary.h
15057 (predicate_t): Remove.
15058 * ipa-fnsummary.c: New file.
15059 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15060 (enum inline_hints_vals, inline_hints, agg_position_info,
15061 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15062 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15063 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15064 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15065 inline_read_summary, inline_write_summary, inline_free_summary,
15066 inline_analyze_function, initialize_inline_failed,
15067 inline_merge_summary, inline_update_overall_summary,
15068 compute_inline_parameters): Move to ipa-fnsummary.h
15069 * ipa-fnsummary.h: New file.
15070 * ipa-inline-transform.h: Include ipa-inline.h.
15071 * ipa-inline.c: LIkewise.
15073 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15075 * ipa-inline.c (edge_badness): Use inlined_time instead of
15076 inline_summaries->get.
15078 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15080 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15082 2017-05-22 Nathan Sidwell <nathan@acm.org>
15084 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15085 (fdump-lang): Document 'raw' option.
15086 * dumpfile.h (TDI_tu): Delete.
15087 * dumpfile.c (dump_files): Remove translation-unit.
15088 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15090 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15092 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15093 command option from $(AWK) call.
15094 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15095 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15096 [FORMAT]: Remove handling of variable.
15097 * config/avr/t-multilib: Regenerate.
15099 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15101 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15103 (dump_inline_summary): Do not print self_time.
15104 (estimate_function_body_sizes): Do not set self_time.
15105 (compute_inline_parameters): Likewise.
15106 (inline_read_section, inline_write_summary): Do not stream self_time.
15107 * ipa-inline.h (inline_summary): Drop self_time.
15109 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15111 * ipa-inline-analysis.c (account_size_time): Rename to ...
15112 (inline_summary::account_size_time): ... this one.
15113 (reset_ipa_call_summary): Turn to ...
15114 (ipa_call_summary::reset): ... this one.
15115 (reset_inline_summary): Turn to ...
15116 (inline_summary::reset): ... this one.
15117 (inline_summary_t::remove): Update.
15118 (inline_summary_t::duplicate): Update.
15119 (ipa_call_summary_t::remove): Update.
15120 (dump_inline_summary): Update.
15121 (estimate_function_body_sizes): Update.
15122 (compute_inline_parameters): Update.
15123 (estimate_node_size_and_time): Update.
15124 (inline_merge_summary): Update.
15125 (inline_update_overall_summary): Update.
15126 (inline_read_section): Update.
15127 (inline_write_summary): Update.
15128 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15129 add account_size_time and reset member functions.
15130 (ipa_call_summary): Add reset function.
15131 * ipa-predicate.h (predicate::operator &): Constify.
15133 2017-05-22 Richard Biener <rguenther@suse.de>
15135 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15137 2017-05-19 Jason Merrill <jason@redhat.com>
15139 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15141 2017-05-19 Marek Polacek <polacek@redhat.com>
15144 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15145 TYPE_OVERFLOW_WRAPS checks.
15147 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15149 * tree-core.h (enum omp_clause_default_kind): Add
15150 "OMP_CLAUSE_DEFAULT_PRESENT".
15151 * tree-pretty-print.c (dump_omp_clause): Handle it.
15152 * gimplify.c (enum gimplify_omp_var_data): Add
15153 "GOVD_MAP_FORCE_PRESENT".
15154 (gimplify_adjust_omp_clauses_1): Map it to
15155 "GOMP_MAP_FORCE_PRESENT".
15156 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15158 * gimplify.c (oacc_default_clause): Clarify.
15160 2017-05-19 Nathan Sidwell <nathan@acm.org>
15162 LANG_HOOK_REGISTER_DUMPS
15163 * toplev.c (general_init): Call register dump lang hook.
15164 * doc/invoke.texi: Document -fdump-lang option family.
15165 * dumpfile.c (dump_files): Remove class dump here.
15166 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15167 * dumpfile.h (tree_dump_index): Remove TDI_class.
15168 * langhooks-def.h (lhd_register_dumps): Declare.
15169 (LANG_HOOKS_REGISTER_DUMPS): Define.
15170 (LANG_HOOKS_INITIALIZER): Add it.
15171 * langhooks.c (lhd_register_dumps): Define.
15172 * langhooks.h (struct lang_hooks): Add register_dumps.
15174 2017-05-19 Nathan Sidwell <nathan@acm.org>
15176 * context.h (context::set_passes): New.
15177 * context.c (context::context): Do not create pass manager.
15178 * toplev.c (general_init): Create pass manager here.
15180 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15182 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15183 use this splitter if two add or or instructions would also work for
15184 the constant we want to generate.
15186 2017-05-19 Richard Biener <rguenther@suse.de>
15189 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15190 predicate evaluation.
15192 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15194 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15196 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15199 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15201 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15202 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15203 (inline_edge_summary_vec): Turn into ...
15204 (ipa_call_summaries): ... this one.
15205 (redirect_to_unreachable, edge_set_predicate,
15206 evaluate_properties_for_edge, inline_summary_alloc,
15207 reset_ipa_call_summary, reset_inline_summary,
15208 inline_summary_t::duplicate): Update.
15209 (inline_edge_duplication_hook): Turn to ...
15210 (ipa_call_summary_t::duplicate): ... this one.
15211 (inline_edge_removal_hook): Turn to ...
15212 (ipa_call_summary_t::remove): ... this one.
15213 (dump_inline_edge_summary): Turn to ...
15214 (dump_ipa_call_summary): ... this one.
15215 (estimate_function_body_sizes): Update.
15216 (inline_update_callee_summaries): Update.
15217 (remap_edge_change_prob): Update.
15218 (remap_edge_summaries): Update.
15219 (inline_merge_summary): Update.
15220 (do_estimate_edge_time): Update.
15221 (inline_generate_summary): Update.
15222 (inline_read_section): Update.
15223 (inline_read_summary): Update.
15224 (inline_free_summary): Update.
15225 * ipa-inline.c (can_inline_edge_p): Update.
15226 (compute_inlined_call_time): Update.
15227 (want_inline_small_function_p): Update.
15228 (edge_badness): Update.
15229 (early_inliner): Update.
15230 * ipa-inline.h (inline_edge_summary): Turn to ...
15231 (ipa_call_summary): ... this one.
15232 (ipa_call_summary_t): New class.
15233 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15234 (ipa_call_summaries): New.
15235 (inline_edge_summary): Remove.
15236 (estimate_edge_growth): Update.
15237 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15238 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15239 * ipa-split.c (execute_split_functions): Update.
15240 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15242 2017-05-19 Richard Biener <rguenther@suse.de>
15244 PR middle-end/80764
15245 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15247 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15249 * config/rs6000/rs6000.c (struct machine_function): Add field
15250 fpr_is_wrapped_separately.
15251 (rs6000_get_separate_components): Use 64 components. Handle the
15252 new FPR components.
15253 (rs6000_components_for_bb): Handle the FPR components.
15254 (rs6000_emit_prologue_components): Handle the FPR components.
15255 (rs6000_emit_epilogue_components): Handle the FPR components.
15256 (rs6000_set_handled_components): Handle the FPR components.
15257 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15258 that are already separately shrink-wrapped.
15259 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15260 that are already separately shrink-wrapped.
15262 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15265 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15268 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15269 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15270 cases where the register allocator uses a move and an offsettable
15271 memory operation to/from a FPR register on ISA 2.06/2.07.
15272 (define_peephole2 for Altivec d-form store): Likewise.
15274 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15277 * config/i386/mmx.md (*mov<mode>_internal): Enable
15278 alternatives 11, 12, 13 and 14 also for 32bit targets.
15279 Remove alternatives 15, 16, 17 and 18.
15280 * config/i386/sse.md (vec_concatv2di): Change
15281 alternative (!x, *y) to (x, ?!*Yn).
15283 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15285 * dumpfile.h (enum dump_kind): Remove stray comma.
15287 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15289 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15290 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15291 predicate::num_conditions
15292 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15293 (CHANGED): turn into predicate::changed.
15294 (agg_position_info): Move to ipa-predicate.h
15295 (add_condition, predicate::add_clause, predicate::operator &=,
15296 predicate::or_with, predicate::evaluate, predicate::probability,
15297 dump_condition, dump_clause, predicate::dump,
15298 predicate::remap_after_duplication, predicate::remap_after_inlining,
15299 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15300 (evaluate_conditions_for_known_args): Update.
15301 (set_cond_stmt_execution_predicate): Update.
15302 * ipa-inline.h: Include ipa-predicate.h
15303 (condition, inline_param_summary, conditions, agg_position_info,
15304 predicate): Move to ipa-predicate.h
15305 * ipa-predicate.c: New file.
15306 * ipa-predicate.h: New file.
15308 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15310 * final.c (leaf_function_p): Check we are not in a sequence.
15312 2017-05-18 Martin Liska <mliska@suse.cz>
15314 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15315 * dumpfile.c (dump_register): Use new enum dump_kind.
15316 (get_dump_file_name): Likewise.
15317 (dump_enable_all): Likewise.
15318 (dump_switch_p_1): Likewise.
15319 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15320 * dumpfile.h (enum dump_kind): New enum type.
15321 (struct dump_file_info): Create constructor and
15322 format fields and comments.
15323 * passes.c (pass_manager::register_one_dump_file):
15325 * statistics.c (statistics_early_init): Likewise.
15326 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15327 TDF_TREE with TDF_SLIM.
15328 (gather_memory_references_ref): Likewise.
15330 2017-05-18 Martin Liska <mliska@suse.cz>
15332 * vec.h (struct vnull): Use it.
15334 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15336 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15337 (true_predicate, false_predicate, true_predicate_p,
15338 false_predicate_p): Remove.
15339 (single_cond_predicate, not_inlined_predicate): Turn to member function
15341 (add_condition): Update.
15342 (add_clause): Turn to...
15343 (predicate::add_clause): ... this one; update; allow passing NULL
15345 (and_predicates): Turn to ...
15346 (predicate::operator &=): ... this one.
15347 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15348 (or_predicates): Turn to ...
15349 (predicate::or_with): ... this one.
15350 (evaluate_predicate): Turn to ...
15351 (predicate::evaluate): ... this one.
15352 (predicate_probability): Turn to ...
15353 (predicate::probability): ... this one.
15354 (dump_condition): Update.
15355 (dump_predicate): Turn to ...
15356 (predicate::dump): ... this one.
15357 (account_size_time): Update.
15358 (edge_set_predicate): Update.
15359 (set_hint_predicate): UPdate.
15360 (evaluate_conditions_for_known_args): Update.
15361 (evaluate_properties_for_edge): Update.
15362 (remap_predicate_after_duplication): Turn to...
15363 (predicate::remap_after_duplication): ... this one.
15364 (remap_hint_predicate_after_duplication): Update.
15365 (inline_summary_t::duplicate): UPdate.
15366 (dump_inline_edge_summary): Update.
15367 (dump_inline_summary): Update.
15368 (set_cond_stmt_execution_predicate): Update.
15369 (set_switch_stmt_execution_predicate): Update.
15370 (compute_bb_predicates): Update.
15371 (will_be_nonconstant_expr_predicate): Update.
15372 (will_be_nonconstant_predicate): Update.
15373 (phi_result_unknown_predicate): Update.
15374 (predicate_for_phi_result): Update.
15375 (array_index_predicate): Update.
15376 (estimate_function_body_sizes): Update.
15377 (estimate_node_size_and_time): Update.
15378 (estimate_ipcp_clone_size_and_time): Update.
15379 (remap_predicate): Rename to ...
15380 (predicate::remap_after_inlining): ... this one.
15381 (remap_hint_predicate): Update.
15382 (inline_merge_summary): Update.
15383 (inline_update_overall_summary): Update.
15384 (estimate_size_after_inlining): Update.
15385 (read_predicate): Rename to ...
15386 (predicate::stream_in): ... this one.
15387 (read_inline_edge_summary): Update.
15388 (write_predicate): Rename to ...
15389 (predicate::stream_out): ... this one.
15390 (write_inline_edge_summary): Update.
15391 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15392 (clause_t): Turn to uint32_t
15393 (predicate): Turn to class; implement constructor and operators
15395 (size_time_entry): Update.
15396 (inline_summary): Update.
15397 (inline_edge_summary): Update.
15399 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15401 * fold-const.c (fold_binary_loc): Move transformation...
15402 * match.pd (C - X CMP X): ... here.
15404 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15406 * config/sparc/sparc.c (sparc_option_override): Set function
15407 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15408 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15410 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15412 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15414 2017-05-18 Marek Polacek <polacek@redhat.com>
15417 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15418 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15419 gimple_assign_load_p.
15421 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15423 PR middle-end/80692
15424 * real.c (do_compare): Give decimal_do_compare preference over
15425 comparing just the signs.
15427 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15429 * doc/md.texi (Canonicalization of Instructions): Describe the
15430 canonical form of instructions that inherently set a condition
15433 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15435 PR middle-end/80775
15436 * tree-cfg.c: Move deletion of unreachable case statements to after
15437 the merging of consecutive case labels.
15439 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15441 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15442 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15443 restoring of callee-saved registers.
15445 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15447 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15448 * config/visium/visium.c (single_set_and_flags): Likewise.
15449 * config/visium/visium.md (Substitutions): Likewise.
15451 2017-05-17 Martin Liska <mliska@suse.cz>
15453 * cfg.c: Introduce dump_flags_t type and
15454 use it instead of int type.
15456 * cfghooks.c: Likewise.
15457 * cfghooks.h (struct cfg_hooks): Likewise.
15458 * cfgrtl.c: Likewise.
15459 * cfgrtl.h: Likewise.
15460 * cgraph.c (cgraph_node::get_body): Likewise.
15461 * coretypes.h: Likewise.
15462 * domwalk.c: Likewise.
15463 * domwalk.h: Likewise.
15464 * dumpfile.c (struct dump_option_value_info): Likewise.
15465 (dump_enable_all): Likewise.
15466 (dump_switch_p_1): Likewise.
15467 (opt_info_switch_p): Likewise.
15468 * dumpfile.h (enum tree_dump_index): Likewise.
15469 (struct dump_file_info): Likewise.
15470 * genemit.c: Likewise.
15471 * generic-match-head.c: Likewise.
15472 * gengtype.c (open_base_files): Likewise.
15473 * gimple-pretty-print.c: Likewise.
15474 * gimple-pretty-print.h: Likewise.
15475 * graph.c (print_graph_cfg): Likewise.
15476 * graphite-scop-detection.c (dot_all_sese): Likewise.
15477 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15478 * loop-unroll.c (report_unroll): Likewise.
15479 * passes.c (pass_manager::register_one_dump_file): Likewise.
15480 * print-tree.c: Likewise.
15481 * statistics.c: Likewise.
15482 * tree-cfg.c: Likewise.
15483 * tree-cfg.h: Likewise.
15484 * tree-dfa.c: Likewise.
15485 * tree-dfa.h: Likewise.
15486 * tree-dump.c (dump_function): Likewise.
15487 * tree-dump.h (struct dump_info): Likewise.
15488 * tree-pretty-print.c: Likewise.
15489 * tree-pretty-print.h: Likewise.
15490 * tree-ssa-live.c: Likewise.
15491 * tree-ssa-live.h: Likewise.
15492 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15493 * tree-vect-loop.c: Likewise.
15494 * tree-vect-slp.c: Likewise.
15496 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15497 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15499 PR tree-optimization/80457
15500 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15501 of all arguments to a statement as scalar_to_vec operations.
15502 (vectorizable_call): Adjust call to vect_model_simple_cost for
15504 (vectorizable_conversion): Likewise.
15505 (vectorizable_assignment): Likewise.
15506 (vectorizable_shift): Likewise.
15507 (vectorizable_operation): Likewise.
15508 (vectorizable_comparison): Likewise.
15509 (vect_is_simple_cond): Record the def types for operands.
15510 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15511 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15512 for statement argument count.
15514 2017-05-16 Carl Love <cel@us.ibm.com>
15516 * config/rs6000/rs6000-c: Add support for built-in functions
15517 vector unsigned long long vec_bperm (vector unsigned long long,
15518 vector unsigned char)
15519 vector signed long long vec_mule (vector signed int,
15521 vector unsigned long long vec_mule (vector unsigned int,
15522 vector unsigned int)
15523 vector signed long long vec_mulo (vector signed int,
15525 vector unsigned long long vec_mulo (vector unsigned int,
15526 vector unsigned int)
15527 vector signed char vec_sldw (vector signed char,
15528 vector signed char,
15530 vector unsigned char vec_sldw (vector unsigned char,
15531 vector unsigned char,
15533 vector signed short vec_sldw (vector signed short,
15534 vector signed short,
15536 vector unsigned short vec_sldw (vector unsigned short,
15537 vector unsigned short,
15539 vector signed int vec_sldw (vector signed int,
15542 vector unsigned int vec_sldw (vector unsigned int,
15543 vector unsigned int,
15545 vector signed long long vec_sldw (vector signed long long,
15546 vector signed long long,
15548 vector unsigned long long vec_sldw (vector unsigned long long,
15549 vector unsigned long long,
15551 * config/rs6000/rs6000-c: Add support for built-in functions
15552 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15553 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15554 * doc/extend.texi: Update the built-in documentation for the
15555 new built-in functions.
15557 2017-05-16 Marek Polacek <polacek@redhat.com>
15561 * tree.c (save_expr): Don't fold the expression.
15563 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15565 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15566 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15567 and (?*y,m). Update insn attributes.
15569 2017-05-16 Martin Liska <mliska@suse.cz>
15571 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15572 flags argument of print_gimple_stmt, print_gimple_expr,
15573 print_generic_stmt and print_generic_expr.
15574 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15575 * coretypes.h: Likewise.
15576 * except.c (dump_eh_tree): Likewise.
15577 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15578 * gimple-pretty-print.h: Likewise.
15579 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15580 (backprop::push_to_worklist): Likewise.
15581 (backprop::pop_from_worklist): Likewise.
15582 (backprop::process_use): Likewise.
15583 (backprop::intersect_uses): Likewise.
15584 (note_replacement): Likewise.
15585 * gimple-ssa-store-merging.c
15586 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15587 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15588 (pass_store_merging::execute): Likewise.
15589 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15590 (ssa_base_cand_dump_callback): Likewise.
15591 (dump_incr_vec): Likewise.
15592 (replace_refs): Likewise.
15593 (replace_mult_candidate): Likewise.
15594 (create_add_on_incoming_edge): Likewise.
15595 (create_phi_basis): Likewise.
15596 (insert_initializers): Likewise.
15597 (all_phi_incrs_profitable): Likewise.
15598 (introduce_cast_before_cand): Likewise.
15599 (replace_one_candidate): Likewise.
15600 * gimplify.c (gimplify_expr): Likewise.
15601 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15602 (set_rename): Likewise.
15603 (rename_uses): Likewise.
15604 (copy_loop_phi_nodes): Likewise.
15605 (add_close_phis_to_merge_points): Likewise.
15606 (copy_loop_close_phi_args): Likewise.
15607 (copy_cond_phi_args): Likewise.
15608 (graphite_copy_stmts_from_block): Likewise.
15609 (translate_pending_phi_nodes): Likewise.
15610 * graphite-poly.c (print_pdr): Likewise.
15611 (dump_gbb_cases): Likewise.
15612 (dump_gbb_conditions): Likewise.
15613 (print_scop_params): Likewise.
15614 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15615 (build_cross_bb_scalars_use): Likewise.
15616 (gather_bbs::before_dom_children): Likewise.
15617 * hsa-dump.c (dump_hsa_immed): Likewise.
15618 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15619 (get_replacement_map): Likewise.
15620 * ipa-inline-analysis.c (dump_condition): Likewise.
15621 (estimate_function_body_sizes): Likewise.
15622 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15623 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15624 * ipa-prop.c (ipa_dump_param): Likewise.
15625 (ipa_print_node_jump_functions_for_edge): Likewise.
15626 (ipa_modify_call_arguments): Likewise.
15627 (ipa_modify_expr): Likewise.
15628 (ipa_dump_param_adjustments): Likewise.
15629 (ipa_dump_agg_replacement_values): Likewise.
15630 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15631 * ipa-pure-const.c (check_stmt): Likewise.
15632 (pass_nothrow::execute): Likewise.
15633 * ipa-split.c (execute_split_functions): Likewise.
15634 * omp-offload.c (dump_oacc_loop_part): Likewise.
15635 (dump_oacc_loop): Likewise.
15636 * trans-mem.c (tm_log_emit): Likewise.
15637 (tm_memopt_accumulate_memops): Likewise.
15638 (dump_tm_memopt_set): Likewise.
15639 (dump_tm_memopt_transform): Likewise.
15640 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15641 (print_loop): Likewise.
15642 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15643 (chkp_gather_checks_info): Likewise.
15644 (chkp_get_check_result): Likewise.
15645 (chkp_remove_check_if_pass): Likewise.
15646 (chkp_use_outer_bounds_if_possible): Likewise.
15647 (chkp_reduce_bounds_lifetime): Likewise.
15648 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15649 (chkp_mark_completed_bounds): Likewise.
15650 (chkp_register_incomplete_bounds): Likewise.
15651 (chkp_mark_invalid_bounds): Likewise.
15652 (chkp_maybe_copy_and_register_bounds): Likewise.
15653 (chkp_build_returned_bound): Likewise.
15654 (chkp_get_bound_for_parm): Likewise.
15655 (chkp_build_bndldx): Likewise.
15656 (chkp_get_bounds_by_definition): Likewise.
15657 (chkp_generate_extern_var_bounds): Likewise.
15658 (chkp_get_bounds_for_decl_addr): Likewise.
15659 * tree-chrec.c (chrec_apply): Likewise.
15660 * tree-data-ref.c (dump_data_reference): Likewise.
15661 (dump_subscript): Likewise.
15662 (dump_data_dependence_relation): Likewise.
15663 (analyze_overlapping_iterations): Likewise.
15664 * tree-inline.c (expand_call_inline): Likewise.
15665 (tree_function_versioning): Likewise.
15666 * tree-into-ssa.c (dump_defs_stack): Likewise.
15667 (dump_currdefs): Likewise.
15668 (dump_names_replaced_by): Likewise.
15669 (dump_update_ssa): Likewise.
15670 (update_ssa): Likewise.
15671 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15672 * tree-parloops.c (build_new_reduction): Likewise.
15673 (try_create_reduction_list): Likewise.
15674 (ref_conflicts_with_region): Likewise.
15675 (oacc_entry_exit_ok_1): Likewise.
15676 (oacc_entry_exit_single_gang): Likewise.
15677 * tree-pretty-print.h: Likewise.
15678 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15679 (get_scalar_evolution): Likewise.
15680 (add_to_evolution): Likewise.
15681 (get_loop_exit_condition): Likewise.
15682 (analyze_evolution_in_loop): Likewise.
15683 (analyze_initial_condition): Likewise.
15684 (analyze_scalar_evolution): Likewise.
15685 (instantiate_scev): Likewise.
15686 (number_of_latch_executions): Likewise.
15687 (gather_chrec_stats): Likewise.
15688 (final_value_replacement_loop): Likewise.
15689 (scev_const_prop): Likewise.
15690 * tree-sra.c (dump_access): Likewise.
15691 (disqualify_candidate): Likewise.
15692 (create_access): Likewise.
15693 (reject): Likewise.
15694 (maybe_add_sra_candidate): Likewise.
15695 (create_access_replacement): Likewise.
15696 (analyze_access_subtree): Likewise.
15697 (analyze_all_variable_accesses): Likewise.
15698 (sra_modify_assign): Likewise.
15699 (initialize_constant_pool_replacements): Likewise.
15700 (find_param_candidates): Likewise.
15701 (decide_one_param_reduction): Likewise.
15702 (replace_removed_params_ssa_names): Likewise.
15703 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15704 * tree-ssa-copy.c (dump_copy_of): Likewise.
15705 (copy_prop_visit_cond_stmt): Likewise.
15706 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15707 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15708 (record_equivalences_from_stmt): Likewise.
15709 * tree-ssa-dse.c (compute_trims): Likewise.
15710 (delete_dead_call): Likewise.
15711 (delete_dead_assignment): Likewise.
15712 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15713 (forward_propagate_into_cond): Likewise.
15714 (pass_forwprop::execute): Likewise.
15715 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15716 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15718 (move_computations_worker): Likewise.
15719 (execute_sm): Likewise.
15720 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15721 (remove_exits_and_undefined_stmts): Likewise.
15722 (remove_redundant_iv_tests): Likewise.
15723 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15724 (adjust_iv_update_pos): Likewise.
15725 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15726 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15727 (value_replacement): Likewise.
15728 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15729 * tree-ssa-pre.c (print_pre_expr): Likewise.
15730 (get_representative_for): Likewise.
15731 (create_expression_by_pieces): Likewise.
15732 (insert_into_preds_of_block): Likewise.
15733 (eliminate_insert): Likewise.
15734 (eliminate_dom_walker::before_dom_children): Likewise.
15735 (eliminate): Likewise.
15736 (remove_dead_inserted_code): Likewise.
15737 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15738 * tree-ssa-reassoc.c (get_rank): Likewise.
15739 (eliminate_duplicate_pair): Likewise.
15740 (eliminate_plus_minus_pair): Likewise.
15741 (eliminate_not_pairs): Likewise.
15742 (undistribute_ops_list): Likewise.
15743 (eliminate_redundant_comparison): Likewise.
15744 (update_range_test): Likewise.
15745 (optimize_range_tests_var_bound): Likewise.
15746 (optimize_vec_cond_expr): Likewise.
15747 (rewrite_expr_tree): Likewise.
15748 (rewrite_expr_tree_parallel): Likewise.
15749 (linearize_expr): Likewise.
15750 (break_up_subtract): Likewise.
15751 (linearize_expr_tree): Likewise.
15752 (attempt_builtin_powi): Likewise.
15753 (attempt_builtin_copysign): Likewise.
15754 (transform_stmt_to_copy): Likewise.
15755 (transform_stmt_to_multiply): Likewise.
15756 (dump_ops_vector): Likewise.
15757 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15758 (print_scc): Likewise.
15759 (set_ssa_val_to): Likewise.
15760 (visit_reference_op_store): Likewise.
15761 (visit_use): Likewise.
15762 (sccvn_dom_walker::before_dom_children): Likewise.
15763 (run_scc_vn): Likewise.
15764 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15766 (expr_hash_elt::print): Likewise.
15767 (const_and_copies::pop_to_marker): Likewise.
15768 (const_and_copies::record_const_or_copy_raw): Likewise.
15769 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15770 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15771 (dump_predicates): Likewise.
15772 (find_uninit_use): Likewise.
15773 (warn_uninitialized_phi): Likewise.
15774 (pass_late_warn_uninitialized::execute): Likewise.
15775 * tree-ssa.c (verify_vssa): Likewise.
15776 (verify_ssa): Likewise.
15777 (maybe_optimize_var): Likewise.
15778 * tree-vrp.c (dump_value_range): Likewise.
15779 (dump_all_value_ranges): Likewise.
15780 (dump_asserts_for): Likewise.
15781 (register_edge_assert_for_2): Likewise.
15782 (vrp_visit_cond_stmt): Likewise.
15783 (vrp_visit_switch_stmt): Likewise.
15784 (vrp_visit_stmt): Likewise.
15785 (vrp_visit_phi_node): Likewise.
15786 (simplify_cond_using_ranges_1): Likewise.
15787 (fold_predicate_in): Likewise.
15788 (evrp_dom_walker::before_dom_children): Likewise.
15789 (evrp_dom_walker::push_value_range): Likewise.
15790 (evrp_dom_walker::pop_value_range): Likewise.
15791 (execute_early_vrp): Likewise.
15793 2017-05-16 Richard Biener <rguenther@suse.de>
15795 * dwarf2out.c (loc_list_from_tree_1): Do not create
15796 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15798 2017-05-16 Richard Biener <rguenther@suse.de>
15800 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15802 (note_variable_value_in_expr): If we resolved the decl ref
15803 do not push to the stack.
15805 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15807 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15808 operations in fast-math mode.
15809 (vaddq_f16): Likewise.
15810 (vmul_f16): Likewise.
15811 (vmulq_f16): Likewise.
15812 (vsub_f16): Likewise.
15813 (vsubq_f16): Likewise.
15814 * config/arm/neon.md (add<mode>3): New.
15816 (fma:<VH:mode>3): New. Also remove outdated comment.
15819 2017-05-16 Martin Liska <mliska@suse.cz>
15823 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15824 (odr_types_equivalent_p): Likewise.
15826 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15828 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15830 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15833 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15834 non-interunit SSE move alternatives with '?'.
15835 (zero-extendsidi peephole2): New peephole to skip intermediate
15836 general register in SSE zero-extend sequence.
15838 2017-05-15 Jeff Law <law@redhat.com>
15840 * reorg.c (relax_delay_slots): Create a new variable to hold
15841 the temporary target rather than clobbering TARGET_LABEL.
15843 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15844 missing argument to extract_bit_field call.
15845 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15847 2017-05-15 Martin Liska <mliska@suse.cz>
15850 * gcc.c (process_command): Do not allow empty argument of -o option.
15852 2017-05-15 Renlin Li <renlin.li@arm.com>
15854 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15855 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15856 * config/aarch64/constraints.md (Usf): Add long call check.
15857 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15858 (call_value): Likewise.
15859 (sibcall): Likewise.
15860 (sibcall_value): Likewise.
15862 (call_value_insn): New.
15863 (sibcall_insn): Update rtx pattern.
15864 (sibcall_value_insn): Likewise.
15865 (call_internal): Remove.
15866 (call_value_internal): Likewise.
15867 (sibcall_internal): Likewise.
15868 (sibcall_value_internal): Likewise.
15869 (call_reg): Likewise.
15870 (call_symbol): Likewise.
15871 (call_value_reg): Likewise.
15872 (call_value_symbol): Likewise.
15874 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15877 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15879 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15881 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15882 compatible with CCGOCmode and with CCZmode.
15884 2017-05-14 Martin Sebor <msebor@redhat.com>
15886 PR middle-end/77671
15887 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15888 (gimple_fold_builtin_snprintf): Same.
15889 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15890 (gimple_fold_builtin_snprintf): Same.
15891 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15892 of character types.
15893 (is_call_safe): New function.
15894 (try_substitute_return_value): Call it.
15895 (try_simplify_call): New function.
15896 (pass_sprintf_length::handle_gimple_call): Call it.
15898 2017-05-14 Martin Sebor <msebor@redhat.com>
15900 PR middle-end/80669
15901 * builtins.c (expand_builtin_stpncpy): Simplify.
15903 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15905 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15906 * config/i386/i386.h
15907 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15908 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15909 (struct machine_function): Add new members call_ms2sysv,
15910 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15911 (struct machine_frame_state): New fields sp_realigned and
15912 sp_realigned_offset.
15913 * config/i386/i386.c
15914 (enum xlogue_stub): New enum.
15915 (enum xlogue_stub_sets): New enum.
15916 (class xlogue_layout): New class.
15917 (struct ix86_frame): New fields stack_realign_allocate_offset,
15918 stack_realign_offset and outlined_save_offset. Modify comments to
15919 detail stack layout when using out-of-line stubs.
15920 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15921 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15922 -mcall-ms2sysv-xlogues.
15923 (stub_managed_regs): New static variable.
15924 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15925 registers managed by out-of-line stub.
15926 (disable_call_ms2sysv_xlogues): New function.
15927 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15928 m->call_ms2sysv when appropriate and compute frame layout for
15930 (sp_valid_at, fp_valid_at): New inline functions.
15931 (choose_basereg): New function.
15932 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15934 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15935 Use align parameter of choose_baseaddr to generated aligned SSE movs
15937 (pro_epilogue_adjust_stack): Modify to track
15938 machine_frame_state::sp_realigned.
15939 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15940 (ix86_nsaved_sseregs): Likewise.
15941 (ix86_emit_save_regs): Likewise.
15942 (ix86_emit_save_regs_using_mov): Likewise.
15943 (ix86_emit_save_sse_regs_using_mov): Likewise.
15944 (get_scratch_register_on_entry): Likewise.
15945 (gen_frame_set): New function.
15946 (gen_frame_load): Likewise.
15947 (gen_frame_store): Likewise.
15948 (emit_outlined_ms2sysv_save): Likewise.
15949 (emit_outlined_ms2sysv_restore): Likewise.
15950 (ix86_expand_prologue): Modify stack re-alignment code and call
15951 emit_outlined_ms2sysv_save when appropriate.
15952 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15953 parameter rtx_insn *insn, which allows the function to be used to only
15954 generate the notes.
15955 (ix86_expand_epilogue): Modify validity checks of frame and stack
15956 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15957 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15958 * config/i386/predicates.md
15959 (save_multiple): New predicate.
15960 (restore_multiple): Likewise.
15961 * config/i386/sse.md
15962 (save_multiple<mode>): New pattern.
15963 (save_multiple_realign<mode>): Likewise.
15964 (restore_multiple<mode>): Likewise.
15965 (restore_multiple_and_return<mode>): Likewise.
15966 (restore_multiple_leave_return<mode>): Likewise.
15967 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15969 2017-05-14 Julia Koval <julia.koval@intel.com>
15971 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15972 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15973 (__builtin_ia32_xsetbv): New builtins.
15974 * config/i386/i386.c (ix86_expand_special_args_builtin):
15976 (ix86_expand_builtin): Special expand for new intrinsics.
15977 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15978 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15979 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15981 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15983 * cfganal.c (inverted_post_order_compute): Change argument type
15985 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15986 * df-core.c (rest_of_handle_df_initialize): Adjust.
15987 (rest_of_handle_df_finish): Likewise.
15988 (df_analyze_1): Likewise.
15989 (df_analyze): Likewise.
15990 (loop_inverted_post_order_compute): Change argument to be a vec *.
15991 (df_analyze_loop): Adjust.
15992 (df_get_n_blocks): Likewise.
15993 (df_get_postorder): Likewise.
15994 * df.h (struct df_d): Change field to be a vec.
15995 * lcm.c (compute_laterin): Adjust.
15996 (compute_available): Likewise.
15997 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15998 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15999 * tree-ssa-pre.c (compute_antic): Likewise.
16001 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16003 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16004 (depth_first_search::depth_first_search): Change structure init
16005 function to this constructor.
16006 (depth_first_search::add_bb): Rename function to this member.
16007 (depth_first_search::execute): Likewise.
16008 (flow_dfs_compute_reverse_finish): Adjust.
16010 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16012 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16013 (longest_simple_path): Likewise.
16014 * shrink-wrap.c (spread_components): Likewise.
16015 (disqualify_problematic_components): Likewise.
16016 (emit_common_heads_for_components): Likewise.
16017 (emit_common_tails_for_components): Likewise.
16018 (insert_prologue_epilogue_for_components): Likewise.
16020 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16022 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16025 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16027 * df-core.c (df_set_blocks): Start using auto_bitmap.
16028 (df_compact_blocks): Likewise.
16029 * df-problems.c (df_rd_confluence_n): Likewise.
16030 * df-scan.c (df_insn_rescan_all): Likewise.
16031 (df_process_deferred_rescans): Likewise.
16032 (df_update_entry_block_defs): Likewise.
16033 (df_update_exit_block_uses): Likewise.
16034 (df_entry_block_bitmap_verify): Likewise.
16035 (df_exit_block_bitmap_verify): Likewise.
16036 (df_scan_verify): Likewise.
16037 * lra-constraints.c (lra_constraints): Likewise.
16038 (undo_optional_reloads): Likewise.
16039 (lra_undo_inheritance): Likewise.
16040 * lra-remat.c (calculate_gen_cands): Likewise.
16041 (do_remat): Likewise.
16042 * lra-spills.c (assign_spill_hard_regs): Likewise.
16043 (spill_pseudos): Likewise.
16044 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16045 (bitmap_set_subtract_values): Likewise.
16047 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16049 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16050 management with auto_bitmap.
16051 (fix_inter_tick): Likewise.
16052 (fix_recovery_deps): Likewise.
16053 * ira.c (add_store_equivs): Likewise.
16054 (find_moveable_pseudos): Likewise.
16055 (split_live_ranges_for_shrink_wrap): Likewise.
16056 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16057 (rtx_reuse_manager::seen_def_p): Likewise.
16058 (rtx_reuse_manager::set_seen_def): Likewise.
16059 * print-rtl.h (class rtx_reuse_manager): Likewise.
16061 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16063 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16065 (migrate_btr_def): Likewise.
16066 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16067 * df-core.c (loop_post_order_compute): Likewise.
16068 (loop_inverted_post_order_compute): Likewise.
16069 * hsa-common.h: Likewise.
16070 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16071 * init-regs.c (initialize_uninitialized_regs): Likewise.
16072 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16073 (inline_small_functions): Likewise.
16074 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16075 * ira.c (combine_and_move_insns): Likewise.
16076 (build_insn_chain): Likewise.
16077 * loop-invariant.c (find_invariants): Likewise.
16078 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16079 * predict.c (tree_predict_by_opcode): Likewise.
16080 (predict_paths_leading_to): Likewise.
16081 (predict_paths_leading_to_edge): Likewise.
16082 (estimate_loops_at_level): Likewise.
16083 (estimate_loops): Likewise.
16084 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16085 (spread_components): Likewise.
16086 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16087 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16088 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16089 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16090 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16091 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16092 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16093 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16094 (mark_threaded_blocks): Likewise.
16095 (thread_through_all_blocks): Likewise.
16096 * tree-ssa.c (verify_ssa): Likewise.
16097 (execute_update_addresses_taken): Likewise.
16098 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16100 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16102 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16104 (post_order_compute): Likewise.
16105 (inverted_post_order_compute): Likewise.
16106 (pre_and_rev_post_order_compute_fn): Likewise.
16108 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16110 * genrecog.c (int_set::int_set): Explicitly construct our
16111 auto_vec base class.
16112 * vec.h (auto_vec::auto_vec): New constructor.
16114 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16116 * bitmap.h (class auto_bitmap): New constructor taking
16117 bitmap_obstack * argument.
16119 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16121 * bitmap.h (class auto_bitmap): Change type of m_bits to
16122 bitmap_head, and adjust ctor / dtor and member operators.
16124 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16126 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16127 when returned register mode doesn't match original mode.
16129 2017-05-12 Jeff Law <law@redhat.com>
16130 Jakub Jelinek <jakub@redhat.com>
16132 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16133 we look for cc setter after the compare-elim changes.
16134 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16135 within the vector to match what compare-elim now expects.
16136 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16137 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16139 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16140 after the compare-elim changes.
16141 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16142 the vector to match what compare-elim now expects.
16143 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16144 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16145 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16146 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16147 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16149 * config/visium/visium.c (single_set_and_flags): Fix where
16150 we look for cc setter after the compare-elim changes.
16151 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16152 with the vector to match what compare-elim now expects.
16153 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16154 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16155 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16156 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16157 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16158 (neg<mode>2_insn_set_overflow): Likewise.
16160 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16162 PR middle-end/79794
16163 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16164 maybe_expand_insn call, set ops[0].target. If still set after call,
16165 set alt_rtl. Add extra arg to recursive calls.
16166 (extract_bit_field): Add alt_rtl argument. Pass to
16168 * expmed.h (extract_bit_field): Fix prototype.
16169 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16170 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16171 to extract_bit_field_calls.
16172 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16173 Pass alt_rtl to extract_bit_field calls.
16174 * calls.c (store_unaligned_arguments_into_psuedos)
16175 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16176 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16178 * optabs.h (struct expand_operand): Add target bitfield.
16180 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16182 * compare-elim.c (try_eliminate_compare): Canonicalize
16183 operation with embedded compare to
16184 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16185 (set (reg) (operation)].
16187 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16189 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16192 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16193 cost of adding a carry flag for ADC instruction.
16194 [case MINUS]: Ignore the cost of subtracting a carry flag
16195 for SBB instruction.
16197 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16199 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16201 * config/rs6000/bmiintrin.h: New file.
16202 * config/rs6000/bmi2intrin.h: New file.
16203 * config/rs6000/x86intrin.h: New file.
16205 2017-05-12 Jeff Law <law@redhat.com>
16207 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16210 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16212 PR middle-end/80707
16213 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16215 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16217 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16218 early expansion of vector divide builtins.
16219 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16220 builtins identified as having unsigned arguments.
16222 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16224 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16225 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16226 expansion of vector logical operations (and, andc, or, xor,
16229 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16231 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16232 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16234 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16236 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16237 early GIMPLE expansion of vector multiplies.
16239 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16241 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16242 TARGET_HAVE_MOVT conditional.
16243 (movt splitter): Likewise.
16245 2017-05-12 Richard Biener <rguenther@suse.de>
16247 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16248 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16249 Fold all stmts not inplace.
16251 2017-05-12 Richard Biener <rguenther@suse.de>
16253 PR tree-optimization/80713
16254 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16255 inserted_exprs bit for not removed stmts.
16257 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16259 PR middle-end/69921
16260 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16261 parallelized" attribute for parallelized OpenACC kernels.
16262 * omp-offload.c (execute_oacc_device_lower): Use it.
16264 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16265 Set "oacc kernels" attribute.
16266 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16267 parameter. Adjust all users.
16268 (oacc_fn_attrib_kernels_p): Remove function.
16269 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16270 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16271 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16272 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16273 assert "oacc kernels" attribute is set.
16275 2017-05-11 Carl Love <cel@us.ibm.com>
16277 * config/rs6000/rs6000-c: Add support for built-in functions
16278 vector unsigned char vec_popcnt (vector signed char)
16279 vector unsigned char vec_popcnt (vector unsigned char)
16280 vector unsigned short vec_popcnt (vector signed short)
16281 vector unsigned short vec_popcnt (vector unsigned short)
16282 vector unsigned int vec_popcnt (vector signed int)
16283 vector unsigned int vec_popcnt (vector unsigned int)
16284 vector unsigned long long vec_popcnt (vector signed long long)
16285 vector unsigned long long vec_popcnt (vector unsigned long long)
16286 vector signed long long vec_slo (vector signed long long,
16287 vector signed char)
16288 vector signed long long vec_slo (vector signed long long,
16289 vector unsigned char)
16290 vector unsigned long long vec_slo (vector unsigned long long,
16291 vector signed char)
16292 vector unsigned long long vec_slo (vector unsigned long long,
16293 vector unsigned char)
16294 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16295 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16296 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16297 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16298 * doc/extend.texi: Update the built-in documentation file for the
16299 new built-in functions.
16301 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16303 * attribs.h (sorted_attr_string): Move machine independent
16304 functions for target clone support from the i386 port to common
16305 code. Rename ix86_function_versions to common_function_versions.
16306 Rename make_name to make_unique_name.
16307 (common_function_versions): Likewise.
16308 (make_unique_name): Likewise.
16309 (make_dispatcher_decl): Likewise.
16310 (is_function_default_version): Likewise.
16311 * attribs.c (attr_strcmp): Likewise.
16312 (sorted_attr_string): Likewise.
16313 (common_function_versions): Likewise.
16314 (make_unique_name): Likewise.
16315 (make_dispatcher_decl): Likewise.
16316 (is_function_default_version): Likewise.
16317 * config/i386/i386.c (attr_strcmp): Likewise.
16318 (sorted_attr_string): Likewise.
16319 (ix86_function_versions): Likewise.
16320 (make_name): Likewise.
16321 (make_dispatcher_decl): Likewise.
16322 (is_function_default_version): Likewise.
16323 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16325 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16328 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16329 Account for direct move costs for vec_construct of integer
16332 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16335 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16336 (UNSPEC_STX_ATOMIC): Ditto.
16337 (loaddi_via_sse): New insn.
16338 (storedi_via_sse): Ditto.
16339 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16340 Update corresponding peephole2 patterns.
16341 (atomic_storedi_fpu): Ditto.
16343 2017-05-11 Julia Koval <julia.koval@intel.com>
16345 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16346 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16348 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16349 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16350 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16352 2017-05-11 Nathan Sidwell <nathan@acm.org>
16354 * graphite-poly.c: Include dumpfile.h.
16356 * dumpfle.h (dump_function): Declare here ...
16357 * tree-dump.h (dump_function): ... not here.
16358 * dumpfile.c: #include tree-cfg.h.
16359 (dump_function): Move here from ...
16360 * tree-dump.c (dump_function): ... here.
16361 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16362 * graphite-poly.c: Don't include tree-dump.h.
16363 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16364 * print-tree.c: Likewise.
16365 * stor-layout.c: Likewise.
16366 * tree-nested.c: Likewise.
16368 * dumpfile.c (dump_start): Use TDF_FLAGS.
16369 (dump_enable_all): Fix TDF_KIND check thinko.
16371 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16373 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16374 array entries to represent two legal parameterizations of the
16375 overloaded __builtin_cmpb function, as represented by the
16376 P6_OV_BUILTIN_CMPB constant.
16377 (altivec_resolve_overloaded_builtin): Add special case handling
16378 for the __builtin_cmpb function, as represented by the
16379 P6_OV_BUILTIN_CMPB constant.
16380 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16381 (BU_P6_64BIT_2): New macro.
16382 (BU_P6_OVERLOAD_2): New macro
16383 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16384 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16385 (CMPB): Add overload support to represent both 32-bit and 64-bit
16386 compare-bytes function.
16387 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16388 support for TARGET_CMPB.
16389 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16390 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16391 documentation of the __builtin_cmpb overloaded built-in function.
16393 2017-05-11 Richard Biener <rguenther@suse.de>
16395 PR tree-optimization/80705
16396 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16397 bases are not vectorizable.
16399 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16401 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16402 when counting register pressure.
16404 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16406 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16407 (iv_ca_more_deps): Renamed to ...
16408 (iv_ca_compare_deps): ... this.
16409 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16413 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16415 (determine_group_iv_costs): ... here.
16416 (find_inv_vars_cb): Record inv var if it's not recorded before.
16418 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16420 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16421 (get_shiftadd_cost): Ditto.
16423 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16425 * tree-ssa-address.c: Include header file.
16426 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16428 (add_to_parts): Refactor.
16429 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16430 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16433 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16435 PR tree-optimization/53090
16436 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16438 (extract_cond_operands): Detect condition with IV on both sides
16439 and return COMP_IV_EXPR_2.
16440 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16441 (rewrite_use_compare): Simplify by removing call to function
16442 extract_cond_operands.
16444 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16446 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16447 (extract_cond_operands): Detect condition comparing against non-
16448 invariant bound and return appropriate enum value.
16449 (find_interesting_uses_cond): Update use of extract_cond_operands.
16450 Handle its return value accordingly.
16451 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16455 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16456 nonlinear iv_use computation in loop invariant sensitive way.
16458 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16460 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16461 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16463 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16465 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16466 (dump_cand): Support iv_cand.inv_exprs.
16467 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16469 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16474 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16476 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16477 as local function. Include necessary header files.
16478 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16482 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16484 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16486 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16487 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16488 RSHIFT_EXPR and BIT_NOT_EXPR.
16490 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16492 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16493 (adjust_setup_cost): New parameter supporting round up adjustment.
16494 (struct address_cost_data): Delete.
16495 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16496 (split_address_cost, ptr_difference_cost): Delete.
16497 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16498 (struct ainc_cost_data): New struct.
16499 (get_address_cost_ainc): New function.
16500 (get_address_cost, get_computation_cost): Reimplement.
16501 (determine_group_iv_cost_address): Record inv_expr for all uses of
16503 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16504 (iv_ca_has_deps): Reimplemented to ...
16505 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16507 (iv_ca_extend): Call iv_ca_more_deps.
16509 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16511 * tree-ssa-address.c (struct mem_address): Move to header file.
16512 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16513 * tree-ssa-address.h (struct mem_address): Move from C file.
16514 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16516 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16518 * tree-affine.h (aff_combination_type): New interface.
16519 (aff_combination_zero_p): Remove static.
16520 (aff_combination_const_p): New interface.
16521 (aff_combination_singleton_var_p): New interfaces.
16523 2017-05-11 Richard Biener <rguenther@suse.de>
16525 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16526 Skip unreachable blocks and destinations.
16527 (eliminate): Move stmt removal and fixup ...
16528 (fini_eliminate): ... here. Skip inserted exprs.
16529 (pass_pre::execute): Move fini_pre after fini_eliminate.
16530 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16531 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16532 PRE to get rid of dead code that has invalid SSA form and
16533 split critical edges again.
16535 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16537 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16539 2017-05-11 Richard Biener <rguenther@suse.de>
16541 * passes.c (execute_function_todo): Verify loops if they are
16542 said to be up-to-date.
16543 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16544 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16546 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16549 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16550 handle calling assemble_external ourself.
16553 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16554 modes with zero size. Enhance comment.
16556 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16558 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16559 built-ins for vec_xl and vec_xst with short and char pointer
16562 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16564 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16565 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16566 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16567 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16568 (_mm_maskz_min_round_ss): New intrinsics.
16569 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16570 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16571 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16572 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16573 (__builtin_ia32_minss_mask_round): New builtins.
16574 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16575 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16576 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16578 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16579 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16581 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16584 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16586 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16587 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16588 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16589 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16590 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16591 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16592 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16593 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16594 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16595 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16596 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16597 (__builtin_ia32_mulss_mask_round): New builtins.
16598 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16599 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16600 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16602 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16603 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16605 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16608 2017-05-10 Julia Koval <julia.koval@intel.com>
16610 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16611 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16612 (_mm256_setr_m128i): New intrinsics.
16614 2017-05-10 Julia Koval <julia.koval@intel.com>
16616 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16617 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16618 (_mm_maskz_rcp14_ss): New intrinsics.
16619 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16620 (__builtin_ia32_rcp14ss_mask): New builtins.
16621 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16623 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16625 PR tree-optimization/51513
16626 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16627 (assert_unreachable_fallthru_edge_p): Use it.
16628 (group_case_labels_stmt): Likewise.
16629 * tree-cfg.h: Prototype it.
16630 * stmt.c: Include cfghooks.h and tree-cfg.h.
16631 (emit_case_dispatch_table) <gap_label>: New local variable.
16632 Use it to fill dispatch table gaps.
16633 Test for default_label before updating probabilities.
16634 (expand_case) <default_label>: Remove unneeded initialization.
16635 Test for unreachable default case statement and remove its edge.
16636 Set default_label accordingly.
16637 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16639 2017-05-10 Carl Love <cel@us.ibm.com>
16641 * config/rs6000/rs6000-c: Add support for built-in functions
16642 vector signed char vec_neg (vector signed char)
16643 vector signed short int vec_neg (vector short int)
16644 vector signed int vec_neg (vector signed int)
16645 vector signed long long vec_neg (vector signed long long)
16646 vector float vec_neg (vector float)
16647 vector double vec_neg (vector double)
16648 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16650 * config/rs6000/altivec.h: Add define for vec_neg
16651 * doc/extend.texi: Update the built-in documentation for the
16652 new built-in functions.
16654 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16656 PR tree-optimization/77644
16657 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16659 2017-05-10 Nathan Sidwell <nathan@acm.org>
16661 * dumpfile.h (TDI_lang_all): New.
16662 (TDF_KIND): New. Renumber others
16663 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16665 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16667 (get_dump_file_name): Adjust suffix generation.
16668 (dump_enable_all): Use TDF_KIND.
16669 * doc/invoke.texi (-fdump-lang-all): Document.
16671 * dumpfile.h: Tabify.
16673 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16676 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16677 Move member access before delete.
16679 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16681 * tree-inline.c (expand_call_inline): Split block at stmt
16684 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16687 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16688 are now unused after splitting mov{sf,sd}_hardfloat.
16689 (f32_lr2): Likewise.
16690 (f32_lm): Likewise.
16691 (f32_lm2): Likewise.
16692 (f32_li): Likewise.
16693 (f32_li2): Likewise.
16694 (f32_lv): Likewise.
16695 (f32_sr): Likewise.
16696 (f32_sr2): Likewise.
16697 (f32_sm): Likewise.
16698 (f32_sm2): Likewise.
16699 (f32_si): Likewise.
16700 (f32_si2): Likewise.
16701 (f32_sv): Likewise.
16702 (f32_dm): Likewise.
16703 (f32_vsx): Likewise.
16704 (f32_av): Likewise.
16705 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16706 For movsf, order stores so the VSX stores occur before the GPR
16707 store which encourages the register allocator to use a traditional
16708 FPR instead of a GPR. For movsd, order the stores so that the GPR
16709 store comes before the VSX stores to allow the power6 to work.
16710 This is due to the power6 not having a 32-bit integer store
16711 instruction from a FPR.
16712 (movsf_hardfloat): Likewise.
16713 (movsd_hardfloat): Likewise.
16715 2017-05-09 Martin Sebor <msebor@redhat.com>
16717 PR translation/80280
16718 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16721 PR translation/80280
16722 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16723 data member added in r247778.
16724 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16726 2017-05-09 Nathan Sidwell <nathan@acm.org>
16728 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16730 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16733 2017-05-09 Marek Polacek <polacek@redhat.com>
16735 * doc/invoke.texi: Fix typo.
16737 2017-05-09 Richard Biener <rguenther@suse.de>
16739 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16740 (vrp_val_is_min): Likewise.
16741 (set_value_range_to_value): Likewise.
16742 (set_value_range_to_nonnegative): Likewise.
16743 (gimple_assign_nonzero_p): Likewise.
16744 (gimple_stmt_nonzero_p): Likewise.
16745 (vrp_int_const_binop): Likewise. Remove unreachable case.
16746 (adjust_range_with_scev): Adjust comments.
16747 (compare_range_with_value): Likewise.
16748 (extract_range_from_phi_node): Likewise.
16749 (test_for_singularity): Likewise.
16751 2017-05-09 Richard Biener <rguenther@suse.de>
16753 * tree-vrp.c (get_single_symbol): Add assert that we don't
16754 get overflowed constants as invariant part.
16755 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16756 checks. Use wi::cmp instead of recursing for integer constants.
16757 (compare_values): Just ignore whether we assumed undefined
16758 overflow instead of failing the compare.
16759 (extract_range_for_var_from_comparison_expr): Add comment before the
16760 TREE_NO_WARNING sets.
16761 (test_for_singularity): Likewise.
16762 (extract_range_from_comparison): Do not disable optimization
16763 when we assumed undefined overflow.
16764 (extract_range_basic): Remove init of unused var.
16766 2017-05-09 Richard Biener <rguenther@suse.de>
16768 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16769 (extract_range_from_multiplicative_op_1): Adjust.
16770 (extract_range_from_binary_expr_1): Use int_const_binop.
16772 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16775 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16776 rs6000_store_data_bypass_p in seven define_bypass directives and
16777 in several comments.
16778 * config/rs6000/rs6000-protos.h: Add prototype for
16779 rs6000_store_data_bypass_p function.
16780 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16781 function implements slightly different (rs6000-specific) semantics
16782 than store_data_bypass_p, returning false rather than aborting
16783 with assertion error when arguments do not satisfy the
16784 requirements of store data bypass.
16785 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16786 rs6000_store_data_bypass_p.
16788 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16790 * config/xtensa/xtensa-protos.h
16791 (xtensa_initial_elimination_offset): New declaration.
16792 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16793 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16794 macro definition, add case for FRAME_POINTER_REGNUM when
16795 FRAME_GROWS_DOWNWARD.
16796 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16797 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16798 xtensa_initial_elimination_offset.
16800 2017-05-08 Nathan Sidwell <nathan@acm.org>
16802 * doc/invoke.texi: Alphabetize -fdump options.
16804 2017-05-08 Martin Sebor <msebor@redhat.com>
16806 PR translation/80280
16807 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16809 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16811 * target.def (compute_frame_layout): New optional target hook.
16812 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16813 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16814 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16816 * reload1.c (verify_initial_elim_offsets): Likewise.
16817 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16818 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16819 (arm_get_frame_offsets): Split up into this ...
16820 (arm_compute_frame_layout): ... and this function.
16822 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16824 * config/aarch64/constraints.md (Usa): New constraint.
16825 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16827 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16829 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16830 with_multilib_list after it has been checked.
16832 2017-05-08 Richard Biener <rguenther@suse.de>
16834 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16835 (bitmap_set_subtract_values): Likewise.
16837 2017-05-08 Richard Biener <rguenther@suse.de>
16839 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16840 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16842 (gimple_stmt_nonzero_warnv_p): Rename to ...
16843 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16845 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16846 (extract_range_basic): Adjust, do not disable propagation on
16847 strict overflow sensitive simplification.
16848 (vrp_visit_cond_stmt): Likewise.
16850 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16852 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16853 body size unconditionally.
16855 2017-05-07 Jeff Law <law@redhat.com>
16858 2017-05-06 Jeff Law <law@redhat.com>
16859 PR tree-optimization/78496
16860 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16863 PR tree-optimization/78496
16864 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16865 (simplify_stmt_using_ranges): Call it.
16866 (vrp_dom_walker::before_dom_children): Extract equivalences
16867 from an ASSERT_EXPR with an equality comparison against a
16870 2017-05-06 Jeff Law <law@redhat.com>
16872 PR tree-optimization/78496
16873 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16876 PR tree-optimization/78496
16877 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16878 (simplify_stmt_using_ranges): Call it.
16879 (vrp_dom_walker::before_dom_children): Extract equivalences
16880 from an ASSERT_EXPR with an equality comparison against a
16883 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16885 * lra-constraints.c (lra_copy_reg_equiv): New function.
16886 (split_reg): Use it to copy equivalence information from the
16887 original register to the spill register.
16889 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16891 PR rtl-optimization/75964
16892 * simplify-rtx.c (simplify_const_relational_operation): Remove
16893 invalid handling of comparisons of integer ABS.
16895 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16897 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16898 initialize to zero.
16899 (init_regs): Remove declaration.
16900 (function_arg_advance_32): Initialize error_p as boolean variable.
16902 2017-05-05 Nathan Sidwell <nathan@acm.org>
16904 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16905 lines. Use for (;;).
16907 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16909 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16910 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16911 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16912 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16913 VF=2 that require versioning.
16915 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16917 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16920 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16922 * diagnostic.h (diagnostic_override_option_index): Convert from
16923 macro to inline function.
16925 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16927 * diagnostic.c (last_module_changed_p): New function.
16928 (set_last_module): New function.
16929 (diagnostic_report_current_module): Convert macro usage to
16930 the above functions.
16931 * diagnostic.h (diagnostic_context::last_module): Strengthen
16932 from const line_map * to const line_map_ordinary *.
16933 (diagnostic_last_module_changed): Delete macro.
16934 (diagnostic_set_last_module): Delete macro.
16936 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16938 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16939 with diagnostic_report_diagnostic.
16940 (diagnostic_n_impl_richloc): Likewise.
16941 * diagnostic.h (report_diagnostic): Delete macro.
16942 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16943 with diagnostic_report_diagnostic.
16944 * substring-locations.c (format_warning_va): Likewise.
16946 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16948 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16949 save/restor of format_spec. Move option-printing code to...
16950 (print_option_information): ...this new function, and
16951 reimplement by simply printing to the pretty_printer,
16952 rather than appending to the format string.
16954 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16956 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16957 handling logic into...
16958 (update_effective_level_from_pragmas): ...this new function.
16960 2017-05-04 Andrew Waterman <andrew@sifive.com>
16962 * config/riscv/riscv.opt (mstrict-align): New option.
16963 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16964 (SLOW_UNALIGNED_ACCESS): Define.
16965 (riscv_slow_unaligned_access): Declare.
16966 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16968 (riscv_slow_unaligned_access): New variable.
16969 (rocket_tune_info): Set slow_unaligned_access to true.
16970 (optimize_size_tune_info): Set slow_unaligned_access to false.
16971 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16972 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16973 (riscv_option_override): Set riscv_slow_unaligned_access.
16974 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16976 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16978 * config/riscv/riscv.md: Unify indentation.
16980 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16985 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16987 (extendsi<mode>2): Add support for doing sign extension via
16988 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16989 don't have ISA 3.0 instructions.
16990 (extendsi<mode>2 splitter): Likewise.
16991 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16992 generate the normal insns since SImode can now go in vector
16993 registers. Disallow the special UNSPECs needed for previous
16994 machines to hide SImode being used. Add new insns
16995 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16996 (fix_trunc<mode>si2_stfiwx): Likewise.
16997 (fix_trunc<mode>si2_internal): Likewise.
16998 (fixuns_trunc<mode>si2): Likewise.
16999 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17000 (fctiw<u>z_<mode>_smallint): Likewise.
17001 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17002 of floating point to 32-bit integer from doing a direct move to
17003 the GPR registers to do a store.
17004 (fctiwz_<mode>): Break long line.
17006 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17008 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17009 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17010 (addr_list, addr_offset_valid_p): New.
17011 (split_address_groups): Check offset validity with above function.
17012 (gt-tree-ssa-loop-ivopts.h): Include header file.
17014 2017-05-05 Nathan Sidwell <nathan@acm.org>
17016 * config.gcc (arm*-*-*): Add missing 'fi'.
17018 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17020 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17021 included in -fopt-info does not matter.
17022 * doc/optinfo.texi (-fopt-info): Fix description of default
17023 behavour. Explicitly say order of options included in -fopt-info
17026 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17028 * config.gcc: Allow combinations of aprofile and rmprofile values for
17029 --with-multilib-list.
17030 * config/arm/t-multilib: New file.
17031 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17032 variables. Remove setting of ISA and floating-point ABI in
17033 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17034 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17035 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17036 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17038 * config/arm/t-rmprofile: Likewise except for the matches changes.
17039 * doc/install.texi (--with-multilib-list): Document the combination of
17040 aprofile and rmprofile values and warn about pitfalls in doing that.
17042 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17044 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17045 (movdi_aarch64): Likewise.
17047 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17049 PR tree-optimization/80632
17050 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17052 (build_arrays): Initialize it for virtual phis.
17053 (fix_phi_nodes): Use it for virtual phis.
17055 PR tree-optimization/80558
17056 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17057 [x, y] op z into [x op, y op z] for op & or | if conditions
17060 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17061 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17064 * config/arm/arm.md (use_literal_pool): Remove.
17065 (64-bit immediate split): No longer takes cost into consideration
17066 if arm_disable_literal_pool is enabled.
17067 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17068 used when arm_disable_literal_pool is enabled.
17069 (arm_max_const_double_inline_cost): Remove use of
17070 arm_disable_literal_pool.
17071 (push_minipool_fix): Add assert.
17072 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17073 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17074 (no_literal_pool_sf_immediate): New.
17076 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17078 PR tree-optimization/80613
17079 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17080 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17082 2017-05-05 Richard Biener <rguenther@suse.de>
17084 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17086 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17088 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17089 of this flag from insn conditions due to removal from r247495.
17091 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17093 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17095 (arm_early_store_addr_dep_ptr): Likewise.
17096 * config/arm/aarch-common-protos.h
17097 (arm_early_load_addr_dep_ptr): Add prototype.
17098 (arm_early_store_addr_dep_ptr): Likewise.
17099 * config/arm/cortex-a53.md: Add new bypasses.
17101 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17103 * tree.c (next_type_uid): Change type to unsigned.
17104 (type_hash_canon): Decrement back next_type_uid if
17105 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17106 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17109 2017-05-04 Martin Sebor <msebor@redhat.com>
17111 * builtins.c: Fix a trivial typo in a comment.
17113 PR middle-end/79234
17114 * builtins.c (check_sizes): Adjust to handle reading past the end.
17115 Avoid printing excessive upper bound of ranges. Use %E to print
17116 tree nodes instead of converting them to %wu.
17117 (expand_builtin_memchr): New function.
17118 (compute_dest_size): Rename...
17119 (compute_objsize): ...to this.
17120 (expand_builtin_memcpy): Adjust.
17121 (expand_builtin_mempcpy): Adjust.
17122 (expand_builtin_strcat): Adjust.
17123 (expand_builtin_strcpy): Adjust.
17124 (check_strncat_sizes): Adjust.
17125 (expand_builtin_strncat): Adjust.
17126 (expand_builtin_strncpy): Adjust and simplify.
17127 (expand_builtin_memset): Adjust.
17128 (expand_builtin_bzero): Adjust.
17129 (expand_builtin_memcmp): Adjust.
17130 (expand_builtin): Handle memcmp.
17131 (maybe_emit_chk_warning): Check strncat just once.
17133 2017-05-04 Martin Sebor <msebor@redhat.com>
17135 PR preprocessor/79214
17136 PR middle-end/79222
17137 PR middle-end/79223
17138 * builtins.c (check_sizes): Add inlining context and issue
17139 warnings even when -Wno-system-headers is set.
17140 (check_strncat_sizes): Same.
17141 (expand_builtin_strncat): Same.
17142 (expand_builtin_memmove): New function.
17143 (expand_builtin_stpncpy): Same.
17144 (expand_builtin): Handle memmove and stpncpy.
17146 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17148 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17149 which is not used any more.
17151 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17153 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17155 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17157 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17158 (cortexa53_tunings): Likewise.
17159 (cortexa57_tunings): Likewise.
17160 (cortexa72_tunings): Likewise.
17161 (cortexa73_tunings): Likewise.
17163 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17165 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17166 Set loop alignment to 8.
17168 2017-05-04 Martin Sebor <msebor@redhat.com>
17170 PR translation/80280
17171 * builtins.c (expand_builtin_object_size): Add missing quoting to
17172 %D and like directives.
17173 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17174 (hsa_type_for_tree_type): Same.
17175 (verify_function_arguments): Same.
17176 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17177 * varasm.c (get_section): Same.
17180 2017-05-04 Martin Sebor <msebor@redhat.com>
17182 PR translation/80280
17183 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17185 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17187 * config/aarch64/aarch64.c (generic_addrcost_table):
17188 Change HI/TI mode setting.
17190 2017-05-04 Martin Jambor <mjambor@suse.cz>
17192 PR tree-optimization/80622
17193 * tree-sra.c (comes_initialized_p): New function.
17194 (build_accesses_from_assign): Only set write lazily when
17195 comes_initialized_p is false.
17196 (analyze_access_subtree): Use comes_initialized_p.
17197 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17198 instead of testing for PARM_DECL.
17200 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17202 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17203 constraint on operand 0 to allow more general addressing modes.
17204 Adjust output template.
17205 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17207 * config/aarch64/aarch64-protos.h
17208 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17209 * config/aarch64/constraints.md (Dp): New address constraint.
17210 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17213 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17215 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17216 update use of estimate_ipcp_clone_size_and_time.
17217 (estimate_local_effects): Update use of
17218 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17219 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17220 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17221 Return nonspecialized time.
17223 2017-05-04 Richard Biener <rguenther@suse.de>
17225 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17226 for the last VUSE which def dominates the PHI. Directly call
17228 (get_continuation_for_phi_1): Remove.
17230 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17232 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17233 to explain the use of truncating division. Cap the number of
17234 iterations to the maximum given by nb_iterations_upper_bound,
17237 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17239 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17240 * configure: Regenerate.
17241 * config.in: Regenerate.
17242 * config/i386/driver-mingw32.c: new file.
17243 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17244 * config.host: Link driver-mingw32.o on MinGW host.
17245 * doc/install.texi: Document new --enable-mingw-wildcard configure
17248 2017-05-04 Marek Polacek <polacek@redhat.com>
17250 PR tree-optimization/80612
17251 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17253 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17254 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17256 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17257 (movt splitter): Likewise.
17258 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17259 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17260 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17261 block for Thumb-1 with MOVT.
17262 (thumb2_legitimate_address_p): Move code block ...
17263 (can_avoid_literal_pool_for_label_p): ... into this new function.
17264 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17266 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17267 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17268 "M-profile targets with the MOVT instruction".
17270 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17272 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17273 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17274 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17276 2017-05-04 Martin Liska <mliska@suse.cz>
17278 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17279 variable cond_code.
17281 2017-05-04 Richard Biener <rguenther@suse.de>
17283 * tree.c (array_at_struct_end_p): Handle arrays at struct
17284 end with flexarrays more conservatively. Refactor and treat
17285 arrays of arrays or aggregates more strict. Fix
17286 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17287 * tree.c (array_at_struct_end_p): Adjust prototype.
17288 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17289 * gimple-fold.c (get_range_strlen): Likewise.
17290 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17292 2017-05-04 Richard Biener <rguenther@suse.de>
17294 PR tree-optimization/31130
17295 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17297 (supports_overflow_infinity): Likewise.
17298 (is_negative_overflow_infinity): Likewise.
17299 (is_positive_overflow_infinity): Likewise.
17300 (is_overflow_infinity): Likewise.
17301 (stmt_overflow_infinity): Likewise.
17302 (overflow_infinity_range_p): Likewise.
17303 (usable_range_p): Remove as always returning true.
17304 (make_overflow_infinity): Remove.
17305 (negative_overflow_infinity): Likewise.
17306 (positive_overflow_infinity): Likewise.
17307 (avoid_overflow_infinity): Likewise.
17308 (set_value_range): Adjust accordingly.
17309 (set_value_range_to_nonnegative): Likewise, remove now unused
17310 overflow_infinity arg.
17311 (vrp_operand_equal_p): Adjust.
17312 (update_value_range): Likewise.
17313 (range_int_cst_singleton_p): Likewise.
17314 (operand_less_p): Likewise.
17315 (compare_values_warnv): Likewise.
17316 (extract_range_for_var_from_comparison_expr): Likewise.
17317 (vrp_int_const_binop): Likewise.
17318 (zero_nonzero_bits_from_vr): Likewise.
17319 (extract_range_from_multiplicative_op_1): Likewise.
17320 (extract_range_from_binary_expr_1): Likewise.
17321 (extract_range_from_unary_expr): Likewise.
17322 (extract_range_from_comparison): Likewise.
17323 (extract_range_basic): Likewise.
17324 (adjust_range_with_scev): Likewise.
17325 (compare_ranges): Likewise.
17326 (compare_range_with_value): Likewise.
17327 (dump_value_range): Likewise.
17328 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17330 (simplify_cond_using_ranges): Adjust.
17332 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17334 * brig-builtins.def: Added a builtin for class_f64.
17335 * builtin-types.def: Added a builtin type needed by class_f64.
17337 2017-05-03 Jason Merrill <jason@redhat.com>
17339 * timevar.def: Add TV_CONSTEXPR.
17341 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17343 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17345 2017-05-03 Martin Jambor <mjambor@suse.cz>
17347 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17348 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17349 * ipa-cp.c (ipcp_propagate_stage): Do not call
17350 ipa_update_after_lto_read.
17351 * ipa-inline.c (ipa_inline): Likewise.
17353 2017-05-03 Martin Jambor <mjambor@suse.cz>
17355 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17356 tag. Added a default constructor and a destructor.
17357 (ipa_edge_args_sum_t): New class;
17358 (ipa_edge_args_sum): Declare.
17359 (ipa_edge_args_vector): Remove declaration.
17360 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17361 (ipa_free_edge_args_substructures): Remove declaration.
17362 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17363 (ipa_edge_args_info_available_for_edge_p): Likewise.
17364 * ipa-prop.c (ipa_edge_args_vector): Removed.
17365 (edge_removal_hook_holder): Likewise.
17366 (edge_duplication_hook_holder): Likewise.
17367 (ipa_edge_args_sum): New variable.
17368 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17369 ipa_edge_args_vector.
17370 (ipa_free_edge_args_substructures): Likewise.
17371 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17372 ipa_edge_args_vector.
17373 (ipa_edge_removal_hook): Turned into method
17374 ipa_edge_args_sum_t::remove.
17375 (ipa_edge_duplication_hook): Turned into method
17376 ipa_edge_args_sum_t::duplicate.
17377 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17378 registering edge hooks.
17379 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17380 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17381 ipa_edge_args_sum instead of ipa_edge_args_vector.
17382 * ipa-profile.c (ipa_profile): Likewise.
17384 2017-05-03 Martin Jambor <mjambor@suse.cz>
17386 * symbol-summary.h (function_summary): New method exists.
17387 (function_summary::symtab_removal): Deallocate through release.
17388 (call_summary): New class.
17389 (gt_ggc_mx): New overload.
17390 (gt_pch_nx): Likewise.
17391 (gt_pch_nx): Likewise.
17393 2017-05-03 Jeff Law <law@redhat.com>
17395 PR tree-optimization/78496
17396 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17397 from simplify_cond_using_ranges. Split off code to walk
17398 backwards through casts into ...
17399 (simplify_cond_using_ranges_2): New function.
17400 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17401 (execute_vrp): After identifying jump threads, call
17402 simplify_cond_using_ranges_2.
17404 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17407 * ipa-inline.h (inline_summary): Add ctor.
17408 (create_ggc): Do not use ggc_cleared_alloc.
17410 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17411 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17413 * gcc.c (handle_braces): Support escaping in switch matching
17415 * doc/invoke.texi (Spec Files): Document it.
17416 Remove superfluous @code markup in items.
17418 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17420 * diagnostic-show-locus.c (struct column_range): New struct.
17421 (get_affected_columns): New function.
17422 (get_printed_columns): New function.
17423 (struct correction): New struct.
17424 (correction::ensure_capacity): New function.
17425 (correction::ensure_terminated): New function.
17426 (struct line_corrections): New struct.
17427 (line_corrections::~line_corrections): New dtor.
17428 (line_corrections::add_hint): New function.
17429 (layout::print_trailing_fixits): Reimplement in terms of the new
17431 (selftest::test_overlapped_fixit_printing): New function.
17432 (selftest::diagnostic_show_locus_c_tests): Call it.
17434 2017-05-03 Nathan Sidwell <nathan@acm.org>
17436 Canonicalize canonical type hashing
17437 * tree.h (type_hash_canon_hash): Declare.
17438 * tree.c (type_hash_list, attribute_hash_list): Move into
17439 type_hash_canon_hash.
17440 (build_type_attribute_qual_variant): Break out hash code calc into
17441 type_hash_canon_hash.
17442 (type_hash_canon_hash): New. Generic type hash computation.
17443 (build_range_type_1, build_array_type_1, build_function_type,
17444 build_method_type_directly, build_offset_type, build_complex_type,
17445 make_vector_type): Call it.
17447 2017-05-03 Richard Biener <rguenther@suse.de>
17449 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17450 When all DRs have unknown misaligned do not always peel
17451 when there is a store but apply the same costing model as if
17452 there were only loads.
17454 2017-05-03 Richard Biener <rguenther@suse.de>
17457 PR tree-optimization/80492
17458 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17459 compare_base_decls returning dont-know properly.
17461 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17463 * config/arm/iterators.md (CCSI): New mode iterator.
17464 (arch): New mode attribute.
17465 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17466 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17467 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17468 code iterator for success result mode.
17469 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17470 the corresponding new insn generators.
17472 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17475 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17476 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17478 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17480 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17481 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17482 (DDR_A): Wrap DDR argument in brackets.
17483 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17484 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17485 (DDR_REVERSED_P): Likewise.
17487 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17489 PR tree-optimization/79472
17490 * tree-switch-conversion.c (struct switch_conv_info): Add
17491 contiguous_range and default_case_nonstandard fields.
17492 (collect_switch_conv_info): Compute contiguous_range and
17493 default_case_nonstandard fields, don't clear final_bb if
17494 contiguous_range and only the default case doesn't have the required
17496 (check_all_empty_except_final): Set default_case_nonstandard instead
17497 of failing if contiguous_range and the default case doesn't have empty
17499 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17500 and only the default case doesn't have the required constants. Skip
17502 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17503 if default_case_nonstandard.
17504 (build_constructors): Build constant 1 just once. Assert that default
17505 values aren't inserted in between cases if contiguous_range. Skip
17507 (build_arrays): Skip virtual phis.
17508 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17509 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17510 Handle virtual phis.
17511 (gen_inbound_check): Handle default_case_nonstandard case.
17512 (process_switch): Adjust check_final_bb caller. Call
17513 gather_default_values with the first non-default case instead of
17514 default case if default_case_nonstandard.
17516 2017-05-02 Nathan Sidwell <nathan@acm.org>
17518 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17519 check. Fix formatting.
17521 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17523 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17524 errors when comparing specialized and unspecialized times.
17526 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17528 * diagnostic-show-locus.c
17529 (layout::should_print_annotation_line_p): Make private.
17530 (layout::print_annotation_line): Make private.
17531 (layout::annotation_line_showed_range_p): Make private.
17532 (layout::show_ruler): Make private.
17533 (layout::print_source_line): Make private. Pass in line and
17534 line_width, rather than calling location_get_source_line. Drop
17536 (layout::print_leading_fixits): New method.
17537 (layout::print_any_fixits): Rename to...
17538 (layout::print_trailing_fixits): ...this, and make private.
17539 Don't print newline fixits.
17540 (diagnostic_show_locus): Move logic for printing one row into...
17541 (layout::print_line): ...this new function. Move the
17542 location_get_source_line call and error-handling from
17543 print_source_line to here. Call print_leading_fixits, and rename
17544 print_any_fixits to print_trailing_fixits.
17545 (selftest::test_fixit_insert_containing_newline): Update now that
17546 newlines are partially supported.
17547 (selftest::test_fixit_insert_containing_newline_2): New test.
17548 (selftest::test_fixit_replace_containing_newline): Update comments.
17549 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17550 * edit-context.c (class added_line): New class.
17551 (class edited_line): Describe newline handling in comment.
17552 (edited_line::actually_edited_p): New method.
17553 (edited_line::print_content): Delete redundant decl.
17554 (edited_line::m_predecessors): New field.
17555 (edited_file::print_content): Call edited_line::print_content.
17556 (edited_file::print_diff): Update to support newlines.
17557 (edited_file::print_diff_hunk): Likewise.
17558 (edited_file::print_run_of_changed_lines): New function.
17559 (edited_file::print_diff_line): Convert to...
17560 (print_diff_line): ...this.
17561 (edited_file::get_effective_line_count): New function.
17562 (edited_line::edited_line): Initialize new field m_predecessors.
17563 (edited_line::~edited_line): Clean up m_predecessors.
17564 (edited_line::apply_fixit): Handle newlines.
17565 (edited_line::get_effective_line_count): New function.
17566 (edited_line::print_content): New function.
17567 (edited_line::print_diff_lines): New function.
17568 (selftest::test_applying_fixits_insert_containing_newline): New
17570 (selftest::test_applying_fixits_replace_containing_newline): New
17572 (selftest::insert_line): New function.
17573 (selftest::test_applying_fixits_multiple_lines): Add example of
17575 (selftest::edit_context_c_tests): Call the new tests.
17577 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17579 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17580 parameter cand. Update dump information.
17581 (get_computation_cost): Update uses.
17583 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17585 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17586 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17587 (get_computation_at, rewrite_use_address): Update use of
17588 get_computation_aff.
17590 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17592 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17593 (get_computation): Delete.
17594 (get_computation_cost): Implement like get_computation_cost_at.
17595 Use get_computation_at.
17596 (get_computation_cost_at): Delete.
17597 (rewrite_use_nonlinear_expr): Use get_computation_at.
17598 (rewrite_use_compare, remove_unused_ivs): Ditto.
17600 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17602 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17604 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17606 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17607 (ivopts_global_cost_for_size): Rename parameter and update uses.
17608 (iv_ca_recount_cost): Update uses.
17609 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17610 candidates seperately in n_invs and n_cands.
17611 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17613 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17615 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17616 (find_inv_vars_cb): New.
17617 (find_depends): Renamed to ...
17618 (find_inv_vars): ... this.
17619 (add_candidate_1, force_var_cost): Call find_inv_vars.
17620 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17622 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17624 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17625 inv_vars. Add inv_exprs.
17626 (struct iv_cand): Rename depends_on to inv_vars.
17627 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17628 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17629 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17630 (dump_cand): Dump inv_vars.
17631 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17632 (record_invariant, find_depends, add_candidate_1): Ditto.
17633 (set_group_iv_cost, force_var_cost): Ditto.
17634 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17635 (get_computation_cost_at, get_computation_cost): Ditto.
17636 (determine_group_iv_cost_generic): Ditto.
17637 (determine_group_iv_cost_address): Ditto.
17638 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17639 (determine_group_iv_costs): Ditto.
17640 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17641 (iv_ca_set_remove_invariants): Renamed to ...
17642 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17643 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17644 (iv_ca_set_add_invariants): Renamed to ...
17645 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17646 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17647 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17648 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17649 (create_new_ivs): Remove useless dump.
17651 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17653 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17655 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17656 (iv_ca_set_no_cp, create_new_iv): Ditto.
17658 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17660 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17662 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17664 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17665 function tree_check2.
17667 2017-05-02 Martin Liska <mliska@suse.cz>
17669 * doc/gcov.texi: Add missing preposition.
17670 * gcov.c (function_info::function_info): Properly fill up
17671 all member variables.
17673 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17675 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17677 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17679 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17681 2017-05-02 Martin Liska <mliska@suse.cz>
17684 * lto-streamer-in.c (lto_read_tree_1): Remove
17685 LTO_STREAMER_DEBUG.
17686 * lto-streamer.c (struct tree_hash_entry): Likewise.
17687 (struct tree_entry_hasher): Likewise.
17688 (tree_entry_hasher::hash): Likewise.
17689 (tree_entry_hasher::equal): Likewise.
17690 (lto_streamer_init): Likewise.
17691 (lto_orig_address_map): Likewise.
17692 (lto_orig_address_get): Likewise.
17693 (lto_orig_address_remove): Likewise.
17694 * lto-streamer.h: Likewise.
17695 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17696 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17698 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17700 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17701 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17702 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17703 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17704 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17705 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17706 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17707 (mm_maskz_sub_ss): New intrinsics.
17708 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17709 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17710 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17711 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17712 (__builtin_ia32_subss_mask_round): New builtins.
17713 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17714 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17715 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17717 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17718 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17720 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17723 2017-05-02 Martin Jambor <mjambor@suse.cz>
17725 PR tree-optimization/78687
17726 * tree-sra.c (access): New field parent.
17727 (process_subtree_disqualification): New function.
17728 (disqualify_candidate): Call it.
17729 (build_accesses_from_assign): Reset write flag if creating an
17731 (build_access_subtree): Fill in parent field and also prpagate
17732 down grp_write flag.
17733 (create_artificial_child_access): New parameter set_grp_write, set
17734 grp_write to its value.
17735 (propagate_subaccesses_across_link): Also propagate grp_write flag
17737 (propagate_all_subaccesses): Push the closest parent back to work
17738 queue if add_access_to_work_queue returned true.
17740 2017-05-02 Richard Biener <rguenther@suse.de>
17742 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17743 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17744 -fstrict-overflow documentation.
17745 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17746 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17747 flag_strict_overflow.
17748 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17749 * lto-opts.c (lto_write_options): Do not stream it.
17750 * lto-wrapper.c (merge_and_complain): Do not handle it.
17751 * opts.c (default_options_table): Do not set -fstrict-overflow.
17752 (finish_options): Likewise do not clear it when sanitizing.
17753 * simplify-rtx.c (simplify_const_relational_operation): Do not
17754 test flag_strict_overflow.
17756 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17758 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17759 using enabled attribute.
17760 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17761 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17762 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17763 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17764 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17765 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17766 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17767 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17768 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17769 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17771 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17773 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17775 2017-05-02 Richard Biener <rguenther@suse.de>
17777 PR tree-optimization/80591
17779 2017-04-10 Richard Biener <rguenther@suse.de>
17781 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17784 2017-05-02 Richard Biener <rguenther@suse.de>
17786 PR tree-optimization/80549
17787 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17788 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17789 headers if they do not have a preheader.
17791 2017-05-02 Martin Liska <mliska@suse.cz>
17794 * common.opt: Fix typo.
17795 * doc/invoke.texi: Likewise.
17797 2017-05-01 Jan Beulich <jbeulich@suse.com>
17799 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17800 swapping, add (x,x,m,x,n) alternative.
17802 2017-05-01 Nathan Sidwell <nathan@acm.org>
17804 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17805 unnecessary unadjusted_alignment check.
17807 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17810 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17812 * gimplify.c (gimplify_cilk_detach): New function.
17813 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17814 * tree-core.h: Document EXPR_CILK_SPAWN.
17815 * tree.h (EXPR_CILK_SPAWN): Define.
17817 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17819 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17820 to use new fixit_hint representation, using the "replace" logic.
17821 (get_line_span_for_fixit_hint): Likewise.
17822 (layout::print_any_fixits): Likewise.
17823 (selftest::test_one_liner_many_fixits): Rename to...
17824 (selftest::test_one_liner_many_fixits_1): ...this, and update
17825 comment and expected output to reflect that the multiple fix-it
17826 hints are now consolidated into one insertion.
17827 (selftest::test_one_liner_many_fixits_2): New test.
17828 (selftest::test_diagnostic_show_locus_one_liner): Update for
17830 (selftest::test_fixit_consolidation): Update for fix-it API
17832 * diagnostic.c (print_parseable_fixits): Likewise.
17833 * edit-context.c (edited_line::m_line_events): Convert from
17834 auto_vec <line_event *> to auto_vec <line_event>.
17835 (class line_event): Convert from abstract base class to a concrete
17836 class, taking over the role of replace_event.
17837 (class insert_event): Delete.
17838 (class replace_event): Rename to class line_event. Convert to
17840 (edit_context::add_fixits): Reimplement.
17841 (edit_context::apply_insert): Delete.
17842 (edit_context::apply_replace): Rename to...
17843 (edit_context::apply_fixit): ...this. Convert to half-open range.
17844 (edited_file::apply_insert): Delete.
17845 (edited_file::apply_replace): Rename to...
17846 (edited_file::apply_fixit): ...this.
17847 (edited_line::~edited_line): Drop deletion of events.
17848 (edited_line::apply_insert): Delete.
17849 (edited_line::apply_replace): Rename to...
17850 (edited_line::apply_fixit): ...this. Convert to half-open range.
17851 Update for change to type of m_line_events.
17852 * edit-context.h (edit_context::apply_insert): Delete.
17853 (edit_context::apply_replace): Rename to...
17854 (edit_context::apply_fixit): ...this.
17856 2017-05-01 Martin Sebor <msebor@redhat.com>
17858 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17861 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17864 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17865 __get_cpuid_max returns 0.
17866 (__get_cpuid_count): Ditto.
17868 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17870 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17871 replacement expression is another instance of one of its arguments.
17873 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17876 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17877 check for stack push/pop autoinc.
17878 * config/i386/i386.c (ix86_agi_dependent): Return false
17879 if the only reason why modified_in_p returned true is that
17880 addr is SP based and set_insn is a push or pop.
17882 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17884 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17887 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17890 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17891 (account_size_time): Use two predicates - exec_pred and
17893 (evaluate_conditions_for_known_args): Compute both clause and
17895 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17896 (inline_summary_t::duplicate): Update.
17897 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17899 (compute_inline_parameters): Likewise.
17900 (estimate_edge_size_and_time): Update caluclation of time.
17901 (estimate_node_size_and_time): Compute both time and nonspecialized
17903 (estimate_ipcp_clone_size_and_time): Update.
17904 (inline_merge_summary): Update.
17905 (do_estimate_edge_time): Update.
17906 (do_estimate_edge_size): Update.
17907 (do_estimate_edge_hints): Update.
17908 (inline_read_section, inline_write_summary): Stream both new predicates.
17909 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17911 (compute_inlined_call_time): Cleanup.
17912 (big_speedup_p): Update.
17913 (edge_badness): Update.
17914 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17915 (size_time_entry): Replace predicate by exec_predicate and
17916 nonconst_predicate.
17917 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17918 (estimate_edge_time): Return also nonspec_time.
17919 (reset_edge_growth_cache): Update.
17921 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17923 PR rtl-optimization/80491
17924 * ifcvt.c (noce_process_if_block): When looking for x setter
17925 with missing else_bb, don't check only the insn right before
17926 cond_earliest, but look for the last insn that x is modified in
17927 within the same bb.
17929 PR rtl-optimization/80491
17930 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17932 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17934 PR tree-optimization/80487
17935 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17937 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17939 PR tree-optimization/79697
17940 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17941 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17942 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17944 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17945 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17947 2017-04-28 Martin Sebor <msebor@redhat.com>
17949 PR tree-optimization/80523
17950 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17951 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17953 (maybe_warn, format_directive, parse_directive): Use new functions.
17954 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17956 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17958 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17960 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17962 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17963 target_header_dir): Set correctly.
17964 * configure: Regenerated.
17965 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17966 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17967 instead of SYSTEM_HEADER_DIR.
17969 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17971 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17972 (estimate_local_effects): Likewise.
17973 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17974 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17975 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17976 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17977 do_estimate_edge_time, estimate_edge_time): Likewise.
17978 * ipa-inline-analysis.c (estimate_node_size_and_time,
17979 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17980 (estimate_time_after_inlining): Remove.
17982 2017-04-28 Martin Liska <mliska@suse.cz>
17984 * doc/gcov.texi: Enhance documentation of gcov.
17986 2017-04-28 Martin Liska <mliska@suse.cz>
17988 * doc/gcov.texi: Sort options in alphabetic order.
17989 * doc/gcov-dump.texi: Likewise.
17990 * doc/gcov-tool.texi: Likewise.
17991 * gcov.c (print_usage): Likewise.
17992 * gcov-dump.c (print_usage): Likewise.
17993 * gcov-tool.c (print_merge_usage_message): Likewise.
17994 (print_rewrite_usage_message): Likewise.
17995 (print_overlap_usage_message): Likewise.
17997 2017-04-28 Martin Liska <mliska@suse.cz>
17999 PR gcov-profile/53915
18000 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18002 2017-04-28 Martin Liska <mliska@suse.cz>
18004 PR gcov-profile/79891
18005 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18006 is marked by compiler as living on a line.
18007 (get_cycles_count): Remove usage of the union.
18008 (output_intermediate_file): Likewise.
18009 (find_source): Fix GNU coding style.
18010 (accumulate_line_counts): Remove old non-all block mode.
18011 (output_lines): Remove usage of the union.
18012 * profile.c (output_location): Include all BBs, even if
18013 belonging to a same line (and file) as a previous BB.
18015 2017-04-28 Martin Liska <mliska@suse.cz>
18017 * gcov.c (process_args): Handle new argument 'w'.
18018 (read_graph_file): Assign ID to BBs.
18019 (output_branch_count): Display BB # if verbose flag is set.
18020 (output_lines): Likewise for arcs.
18021 (print_usage): Add '--verbose' option help.
18022 * doc/gcov.texi: Document --verbose (-w) option.
18024 2017-04-28 Martin Liska <mliska@suse.cz>
18026 * gcov.c (struct block_location_info): New struct.
18027 (process_file): Fill up the new structure.
18028 (read_graph_file): Replace usage of encoding by the newly added
18030 (add_line_counts): Likewise.
18031 (accumulate_line_counts): Remove usage of the union.
18032 (function_info::function_info): New function.
18033 (function_info::~function_info): Likewise.
18034 (process_file): Call delete instead of release_function.
18035 (release_function): Release the function.
18036 (release_structures): Call delete instead of release_function.
18037 (solve_flow_graph): Replace usage of num_blocks.
18038 (find_exception_blocks): Likewise.
18039 (output_lines): Fix GNU coding style.
18041 2017-04-28 Martin Liska <mliska@suse.cz>
18044 * coverage.c (coverage_remove_note_file): New function.
18045 * coverage.h: Declare the function.
18046 * toplev.c (finalize): Clean if an error has been seen.
18048 2017-04-28 Martin Liska <mliska@suse.cz>
18050 PR gcov-profile/80031
18051 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18052 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18053 * gcov.c (read_graph_file): Read just number of blocks.
18054 * profile.c (branch_prob): Do not stream 0 flags per a basic
18057 2017-04-28 Martin Liska <mliska@suse.cz>
18059 * gcov-dump.c (tag_*): Add new argument to declarations.
18060 (dump_gcov_file): Likewise.
18061 (tag_blocks): Add and use new argument depth.
18062 (tag_arcs): Likewise.
18063 (tag_lines): Likewise.
18064 (tag_counters): Likewise.
18065 (tag_summary): Likewise.
18066 (dump_working_sets): Use depth to do a proper indentation.
18068 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18071 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18072 * symtab.c (symtab_node::debug_symtab): Move definition here.
18074 2017-04-28 Richard Biener <rguenther@suse.de>
18076 * lto-streamer.h (LTO_major_version): Bump to 7.
18078 2017-04-28 Richard Biener <rguenther@suse.de>
18080 * tree-vrp.c (assert_info): New struct.
18081 (add_assert_info): New helper.
18082 (register_edge_assert_for_2): Refactor to add asserts to a vector
18084 (register_edge_assert_for_1): Likewise.
18085 (register_edge_assert_for): Likewise.
18086 (finish_register_edge_assert_for): New helper actually registering
18087 asserts where live on edge.
18088 (find_conditional_asserts): Adjust.
18089 (find_switch_asserts): Likewise.
18090 (evrp_dom_walker::try_find_new_range): Generalize.
18091 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18093 2017-04-27 Marek Polacek <polacek@redhat.com>
18096 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18097 arg10 and arg11 to itype.
18099 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18101 * doc/extend.texi (Object Size Checking): Improve grammar.
18103 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18106 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18107 that the logic for permitting reciprocal estimates matches that
18110 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18113 * tree.c (type_cache_hasher::equal): Only compare
18114 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18115 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18116 non-aggregate element types.
18117 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18118 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18120 2017-04-27 Richard Biener <rguenther@suse.de>
18122 PR middle-end/80533
18123 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18124 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18125 keeping a reference to a trailing array.
18127 2017-04-27 Richard Biener <rguenther@suse.de>
18129 PR middle-end/80539
18130 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18131 being in loop-closed SSA form conservatively.
18132 (chrec_fold_multiply_poly_poly): Likewise.
18134 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18136 PR middle-end/79665
18137 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18138 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18140 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18143 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18144 (aarch64_function_arg_alignment): Return unsigned int again, but still
18145 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18146 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18147 Don't emit -Wpsabi note.
18148 (aarch64_function_arg_boundary): Likewise.
18149 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18152 2017-04-26 Nathan Sidwell <nathan@acm.org>
18154 * tree.h (crc32_unsigned_n): Declare.
18155 (crc32_unsigned, crc32_unsigned): Make inline.
18156 * tree.c (crc32_unsigned_bits): Replace with ...
18157 (crc32_unsigned_n): ... this.
18158 (crc32_unsigned, crc32_byte): Remove.
18159 (crc32_string): Remove unnecessary braces.
18161 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18163 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18164 * ipa-inline-analysis.c (MAX_TIME): Remove.
18165 (account_size_time): Use sreal for time.
18166 (dump_inline_summary): Update.
18167 (estimate_function_body_sizes): Update.
18168 (estimate_edge_size_and_time): Update.
18169 (estimate_calls_size_and_time): Update.
18170 (estimate_node_size_and_time): Update.
18171 (inline_merge_summary): Update.
18172 (inline_update_overall_summary): Update.
18173 (estimate_time_after_inlining): Update.
18174 (inline_read_section): Update.
18175 (inline_write_summary): Update.
18176 * ipa-inline.c (compute_uninlined_call_time): Update.
18177 (compute_inlined_call_time): Update.
18178 (recursive_inlining): Update.
18179 (inline_small_functions): Update.
18180 (dump_overall_stats): Update.
18181 * ipa-inline.h: Include sreal.h.
18182 (size_time_entry): Turn time to sreal.
18183 (inline_summary): Turn self_time nad time to sreal.
18185 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18187 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18189 (sreal::stream_out, sreal::stream_in): New.
18190 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18192 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18194 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18197 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18200 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18202 (dimode_scalar_chain::compute_convert_gain): Ditto.
18203 (dimode_scalar_chain::make_vector_copies): Ditto.
18204 (dimode_scalar_chain::convert_reg): Ditto.
18205 (dimode_scalar_chain::convert_insn): Ditto.
18206 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18207 (VI248_AVX512BW_1): New mode iterator.
18208 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18209 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18212 2017-04-25 Martin Sebor <msebor@redhat.com>
18214 PR tree-optimization/80497
18215 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18216 constants are representable in HOST_WIDE_INT.
18217 (parse_directive): Ditto.
18219 2017-04-25 Martin Sebor <msebor@redhat.com>
18222 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18223 (new_zero_array): Adjust signature.
18224 (dom_info::dom_init): Used unsigned rather that size_t.
18225 (dom_info::dom_info): Same.
18227 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18228 Jakub Jelinek <jakub@redhat.com>
18231 * config/arm/arm.c: Include gimple.h.
18232 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18233 returns negative, increment ncrn only if it returned positive.
18234 (arm_needs_doubleword_align): Return int instead of bool,
18235 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18236 members, but if there is any such non-FIELD_DECL
18237 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18238 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18239 returns negative, increment nregs only if it returned positive.
18240 (arm_setup_incoming_varargs): Likewise.
18241 (arm_function_arg_boundary): Emit -Wpsabi note if
18242 arm_needs_doubleword_align returns negative, return
18243 DOUBLEWORD_ALIGNMENT only if it returned positive.
18245 2017-04-25 Marek Polacek <polacek@redhat.com>
18248 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18249 first argument to type.
18251 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18254 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18255 type checks to test for compatibility instead of equality.
18257 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18258 Jakub Jelinek <jakub@redhat.com>
18261 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18263 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18264 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18265 the alignment computation, but return their maximum in warn_alignment.
18266 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18267 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18269 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18270 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18273 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18275 * config/arc/simdext.md (dmpyh): Fix typo.
18277 2017-04-25 Richard Biener <rguenther@suse.de>
18279 PR tree-optimization/80492
18280 * alias.c (compare_base_decls): Handle registers with asm
18281 specification conservatively.
18282 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18283 compare_base_decls returning dont-know properly.
18285 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18287 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18288 (legitimate_offset_address_p): New function.
18289 (arc_legitimate_address_p): Use above function.
18291 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18293 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18295 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18297 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18298 ACCH registers whenever they are available.
18300 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18302 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18303 double regs fix when not used.
18305 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18307 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18309 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18310 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18312 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18314 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18315 when emitting short ADD instructions.
18317 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18319 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18321 (cmpsi_cc_c_insn): Likewise.
18322 (cbranchsi4_scratch): Compute proper instruction length using
18323 compact_hreg_operand.
18324 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18326 2017-04-25 Richard Biener <rguenther@suse.de>
18328 PR middle-end/80509
18329 * passes.c (pass_manager::pass_manager): Initialize
18330 m_name_to_pass_map.
18332 2017-04-25 Richard Biener <rguenther@suse.de>
18334 PR tree-optimization/79201
18335 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18337 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18340 * config/s390/vector.md: Split MEM->GPR vector moves for
18341 non-s_operand addresses.
18343 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18346 * config/s390/predicates.md (reload_const_wide_int_operand): New
18348 * config/s390/s390.md ("movti"): Remove d/P alternative.
18349 ("movti_bigconst"): New pattern definition.
18351 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18354 * s390-protos.h (s390_expand_cs_hqi): Removed.
18355 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18356 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18357 modes as well as CCZ1mode and CCZmode.
18358 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18359 signature of s390_emit_compare_and_swap.
18360 (s390_expand_cs_hqi): Likewise, make static.
18361 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18362 compare-and-swap, in some cases.
18363 (s390_expand_cs): Wrapper function.
18364 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18366 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18367 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18368 patterns for small and large integers. Forbid symref memory operands.
18369 Move expander to s390.c. Require cc register.
18370 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18371 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18372 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18373 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18374 symref memory operands. Remove CC mode and call s390_match_ccmode
18376 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18378 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18380 * config/s390/s390.md (define_peephole2): New peephole to help
18381 combining the load-and-test pattern with volatile memory.
18383 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18385 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18386 with CCZmode for TARGET_Z196.
18388 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18390 PR rtl-optimization/80501
18391 * combine.c (make_compound_operation_int): Set subreg_code to SET
18392 even for AND with mask of the sign bit of mode.
18394 PR rtl-optimization/80500
18395 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18396 sum's initial value.
18398 2017-04-25 Julian Brown <julian@codesourcery.com>
18399 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18401 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18403 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18405 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18407 2017-04-25 Julian Brown <julian@codesourcery.com>
18408 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18410 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18411 (thunderx2t99_sha): New Reservation.
18413 2017-04-25 Julian Brown <julian@codesourcery.com>
18414 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18416 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18417 type for 1-element load.
18419 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18421 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18423 2017-04-24 Martin Jambor <mjambor@suse.cz>
18425 PR tree-optimization/80293
18426 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18427 char arrays not totally scalarizable if it is false.
18428 (analyze_all_variable_accesses): Pass correct value in the new
18429 parameter. Add a statistics counter.
18431 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18433 PR middle-end/79931
18434 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18436 2017-04-24 Richard Biener <rguenther@suse.de>
18438 PR tree-optimization/80494
18439 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18440 out for complex types.
18442 2017-04-24 Richard Biener <rguenther@suse.de>
18444 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18445 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18446 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18447 (DFS): Adjust and never fail.
18448 (sccvn_dom_walker::fail): Remove.
18449 (sccvn_dom_walker::before_dom_children): Adjust.
18450 (run_scc_vn): Likewise and never fail.
18451 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18452 (pass_fre::execute): Likewise.
18454 2017-04-24 Richard Biener <rguenther@suse.de>
18456 PR tree-optimization/79725
18457 * tree-ssa-sink.c (statement_sink_location): Return whether
18458 failure reason was zero uses. Move that check later.
18459 (sink_code_in_bb): Deal with zero uses by removing the stmt
18462 2017-04-24 Richard Biener <rguenther@suse.de>
18465 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18466 pointer-based references.
18468 2017-04-24 Richard Biener <rguenther@suse.de>
18471 * pass_manager.h (pass_manager::operator new): Remove.
18472 (pass_manager::operator delete): Likewise.
18473 * passes.c (pass_manager::operator new): Remove.
18474 (pass_manager::operator delete): Likewise.
18475 (pass_manager::pass_manager): Zero individual pass members.
18477 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18480 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18481 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18482 Check "XEXP (src, 1)" operand here.
18483 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18484 Check "XEXP (src, 1)" operand here.
18485 (dimode_scalar_chain::make_vector_copies): Detect count register
18486 of a shift instruction. Zero extend count register from QImode
18487 to DImode to satisfy vector shift pattern count operand predicate.
18488 Substitute vector shift count operand with a DImode copy.
18489 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18492 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18494 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18495 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18496 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18497 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18498 (UNSPEC_NOREX_MEM): Remove definition.
18500 2017-04-21 Richard Biener <rguenther@suse.de>
18502 PR tree-optimization/79547
18503 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18504 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18505 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18506 without any constraints.
18508 2017-04-21 Richard Biener <rguenther@suse.de>
18510 PR tree-optimization/78847
18511 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18513 2017-04-21 Richard Biener <rguenther@suse.de>
18515 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18516 (build_distinct_type_copy): Likewise.
18517 (build_variant_type_copy): Likewise.
18518 * tree.c (build_qualified_type): Pass down mem-stat info.
18519 (build_distinct_type_copy): Likewise.
18520 (build_variant_type_copy): Likewise.
18522 2017-04-21 Richard Biener <rguenther@suse.de>
18524 PR tree-optimization/80237
18525 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18527 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18528 for a simplified result.
18530 2016-04-21 Richard Biener <rguenther@suse.de>
18532 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18533 sth as strict as a simple_iv but a chrec without symbols and an
18534 operand defined in the loop we are peeling (and not some subloop).
18535 (propagate_constants_for_unrolling): Propagate all constants.
18537 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18540 * config/i386/i386.c (print_reg): Remove assert for disalowed
18541 regno values, call output_operand_lossage instead.
18543 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18546 * config/i386/constraints.md (Yc): New register constraint.
18547 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18548 Use Yc constraint for alternative 2 of operand 0. Remove
18549 preferred_for_speed attribute.
18551 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18553 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18554 lastprivate clauses in SIMT case.
18556 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18558 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18560 2017-04-20 Richard Biener <rguenther@suse.de>
18562 PR tree-optimization/57796
18563 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18564 as N scalar stores.
18565 (vect_model_load_cost): Cost gathers as N scalar loads.
18567 2017-04-20 Richard Biener <rguenther@suse.de>
18569 * ggc-page.c (ggc_allocated_p): Rename to ...
18570 (safe_lookup_page_table_entry): ... this and return the lookup
18572 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18574 2017-04-20 Richard Biener <rguenther@suse.de>
18576 PR tree-optimization/80453
18577 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18578 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18579 from the conditions.
18580 (vn_phi_eq): Pass them down.
18581 (vn_phi_lookup): Record them.
18582 (vn_phi_insert): Likewise.
18584 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18586 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18587 uninitialized variable warning to avoid buffer overrun.
18589 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18592 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18593 is suppressed for '{ 0 }' in C.
18595 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18597 * BASE-VER: Set to 8.0.0.
18599 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18601 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18602 priority .init_array and .fini_array section with SECTION_NOTYPE
18605 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18607 PR middle-end/80423
18608 * tree.h (build_array_type): Add typeless_storage default argument.
18609 * tree.c (type_cache_hasher::equal): Also compare
18610 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18611 (build_array_type): Add typeless_storage argument, set
18612 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18614 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18615 (build_array_type): Likewise. Add typeless_storage argument.
18617 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18618 Jakub Jelinek <jakub@redhat.com>
18620 PR tree-optimization/80426
18621 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18622 operation on symbolic operands, also compute the overflow for the
18623 invariant part when the operation degenerates into a negation.
18625 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18628 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18629 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18632 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18634 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18637 * config/avr/avr.c (tree.h): Include it.
18638 (cgraph.h): Include it.
18639 (avr_encode_section_info): Don't warn for uninitialized progmem
18640 variable if it's just an alias.
18642 2017-04-19 Richard Biener <rguenther@suse.de>
18645 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18646 when needed by AutoPGO.
18648 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18651 * doc/lto.texi: Remove an extra 'that'.
18653 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18655 PR rtl-optimization/80429
18656 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18657 are only used in debug insns.
18659 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18660 Vladimir Makarov <vmakarov@redhat.com>
18662 * config/sparc/predicates.md (input_operand): Add comment. Return
18663 true for any memory operand when LRA is in progress.
18664 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18666 2017-04-18 Jeff Law <law@redhat.com>
18669 * mips.md ({return,simple_return}_internal): Do not overwrite
18672 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18674 PR tree-optimization/80443
18675 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18676 instead of adding 1, subtract -1 and similarly instead of subtracting
18679 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18681 PR rtl-optimization/80357
18682 * haifa-sched.c (tmp_bitmap): New variable.
18683 (model_recompute): Handle duplicate use records.
18684 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18685 (free_global_sched_pressure_data): Free it.
18687 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18690 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18691 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18692 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18693 instead of SYSTEM_HEADER_DIR.
18695 2017-04-18 Jeff Law <law@redhat.com>
18697 PR middle-end/80422
18698 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18699 predecessors after walking up the insn chain.
18701 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18704 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18705 sizetype type into debug info.
18707 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18710 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18711 unneeded test for TARGET_UPPER_REGS_SF.
18712 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18714 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18717 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18718 instead of gsi_after_labels.
18720 2017-04-18 Jeff Law <law@redhat.com>
18722 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18726 2017-04-13 Jeff Law <law@redhat.com>
18727 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18728 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18730 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18733 * config/avr/avr.c (intl.h): Include it.
18734 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18736 2017-04-18 Martin Liska <mliska@suse.cz>
18738 PR gcov-profile/78783
18739 * gcov-tool.c (gcov_output_files): Validate that destination
18740 file is either removed by the tool or by a user.
18742 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18743 Guy Benyei <guybe@mellanox.com>
18745 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18746 block, and do not negate it, the stored id is already negative.
18748 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18750 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18752 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18755 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18756 masks of options that should be turned off if the VSX vector
18757 options are turned off.
18758 (OTHER_P8_VECTOR_MASKS): Likewise.
18759 (OTHER_VSX_VECTOR_MASKS): Likewise.
18760 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18761 rs6000_disable_incompatible_switches to validate no type switches
18763 (rs6000_incompatible_switch): New function to disallow turning on
18764 other vector options if -mno-vsx, -mno-power8-vector, or
18765 -mno-power9-vector are specified.
18767 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18769 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18771 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18773 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18774 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18775 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18776 (ARG_POINTER_CFA_OFFSET): Likewise.
18778 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18780 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18781 conditions to take advantage of various optimizations.
18783 2017-04-13 Jeff Law <law@redhat.com>
18785 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18786 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18787 (zero_extendsidi2_dext): Likewise.
18789 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18792 * fold-const.c (fold_ternary_loc): Revert
18793 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18796 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18798 PR rtl-optimization/80343
18799 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18800 new scratch pseudo.
18802 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18805 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18806 to ubsan_encode_value.
18808 2017-04-13 Jeff Law <law@redhat.com>
18810 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18811 appearing in DEBUG_INSNs.
18813 2017-04-13 Martin Liska <mliska@suse.cz>
18815 PR gcov-profile/80413
18816 * gcov-io.c (gcov_write_string): Copy to buffer just when
18817 allocated size is greater than zero.
18819 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18822 * dwarf2out.c (decls_for_scope): Ignore declarations of
18823 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18825 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18828 * ipa-visibility.c (non_local_p): Fix typos.
18829 (localize_node): When localizing symbol in same comdat group,
18830 dissolve the group only when we know external symbols are going
18832 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18834 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18836 PR tree-optimization/79390
18837 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18838 order does not result in usable sequence, retry with reversed operand
18844 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18845 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18846 op0 instead of fold_convert_loc (loc, type, arg0).
18848 2017-04-12 Jeff Law <law@redhat.com>
18850 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18851 has a delay slot in the generated code.
18853 * config/cris/cris.md (cris_preferred_reload_class): Return
18854 GENNONACR_REGS rather than GENERAL_REGS.
18856 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18859 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18860 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18861 signedness of the result type.
18863 2017-04-12 Richard Biener <rguenther@suse.de>
18864 Jeff Law <law@redhat.com>
18866 PR tree-optimization/80359
18867 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18868 trim stores to TARGET_MEM_REFs.
18870 2017-04-12 Richard Biener <rguenther@suse.de>
18872 PR tree-optimization/79390
18873 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18874 threading case even more.
18876 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18879 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18880 for quad_address_p for TImode, instead of just not indexed_address.
18882 2017-04-12 Richard Biener <rguenther@suse.de>
18883 Bernd Edlinger <bernd.edlinger@hotmail.de>
18885 PR middle-end/79671
18886 * alias.c (component_uses_parent_alias_set_from): Handle
18887 TYPE_TYPELESS_STORAGE.
18888 (get_alias_set): Likewise.
18889 * tree-core.h (tree_type_common): Add typeless_storage flag.
18890 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18891 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18892 for types containing members with TYPE_TYPELESS_STORAGE.
18893 (place_field): Likewise.
18894 (layout_type): Likewise for ARRAY_TYPE.
18895 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18896 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18897 TYPE_TYPELESS_STORAGE.
18898 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18900 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18903 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18904 first argument to type.
18906 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18910 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18911 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18912 (rs6000_expand_binop_builtin): Likewise.
18913 (rs6000_expand_ternop_builtin): Likewise; also add missing
18914 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18915 vshasigma built-ins.
18916 * doc/extend.texi: Document that vec_xxpermdi's third argument
18917 must be a constant.
18919 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18921 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18922 Use shift_const cost parameter when calculating gain of STV shifts.
18924 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18926 PR rtl-optimization/70478
18927 * lra-constraints.c (process_alt_operands): Check memory for
18928 disfavoring memory insn operand.
18930 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18932 PR middle-end/80100
18933 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18934 left shift in unsigned HOST_WIDE_INT type.
18936 PR rtl-optimization/80385
18937 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18938 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18941 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18942 if they have any depend clauses.
18944 2017-04-11 Martin Liska <mliska@suse.cz>
18947 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18948 * ipa-split.c (split_function): Create a local comdat symbol
18949 if caller is in a comdat group.
18951 2017-04-11 Martin Liska <mliska@suse.cz>
18954 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18957 2017-04-11 Martin Sebor <msebor@redhat.com>
18959 PR middle-end/80364
18960 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18961 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18963 (directive::set_width, directive::set_precision, format_character):
18965 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18968 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18971 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18972 conflict, set target->arch_name instead of target->cpu_name.
18974 2017-04-11 Richard Biener <rguenther@suse.de>
18976 PR tree-optimization/80374
18977 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18978 build_zero_cst, remove fold_convertible_p check again.
18980 2017-04-11 Martin Liska <mliska@suse.cz>
18983 * ubsan.c (instrument_object_size): Do not instrument register
18986 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18989 * config/i386/i386-builtin-types.def
18990 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18991 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18992 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18993 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18994 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18995 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18996 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18997 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18998 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18999 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19000 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19001 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19002 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19003 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19004 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19005 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19006 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19007 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19008 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19009 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19010 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19011 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19012 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19013 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19014 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19015 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19016 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19017 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19018 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19019 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19020 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19021 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19022 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19023 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19024 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19025 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19026 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19027 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19028 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19029 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19030 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19031 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19032 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19033 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19034 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19035 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19037 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19038 flag to second_arg_count, handle 4 argument function type _COUNT
19039 aliases, handle second_arg_count on second argument rather than last.
19041 2017-04-10 Jeff Law <law@redhat.com>
19043 PR tree-optimization/80374
19044 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19045 record anything if we can not convert integer_zero_node to the
19048 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19051 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19052 Enhance special handling given to the TARGET_P9_MINMAX option in
19053 relation to certain other options.
19055 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19057 PR tree-optimization/80153
19058 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19059 remove POINTER_PLUS_EXPR's base part directly, rather than through
19062 2017-04-10 Richard Biener <rguenther@suse.de>
19063 Bin Cheng <bin.cheng@arm.com>
19065 PR tree-optimization/80153
19066 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19067 the first element of pointer type aff_tree. Build result expr in
19069 (add_elt_to_tree): Convert to type unconditionally. Remove other
19070 fold_convert calls.
19071 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19072 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19074 2017-04-10 Richard Biener <rguenther@suse.de>
19076 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19079 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19081 PR rtl-optimization/70478
19082 * lra-constraints.c (curr_small_class_check): New.
19083 (update_and_check_small_class_inputs): New.
19084 (process_alt_operands): Update curr_small_class_check. Disfavor
19085 alternative insn memory operands. Check available regs for small
19088 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19091 * config/mips/mips.opt (-mvirt): Update description.
19092 * doc/invoke.texi (-mvirt): Likewise.
19094 2017-04-10 Richard Biener <rguenther@suse.de>
19096 PR middle-end/80362
19097 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19098 looking for NEGATE_EXPR in -A / -B to A / B folding.
19100 2017-04-10 Martin Liska <mliska@suse.cz>
19102 PR gcov-profile/80224
19103 * gcov.c (print_usage): Fix usage string.
19104 (get_gcov_intermediate_filename): Remove.
19105 (output_gcov_file): Use both for normal and intermediate format.
19106 (generate_results): Do not initialize special file for
19107 intermediate format.
19109 2017-04-10 Richard Biener <rguenther@suse.de>
19111 PR tree-optimization/80304
19112 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19115 2017-04-10 Nathan Sidwell <nathan@acm.org>
19118 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19119 (rs6000_init_builtins): Use it.
19121 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19123 * config/arm/arm.md (<mrc>): Add mode to SET source.
19124 (<mrrc>): Likewise.
19126 2017-04-10 Richard Biener <rguenther@suse.de>
19128 PR middle-end/80344
19129 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19131 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19134 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19135 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19136 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19137 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19138 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19139 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19140 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19141 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19142 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19143 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19144 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19145 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19146 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19147 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19148 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19149 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19150 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19151 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19152 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19153 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19154 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19155 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19156 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19158 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19160 PR rtl-optimization/70478
19161 * lra-constraints.c: Reverse the last patch.
19163 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19165 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19166 Add comment for WCHAR_T.
19168 2017-04-08 Martin Liska <mliska@suse.cz>
19171 2017-04-07 Martin Liska <mliska@suse.cz>
19174 * ipa-split.c (split_function): Add function part to a same comdat
19177 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19180 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19182 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19184 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19185 (vec_pairing): ...to this.
19186 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19187 (rs6000_sched_init): Adjust for name change.
19188 (struct rs6000_sched_context): Likewise.
19189 (rs6000_init_sched_context): Likewise.
19190 (rs6000_set_sched_context): Likewise.
19192 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19198 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19200 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19201 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19202 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19204 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19206 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19208 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19210 PR rtl-optimization/70703
19211 * ira-color.c (update_conflict_hard_regno_costs): Use
19212 int64_t instead of HOST_WIDE_INT.
19214 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19216 PR rtl-optimization/70478
19217 * lra-constraints.c (process_alt_operands): Disfavor alternative
19218 insn memory operands.
19220 2017-04-07 Jeff Law <law@redhat.com>
19222 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19223 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19225 2017-04-07 Martin Liska <mliska@suse.cz>
19228 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19229 Show error message instead of an ICE.
19231 2017-04-07 Martin Liska <mliska@suse.cz>
19234 * ipa-split.c (split_function): Add function part to a same comdat
19237 2017-04-07 Richard Biener <rguenther@suse.de>
19239 PR middle-end/80341
19240 * tree.c (get_unwidened): Also handle ! for_type case for
19242 * convert.c (do_narrow): Split out from ...
19243 (convert_to_integer_1): ... here. Do not pass final truncation
19244 type to get_unwidened for TRUNC_DIV_EXPR.
19246 2017-04-07 Richard Biener <rguenther@suse.de>
19248 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19250 (aff_combination_const): Adjust.
19251 (aff_combination_scale): Likewise.
19252 (aff_combination_add_elt): Likewise.
19253 (aff_combination_add_cst): Likewise.
19254 (aff_combination_convert): Likewise.
19255 (add_elt_to_tree): Likewise. Remove unused argument.
19256 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19258 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19260 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19262 * config/arm/arm.c (arm_default_short_enums): Use
19263 ARM_DEFAULT_SHORT_ENUMS.
19264 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19266 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19269 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19270 members with redundant out-of-class redeclaration.
19272 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19275 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19276 * config/i386/i386.md (*zero_extendsidi2):
19277 Add (?*x,*x) and (?*v,*v) alternatives.
19279 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19282 * config/i386/i386.c (ix86_expand_builtin)
19283 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19284 mode from insn data. Convert operands to insn operand mode.
19285 Copy operands that don't satisfy insn predicate to a register.
19287 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19289 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19292 2017-04-06 Richard Biener <rguenther@suse.de>
19294 PR tree-optimization/80334
19295 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19296 preserve alignment of accesses.
19298 2017-04-06 Richard Biener <rguenther@suse.de>
19300 PR tree-optimization/80262
19301 * tree-sra.c (build_ref_for_offset): Preserve address-space
19303 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19304 Drop useless address-space information on MEM_REF offsets.
19306 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19308 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19310 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19312 PR rtl-optimization/70703
19313 * ira-color.c (update_conflict_hard_regno_costs): Use
19314 HOST_WIDE_INT instead of long.
19316 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19319 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19320 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19322 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19323 for x86_64 target. Handle -m3dnowa option.
19325 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19327 PR rtl-optimization/70703
19328 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19329 (update_conflict_hard_regno_costs): Use long instead of unsigned
19330 arithmetic for cost calculation.
19332 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19333 Bernd Edlinger <bernd.edlinger@hotmail.de>
19336 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19339 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19342 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19343 ptr_mode with Pmode throughout.
19344 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19345 into probe_stack_range and use DImode.
19347 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19350 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19351 call_eh_return is true.
19353 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19355 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19356 Initialize last_match_fntype_index.
19358 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19361 * tree-nvr.c: Include internal-fn.h.
19362 (pass_return_slot::execute): Ignore internal calls without
19365 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19366 Richard Biener <rguenther@suse.de>
19369 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19370 captures used multiple times, except for the last use.
19371 * generic-match-head.c: Include gimplify.h.
19373 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19375 PR tree-optimization/79390
19376 * target.h (struct noce_if_info): Declare.
19377 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19378 * target.def (noce_conversion_profitable_p): New target hook.
19379 * ifcvt.h (struct noce_if_info): New type, moved from ...
19380 * ifcvt.c (struct noce_if_info): ... here.
19381 (noce_conversion_profitable_p): Renamed to ...
19382 (default_noce_conversion_profitable_p): ... this. No longer
19384 (noce_try_store_flag_constants, noce_try_addcc,
19385 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19386 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19387 instead of noce_conversion_profitable_p.
19388 * config/i386/i386.c: Include ifcvt.h.
19389 (ix86_option_override_internal): Don't override
19390 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19391 (ix86_noce_conversion_profitable_p): New function.
19392 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19393 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19394 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19395 * doc/tm.texi: Regenerated.
19397 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19399 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19402 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19405 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19406 instructions for small multiply cores.
19408 2017-04-04 Jeff Law <law@redhat.com>
19410 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19412 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19413 that are not set by the loop over the elements.
19415 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19418 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19419 int mode, convert_modes it to mode as unsigned, otherwise use
19420 lowpart_subreg to mode rather than SImode.
19421 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19422 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19423 Use DImode instead of SImode for the shift count operand.
19424 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19427 2017-04-04 Richard Biener <rguenther@suse.de>
19429 PR middle-end/80281
19430 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19431 arithmetic done for the negate or the plus. Simplify.
19432 (A - (-B) -> A + B): Likewise.
19433 * fold-const.c (split_tree): Make sure to not negate pointers.
19435 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19437 PR rtl-optimization/60818
19438 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19439 a compare of comparisons with the thing compared if this results
19440 in a different machine mode.
19442 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19444 * alias.c (base_alias_check): Fix typo in comment.
19445 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19446 * cgraphunit.c (symbol_table::compile): Likewise.
19447 * collect2.c (maybe_run_lto_and_relink): Likewise.
19448 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19449 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19450 * config/avr/avr.c (avr_map_op_t): Likewise.
19451 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19452 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19453 * config/epiphany/epiphany.md (movcc): Likewise.
19454 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19455 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19457 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19458 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19459 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19460 * config/sh/sh.c (sh_rtx_costs): Likewise.
19461 * fold-const.c (fold_truth_andor): Likewise.
19462 * genautomata.c (collapse_flag): Likewise.
19463 * gengtype.h (struct type::u::s): Likewise.
19464 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19465 * input.c (FORMAT_AMOUNT): Likewise.
19466 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19467 (known_aggs_to_agg_replacement_list): Likewise.
19468 * ipa-inline-analysis.c: Likewise.
19469 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19470 * ipa-polymorphic-call.c
19471 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19472 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19473 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19475 * modulo-sched.c (apply_reg_moves): Likewise.
19476 * omp-expand.c (build_omp_regions_1): Likewise.
19477 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19478 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19479 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19480 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19481 * value-prof.c: Likewise.
19482 * var-tracking.c (val_reset): Likewise.
19484 2017-04-03 Richard Biener <rguenther@suse.de>
19486 PR tree-optimization/80275
19487 * fold-const.c (split_address_to_core_and_offset): Handle
19490 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19492 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19493 descriptors is at least equal to that of functions.
19495 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19497 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19499 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19502 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19503 (mov<IMOD4:mode>): New expander.
19504 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19506 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19508 PR rtl-optimization/79405
19509 * fwprop.c (propagations_left): New variable.
19510 (forward_propagate_into): Decrement it.
19511 (fwprop_init): Initialize it.
19512 (fw_prop): If the variable has reached zero, stop propagating.
19513 (fwprop_addr): Ditto.
19515 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19518 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19519 a FUNCTION_DECL, pass it as decl instead of origin to
19522 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19524 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19527 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19530 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19531 TARGET_VSX_SMALL_INTEGER.
19533 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19535 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19536 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19538 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19540 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19541 extraction from odd-numbered MSA register.
19543 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19545 PR middle-end/80173
19546 * expmed.c (store_bit_field_1): Don't attempt to create
19547 a word subreg out of hard registers wider than word if they
19548 have HARD_REGNO_NREGS of 1 for their mode.
19550 PR middle-end/80163
19551 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19552 conversions to integer types wider than word and pointer.
19555 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19556 (rtx_equal_for_cselib_p): Pass 0 to it.
19557 * cselib.c (cselib_hasher::equal): Likewise.
19558 (rtx_equal_for_cselib_1): Add depth argument. If depth
19559 is 128, don't look up VALUE locs and punt. Increment
19560 depth in recursive calls when walking VALUE locs.
19562 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19564 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19565 (make_gcov_file_name): Use the canonical path name for generating
19567 (read_line): Fix handling of files with ascii null bytes.
19569 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19571 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19572 to initialise a vector register instead
19573 of using a const_int.
19575 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19577 PR translation/80189
19578 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19579 diagnostic messages.
19581 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19584 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19585 (dfp_diex_<mode>): Update mode of operand 1.
19586 * doc/extend.texi (dxex, dxexq): Document change to return type.
19587 (diex, diexq): Document change to argument type.
19589 2017-03-30 Martin Jambor <mjambor@suse.cz>
19592 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19593 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19594 it reflects the signature changes performed at the callee side.
19595 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19596 to cgraph_build_function_type_skip_args.
19597 (build_function_decl_skip_args): Adjust call to the above function.
19599 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19602 * config/i386/sse.md
19603 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19604 register as dest whenever it is a MEM not rtx_equal_p to the
19605 corresponding dup operand, and when forcing into reg move the
19606 reg into the memory afterwards.
19607 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19608 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19609 for the force_reg mode.
19610 (avx512vl_vextractf128<mode>): Use register as dest either
19611 always when a MEM, or when it is a MEM not rtx_equal_p to the
19612 corresponding dup operand, or even not when it is a CONST_VECTOR
19613 depending on the mode and lo vs. hi.
19614 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19616 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19617 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19618 Likewise. Require that operands[2] is even.
19619 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19620 Remove extraneous parens. Require that operands[2] is a multiple
19622 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19623 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19624 disallow memory then.
19626 2017-03-30 Richard Biener <rguenther@suse.de>
19628 PR tree-optimization/77498
19629 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19630 to non-constants over backedges.
19632 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19634 PR rtl-optimization/80233
19635 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19636 as last_combined_insn. Do not test for BARRIER_P separately.
19638 2017-03-29 Andreas Schwab <schwab@suse.de>
19641 * calls.c (prepare_call_address): Convert funexp to Pmode before
19642 copying to temp reg.
19644 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19646 PR tree-optimization/80158
19647 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19648 Handle possible future case of more than one alternate
19650 (replace_rhs_if_not_dup): Likewise.
19651 (replace_one_candidate): Likewise.
19653 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19655 PR rtl-optimization/80193
19656 * ira.c (ira): Do not check allocation for LRA.
19658 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19660 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19661 (nvptx_output_simt_exit): Declare.
19662 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19663 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19664 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19665 (nvptx_declare_function_name): ...here. Emit declaration of local
19666 memory space buffer for omp_simt_enter insn.
19667 (nvptx_output_unisimt_switch): New.
19668 (nvptx_output_softstack_switch): New.
19669 (nvptx_output_simt_enter): New.
19670 (nvptx_output_simt_exit): New.
19671 * config/nvptx/nvptx.h (struct machine_function): New fields
19672 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19673 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19674 (UNSPECV_SIMT_EXIT): Ditto.
19675 (omp_simt_enter_insn): New insn.
19676 (omp_simt_enter): New expansion.
19677 (omp_simt_exit): New insn.
19678 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19680 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19681 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19682 (expand_GOMP_SIMT_EXIT): New.
19683 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19684 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19685 (GOMP_SIMT_EXIT): Ditto.
19686 * target-insns.def (omp_simt_enter): New insn.
19687 (omp_simt_exit): Ditto.
19688 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19690 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19691 (lower_rec_input_clauses): Likewise.
19692 (lower_lastprivate_clauses): Handle SIMT privatization.
19694 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19695 (ompdevlow_adjust_simt_enter): New.
19696 (find_simtpriv_var_op): New.
19697 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19698 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19700 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19701 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19702 (copy_decl_for_dup_finish): Ditto.
19704 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19706 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19709 * config/i386/i386.c (ix86_option_override_internal): Always
19710 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19712 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19714 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19716 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19718 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19719 mark new edge's irreducible flag accordign to it.
19720 (vect_do_peeling): Check loop preheader edge's irreducible flag
19721 and pass it to function slpeel_add_loop_guard.
19723 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19725 PR tree-optimization/80218
19726 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19727 Update block frequencies and counts.
19729 2017-03-28 Richard Biener <rguenther@suse.de>
19731 PR tree-optimization/78644
19732 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19733 of a simplification result we may not use it at all.
19735 2017-03-28 Richard Biener <rguenther@suse.de>
19738 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19739 without arguments, generate default definition of a SSA name.
19741 2017-03-28 Richard Biener <rguenther@suse.de>
19743 PR middle-end/80222
19744 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19745 TYPE_REF_CAN_ALIAS_ALL references.
19746 * fold-const.c (fold_indirect_ref_1): Likewise.
19748 2017-03-28 Martin Liska <mliska@suse.cz>
19751 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19752 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19754 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19755 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19757 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19758 (EXTRA_SPECS): Define.
19759 (SUBTARGET_EXTRA_SPECS): Likewise.
19760 (SUBTARGET_CPP_SPEC): Likewise.
19761 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19762 SUBTARGET_EXTRA_SPECS.
19763 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19765 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19767 * config/arc/simdext.md (vst64_insn): Update pattern.
19768 (vld32wh_insn): Likewise.
19769 (vld32wl_insn): Likewise.
19770 (vld64_insn): Likewise.
19771 (vld32_insn): Likewise.
19773 2017-03-28 Marek Polacek <polacek@redhat.com>
19776 * fold-const.c (fold_comparison): Use protected_set_expr_location
19777 instead of SET_EXPR_LOCATION.
19779 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19781 * tree.c (add_expr): Avoid name lookup warning.
19783 2017-03-27 Jeff Law <law@redhat.com>
19785 PR tree-optimization/80216
19786 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19787 function name. Limit recursion depth.
19788 (record_temporary_equivalences): Corresponding changes.
19790 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19792 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19795 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19798 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19800 * cfgcleanup.c (reg_note_cfa_p): New array.
19801 (insns_have_identical_cfa_notes): New function.
19802 (old_insns_match_p): Don't cross-jump in between /f
19803 and non-/f instructions. If both i1 and i2 are frame related,
19804 verify all CFA notes, their order and content.
19806 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19809 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19810 HImode and SImode with zero extend to DImode to one insn.
19811 (bswap<mode>2_extenddi): Likewise.
19812 (bswapsi2_extenddi): Likewise.
19813 (bswaphi2_extendsi): Likewise.
19814 (bswaphi2): Combine bswap HImode and SImode into one insn.
19815 Separate memory insns from swapping register.
19816 (bswapsi2): Likewise.
19817 (bswap<mode>2): Likewise.
19818 (bswaphi2_internal): Delete, no longer used.
19819 (bswapsi2_internal): Likewise.
19820 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19821 store, and gpr<-gpr swap insns.
19822 (bswap<mode>2_store): Likewise.
19823 (bswaphi2_reg): Register only splitter, combine with the splitter.
19824 (bswaphi2 splitter): Likewise.
19825 (bswapsi2_reg): Likewise.
19826 (bswapsi2 splitter): Likewise.
19827 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19828 the insns into load, store, and register/register insns.
19829 (bswapdi2_ldbrx): Likewise.
19830 (bswapdi2_load): Likewise.
19831 (bswapdi2_store): Likewise.
19832 (bswapdi2_reg): Likewise.
19834 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19836 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19837 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19839 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19842 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19844 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19845 special handling for target option conflicts between dform
19846 options (-mpower9-dform, -mpower9-dform-vector,
19847 -mpower9-dform-scalar) and -mno-direct-move.
19849 2017-03-27 Richard Biener <rguenther@suse.de>
19851 PR tree-optimization/80181
19852 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19854 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19856 * config/arc/predicates.md (move_double_src_operand): Replace the
19857 call to move_double_src_operand with a call to address_operand.
19859 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19861 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19862 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19863 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19865 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19867 * config/arc/predicates.md (long_immediate_loadstore_operand):
19868 Consider scaled addresses cases.
19870 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19872 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19873 restored when in interrupt.
19874 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19875 doesn't have delay slot.
19877 2017-03-27 Richard Biener <rguenther@suse.de>
19880 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19881 inlined thunk clones.
19883 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19886 * asan.c (instrument_derefs): Copy over last operand from
19887 original COMPONENT_REF to the new COMPONENT_REF with
19888 DECL_BIT_FIELD_REPRESENTATIVE.
19889 * ubsan.c (instrument_object_size): Likewise.
19891 2017-03-27 Richard Biener <rguenther@suse.de>
19893 PR tree-optimization/80170
19894 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19895 sure DR/SCEV didnt fold in constants we do not see when looking
19896 at the reference base alignment.
19898 2017-03-27 Richard Biener <rguenther@suse.de>
19900 PR middle-end/80171
19901 * gimple-fold.c (fold_ctor_reference): Properly guard against
19902 NULL return value from canonicalize_constructor_val.
19904 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19907 * config/i386/i386.c (ix86_expand_builtin)
19908 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19909 flags reg setting and flags reg using instructions.
19910 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19911 clobbering instructions to zero extend op2.
19913 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19915 * doc/install.texi (Configuration) <--with-aix-soname>:
19916 Update link to AIX ld.
19918 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19920 PR rtl-optimization/80160
19921 PR rtl-optimization/80159
19922 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19923 reg_alternate_class into account.
19925 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19928 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19929 to consider in curr_insn_transform.
19931 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19933 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19934 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19935 and emit_mode_inner.
19937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19939 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19940 argument to the overloaded builtin variants. Use the new flag to
19941 deprecate certain builtin variants.
19942 * config/s390/s390-builtin-types.def: Add new builtin types.
19943 * config/s390/s390-builtins.h: Support new flags field for
19944 overloaded builtins.
19945 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19946 (s390_macro_to_expand): Enable vector float data type.
19947 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19948 builtins by incrementing the __VEC__ version number.
19949 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19951 (s390_resolve_overloaded_builtin): Emit error messages depending
19952 on the builtin flags.
19953 * config/s390/s390.c (s390_expand_builtin): Support additional
19954 flags argument. Change error message to match the messages
19955 emitted in s390-c.c.
19956 * config/s390/s390.md: New UNSPEC_* constants.
19957 (op_type): Add new instruction types.
19958 * config/s390/vecintrin.h: Add new builtins and test data class
19960 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19961 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19962 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19963 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19964 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19965 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19967 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19968 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19969 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19970 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19972 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19973 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19974 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19975 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19977 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19978 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19979 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19980 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19981 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19982 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19983 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19985 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19986 ("vec_scatter_element<V_HW_4:mode>_DI")
19987 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19988 ("vec_fpint<mode>", "vflls")
19989 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19990 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19991 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19992 ("*vec_cmphe<mode>_cc"): ... these.
19994 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19995 mode constant instead of magic value.
19997 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19999 * config/s390/s390.c (s390_expand_vec_compare): Support other
20000 vector floating point modes than just V2DF.
20001 (s390_expand_vcond): Likewise.
20002 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20003 (s390_cannot_change_mode_class): Prevent mode changes between TF
20004 and V1TF in vector registers.
20005 * config/s390/s390.md (DF, SF): New mode attributes.
20006 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20007 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20008 SFmode support for VRs.
20009 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20011 (VFT, VF_HW): New mode iterators.
20012 (vw, sdx): New mode attributes.
20013 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20014 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20015 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20016 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20017 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20018 also the new vector floating point modes. Renaming to ...
20020 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20021 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20022 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20023 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20024 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20025 ("vec_unordered<mode>"): ... these.
20027 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20028 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20029 ("*vec_extendv2df"): New insn definitions.
20031 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20033 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20034 ("mulditi3_2", "*muldi3_sign"): New patterns.
20035 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20036 rename the pattern definition.
20038 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20040 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20042 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20046 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20047 instruction if possible.
20048 * config/s390/vector.md (vec_halfnumelts): New mode
20050 ("*vec_vllezlf<mode>"): New pattern.
20052 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20054 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20055 ("popcountv4si2", "popcountv2di2"): Rename to ...
20056 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20057 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20059 ("popcount<mode>2_vxe"): New pattern.
20061 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20063 * common/config/s390/s390-common.c (processor_flags_table): Add
20065 * config.gcc: Add arch12.
20066 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20067 Default to arch12 for unknown CPU model numbers.
20068 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20069 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20070 PROCESSOR_max sanity check.
20071 * config/s390/s390-opts.h (enum processor_type): Add
20073 * config/s390/s390.c (processor_table): Add arch12.
20074 (s390_expand_builtin): Add check for B_VXE flag.
20075 (s390_issue_rate): Add PROCESSOR_ARCH12.
20076 (s390_get_sched_attrmask): Likewise.
20077 (s390_get_unit_mask): Likewise.
20078 (s390_sched_score): Enable z13 scheduling for arch12.
20079 (s390_sched_reorder): Likewise.
20080 (s390_sched_variable_issue): Likewise.
20081 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20083 (s390_tune_attr): Use z13 scheduling also for arch12.
20084 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20085 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20086 (TARGET_VXE_P): New macros.
20087 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20088 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20089 * config/s390/s390.opt: Add arch12 as processor_type.
20091 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20093 * config/s390/s390.md
20094 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20095 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20096 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20098 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20099 Rename expanders to ...
20101 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20102 ("fixuns_truncdddi2_emu"): ... these.
20104 ("fixuns_trunc<mode>si2_emu"): New expander.
20106 ("*fixuns_truncdfdi2_z13"): Rename to ...
20107 ("*fixuns_truncdfdi2_vx"): ... this.
20109 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20111 * config/s390/2964.md: Remove the single element vector compare
20112 instructions which are no longer used.
20113 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20115 (s390_canonicalize_comparison): Remove handling of DFmode
20117 (s390_expand_vec_compare_scalar): Remove function.
20118 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20119 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20121 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20125 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20127 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20128 will anyway by matched by mov<mode>_64dfp.
20130 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20132 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20133 vlef/vstef. Add missing operand to vleif.
20135 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20137 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20138 pair for all vector types with 64 bit elements.
20139 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20140 * config/s390/vector.md (V_HW_64): ... here.
20141 (V_128_NOSINGLE): New mode iterator.
20142 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20143 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20144 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20145 ("*vec_load_pairv2di"): Change to ...
20146 ("*vec_load_pair<mode>"): ... this one.
20148 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20150 * config/s390/constraints.md: Add comments.
20151 (jKK): Reject element sizes > 8 bytes.
20152 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20154 * config/s390/s390.md: Add the s_operand checks formerly in
20155 s390_split_ok_p to various splitters where they are still
20157 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20158 for 128 bit vectors. Plus two splitters.
20160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20162 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20165 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20168 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20169 error if the boundary argument is not constant.
20171 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20173 PR rtl-optimization/80112
20174 * loop-doloop.c (doloop_condition_get): Don't check condition
20175 if cmp isn't SET with IF_THEN_ELSE src.
20177 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20179 PR tree-optimization/80158
20180 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20181 replacing a candidate statement, also replace it for the
20182 candidate's alternate interpretation.
20183 (replace_rhs_if_not_dup): Likewise.
20184 (replace_one_candidate): Likewise.
20186 2017-03-24 Richard Biener <rguenther@suse.de>
20188 PR tree-optimization/80167
20189 * graphite-isl-ast-to-gimple.c
20190 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20192 (translate_isl_ast_to_gimple::get_rename): Likewise.
20194 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20196 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20197 handling of certain combinations of target options, including the
20198 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20199 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20201 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20204 * config/arm/arm.md (*load_multiple): Add reload_completed to
20205 matching condition.
20207 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20208 Richard Biener <rguenth@suse.de>
20210 PR tree-optimization/79908
20211 PR tree-optimization/80136
20212 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20213 been cast away, gimplify_and_add suffices.
20215 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20217 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20219 2017-03-23 Richard Biener <rguenther@suse.de>
20221 PR tree-optimization/80032
20222 * gimplify.c (gimple_push_cleanup): Forced unconditional
20223 cleanups still have to go to the conditional_cleanups
20226 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20228 PR tree-optimization/80072
20229 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20231 (next_operand_entry_id): Change type to unsigned int.
20232 (sort_by_operand_rank): Make sure to return the right return value
20233 even if unsigned fields are bigger than INT_MAX.
20234 (struct oecount): Change cnt and id type to unsigned int.
20235 (oecount_hasher::equal): Formatting fix.
20236 (oecount_cmp): Make sure to return the right return value
20237 even if unsigned fields are bigger than INT_MAX.
20238 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20241 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20242 TREE_READONLY on result if writing it more than once.
20245 * doc/invoke.texi (-fsanitize=thread): Document that with
20246 -fnon-call-exceptions atomics are not able to throw
20250 * tsan.c: Include tree-eh.h.
20251 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20252 maybe_clean_or_replace_eh_stmt where needed.
20253 (instrument_memory_accesses): Add cfg_changed argument.
20254 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20255 if it returned true.
20256 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20258 PR rtl-optimization/63191
20259 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20260 wrapper function, moved the whole old content into ...
20261 (ix86_delegitimize_address_1): ... this. New inline function.
20262 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20263 true as last argument instead of ix86_delegitimize_address.
20265 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20267 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20268 cortexa57_branch_cost.
20270 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20272 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20274 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20277 * doc/md.texi (Constraints): Document wA constraint.
20278 * config/rs6000/constraints.md (wA): New.
20279 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20280 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20281 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20282 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20284 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20287 * gimplify.c (is_oacc_declared): New function.
20288 (oacc_default_clause): Use it to set default flags for acc declared
20289 variables inside parallel regions.
20290 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20291 declared variables.
20292 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20293 declare attribute to any decl as necessary.
20295 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20298 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20299 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20300 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20301 (arm_arch_lpae): This.
20302 * config/arm/arm.c (arm_arch7ve): Rename into ...
20303 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20304 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20307 2017-03-22 Martin Liska <mliska@suse.cz>
20310 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20311 error message instead of an ICE.
20313 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20315 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20317 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20319 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20321 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20324 2017-03-21 Martin Sebor <msebor@redhat.com>
20326 * doc/extend.texi: Use "cannot" instead of "can't."
20327 * doc/hostconfig.texi: Same.
20328 * doc/install.texi: Same.
20329 * doc/invoke.texi: Same.
20330 * doc/loop.texi: Same.
20331 * doc/md.texi: Same.
20332 * doc/objc.texi: Same.
20333 * doc/rtl.texi: Same.
20334 * doc/tm.texi: Same.
20335 * doc/tm.texi.in: Same.
20336 * doc/trouble.texi: Same.
20338 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20341 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20342 (collect_checksum_attributes): Set it.
20343 (die_checksum_ordered): Use it.
20345 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20347 PR tree-optimization/79908
20348 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20349 change: For a VA_ARG whose LHS has been cast away, use
20350 force_gimple_operand to construct the side effects.
20352 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20354 PR translation/80001
20355 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20356 more amenable to translation.
20357 (oacc_loop_auto_partitions): Likewise.
20359 2017-03-21 Marek Polacek <polacek@redhat.com>
20360 Martin Sebor <msebor@redhat.com>
20362 PR tree-optimization/80109
20363 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20364 on INTEGRAL_TYPE_P.
20366 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20367 Segher Boessenkool <segher@kernel.crashing.org>
20370 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20371 check reg_used_between_p between insn and one of succ or succ2
20372 depending on if succ is artificial insn not inserted into insn
20375 2017-03-21 Martin Liska <mliska@suse.cz>
20377 PR gcov-profile/80081
20378 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20379 * doc/gcc.texi: Include gcov-dump stuff.
20380 * doc/gcov-dump.texi: New file.
20382 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20384 PR rtl-optimization/79150
20385 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20386 conditional jump, if the jump is the last insn of the loop.
20388 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20389 Richard Biener <rguenth@suse.de>
20391 PR tree-optimization/79908
20392 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20393 been cast away, use force_gimple_operand to construct the side
20396 2017-03-21 Martin Liska <mliska@suse.cz>
20398 PR libfortran/79956
20399 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20402 2017-03-21 Brad Spengler <spender@grsecurity.net>
20405 * plugin.c (htab_hash_plugin): New function.
20406 (add_new_plugin): Use it and adjust.
20407 (parse_plugin_arg_opt): Adjust.
20408 (init_one_plugin): Likewise.
20410 2017-03-21 Richard Biener <rguenther@suse.de>
20412 PR tree-optimization/80032
20413 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20414 if set force the cleanup to happen unconditionally.
20415 (gimplify_target_expr): Push inserted clobbers with force_uncond
20416 to avoid them being removed by control-dependent DCE.
20418 2017-03-21 Richard Biener <rguenther@suse.de>
20420 PR tree-optimization/80122
20421 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20422 va_arg_pack_len when the inlined call stmt requires pack
20424 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20426 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20429 * tsan.c (instrument_builtin_call): If the memory model argument
20430 is not a constant, assume it is valid.
20433 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20436 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20438 PR rtl-optimization/79910
20439 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20440 if its dest is used by an insn before I2 (other than the combined
20441 insns themselves, which are properly handled already).
20443 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20446 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20448 * combine.c (record_used_regs): New static function.
20449 (try_combine): Handle situations where there is an additional
20450 instruction between I2 and I3 which needs to have a LOG_LINK
20454 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20456 * combine.c (try_combine): Delete redundant i1 test. Call
20457 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20459 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20462 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20463 alternatives 13/14.
20465 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20467 PR tree-optimization/80054
20468 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20469 the optimization if a PHI or any of its arguments is not dominated
20470 by the candidate's basis. Use gphi* rather than gimple* as
20472 (replace_profitable_candidates): Clean up a gimple* variable that
20473 should be a gphi* variable.
20475 2017-03-20 Martin Sebor <msebor@redhat.com>
20478 * doc/extend.texi (attribute constructor): Document present limitation.
20480 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20483 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20484 __POWER9_VECTOR__ #ifdef control, change template definition to
20485 use Power9-specific built-in function.
20486 (vec_any_eq): Likewise.
20487 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20488 to control outcomes from this test.
20489 (vector_ae_<mode>p): For VEC_F modes, likewise.
20491 2017-03-20 Ian Lance Taylor <iant@google.com>
20493 * config/i386/i386.c (ix86_function_regparm): Save an extra
20494 register for -fsplit-stack with DECL_STATIC_CHAIN.
20496 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20499 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20500 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20502 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20504 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20506 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20509 2017-03-20 Marek Polacek <polacek@redhat.com>
20512 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20514 2017-03-20 Richard Biener <rguenther@suse.de>
20516 PR tree-optimization/80113
20517 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20518 allocate extra SSA name for PHI def.
20519 (add_close_phis_to_outer_loops): Likewise.
20520 (add_close_phis_to_merge_points): Likewise.
20521 (copy_loop_close_phi_args): Likewise.
20522 (copy_cond_phi_nodes): Likewise.
20524 2017-03-20 Martin Liska <mliska@suse.cz>
20526 PR middle-end/79753
20527 * tree-chkp.c (chkp_build_returned_bound): Do not build
20528 returned bounds for a LHS that's not a BOUNDED_P type.
20530 2017-03-20 Martin Liska <mliska@suse.cz>
20534 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20535 COMPLEX_CST and VECTOR_CST.
20537 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20540 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20541 target operand. A new splitter adds the clobber statement in case
20542 the target operand is dead anyway.
20544 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20546 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20547 to age-old versions of binutils and glibc.
20549 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20551 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20553 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20555 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20557 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20559 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20560 requirement for binutils 2.13.
20562 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20564 * combine.c (try_combine): Delete redundant i1 test. Call
20565 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20567 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20569 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20570 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20572 <riscv64-*-elf>: Re-arrange section
20573 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20574 <riscv32-*-linux>: Likewise.
20575 <riscv64-*-elf>: Likewise
20576 <riscv64-*-linux>: Likewise.
20578 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20581 * aarch64.opt(verbose-cost-dump): Fix typo.
20583 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20586 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20587 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20589 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20591 * reload.c (find_reloads): When reloading a nonoffsettable address,
20592 use RELOAD_OTHER for it and its address reloads.
20594 PR rtl-optimization/79910
20595 * combine.c (record_used_regs): New static function.
20596 (try_combine): Handle situations where there is an additional
20597 instruction between I2 and I3 which needs to have a LOG_LINK
20600 2017-03-17 Jeff Law <law@redhat.com>
20602 PR tree-optimization/71437
20603 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20604 conditional in the hash table first.
20605 (vrp_dom_walker::before_dom_children): Extract condition from
20606 ASSERT_EXPR. Record condition, its inverion and any implied
20607 conditions as well.
20609 2017-03-17 Marek Polacek <polacek@redhat.com>
20610 Markus Trippelsdorf <markus@trippelsdorf.de>
20612 PR tree-optimization/80079
20613 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20616 2017-03-17 Richard Biener <rguenther@suse.de>
20618 PR middle-end/80075
20619 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20620 Properly verify the LHS before the RHS possibly claims to be
20622 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20625 2017-03-17 Martin Jambor <mjambor@suse.cz>
20627 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20628 (List of -O2 options): Likewise.
20629 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20632 2017-03-17 Tom de Vries <tom@codesourcery.com>
20634 * gcov-dump.c (print_usage): Print bug_report_url.
20636 2017-03-17 Richard Biener <rguenther@suse.de>
20638 PR middle-end/80050
20639 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20640 (parser::peek): Likewise.
20642 2017-03-17 Richard Biener <rguenther@suse.de>
20644 PR tree-optimization/80048
20645 * sese.c (free_sese_info): Properly release rename_map and
20646 copied_bb_map elements.
20648 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20650 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20651 Add linked-list forward and backlinks. Insert on
20652 construction, remove on destruction.
20653 (class pass_store_merging): Add m_stores_head field.
20654 (pass_store_merging::terminate_and_process_all_chains):
20655 Iterate over m_stores_head list.
20656 (pass_store_merging::terminate_all_aliasing_chains):
20658 (pass_store_merging::execute): Check for debug stmts first.
20659 Push new chains onto the m_stores_head stack.
20661 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20664 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20665 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20666 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20668 2017-03-16 Jeff Law <law@redhat.com>
20670 PR tree-optimization/71437
20671 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20672 member function. Implementation moved into after_dom_children
20673 member function and into the threader's thread_outgoing_edges
20675 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20676 some code into new thread_outgoing_edges.
20677 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20678 definition. Simplify marker handling (do it here). Assume we always
20679 have the available expression and the const/copies tables.
20680 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20682 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20683 * tree-vrp.c (equiv_stack): No longer file scoped.
20684 (vrp_dom_walker): New class.
20685 (vrp_dom_walker::before_dom_children): New member function.
20686 (vrp_dom_walker::after_dom_children): Likewise.
20687 (identify_jump_threads): Setup domwalker. Use it rather than
20688 walking edges in a random order by hand. Simplify setup/finalization.
20689 (finalize_jump_threads): Remove.
20690 (vrp_finalize): Do not call identify_jump_threads here.
20691 (execute_vrp): Do it here instead and call thread_through_all_blocks
20694 PR tree-optimization/71437
20695 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20697 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20699 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20700 (dom_opt_dom_walker::thread_across_edge): Remove
20701 handle_dominating_asserts argument. All callers changed.
20702 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20703 changes. Remove calls to lhs_of_dominating_assert. Other
20704 uses of handle_dominating_asserts turn into unconditional code
20705 (simplify_control_stmt_condition_1): Likewise.
20706 (simplify_control_stmt_condition): Likewise.
20707 (thread_through_normal_block, thread_across_edge): Likewise.
20708 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20709 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20710 object if it is not an SSA_NAME.
20711 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20712 before calling into the VRP specific simplifiers.
20713 (identify_jump_threads): Remove handle_dominating_asserts
20716 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20719 * tree-diagnostic.c (default_tree_printer): No longer static.
20720 * tree-diagnostic.h (default_tree_printer): New prototype.
20722 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20724 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20725 Change ins into fmov.
20727 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20729 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20730 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20731 Use h_con constraint for operand 1.
20732 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20733 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20735 2017-03-15 Jeff Law <law@redhat.com>
20737 PR tree-optimization/71437
20738 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20739 (record_temporary_equivalences): Use it.
20741 PR tree-optimization/71437
20742 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20743 tree-ssa-scopedtables.
20744 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20745 (record_conditions, record_cond, vuse_eq): Likewise.
20746 (record_edge_info): Adjust to API tweak of record_conditions.
20747 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20748 (record_temporary_equivalences, optimize_stmt): Likewise.
20749 (eliminate_redundant_computations): Likewise.
20750 (record_equivalences_from_stmt): Likewise.
20751 * tree-ssa-scopedtables.c: Include options.h and params.h.
20752 (vuse_eq): New function, moved from tree-ssa-dom.c
20753 (build_and_record_new_cond): Likewise.
20754 (record_conditions): Likewise. Accept vector of conditions rather
20755 than edge_equivalence structure for first argument.
20756 for the first argument.
20757 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20758 from tree-ssa-dom.c.
20759 (avail_exprs_stack::record_cond): Likewise.
20760 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20761 from tree-ssa-dom.c.
20762 (avail_exprs_stack): Add new member functions lookup_avail_expr
20764 (record_conditions): Declare.
20766 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20769 * lra-constraints.c (process_alt_operands): Increase reject for
20770 reloading an input/output operand.
20772 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20775 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20776 insns to convert from signed/unsigned char/short to IEEE 128-bit
20778 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20780 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20783 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20784 subreg of inner mode for values already in registers.
20786 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20788 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20789 iteration reg is used after the loop.
20791 2017-03-14 Martin Sebor <msebor@redhat.com>
20793 PR tree-optimization/79800
20794 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20795 precision in negative-positive range.
20796 (format_floating): Call non-const overload with adjusted precision.
20798 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20801 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20804 2017-03-14 Martin Sebor <msebor@redhat.com>
20806 PR middle-end/80020
20807 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20808 * builtins.def (aligned_alloc): Use it.
20811 * Makefile.in (GTFILES): Add calls.c.
20812 * calls.c: Include "gt-calls.h".
20814 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20816 PR rtl-optimization/79728
20817 * regs.h (struct target_regs): New field
20818 x_contains_allocatable_regs_of_mode.
20819 (contains_allocatable_regs_of_mode): New macro.
20820 * reginfo.c (init_reg_sets_1): Initialize it, and change
20821 contains_reg_of_mode so it includes global regs as well.
20822 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20823 rather than contains_regs_of_mode.
20825 2017-03-14 Martin Liska <mliska@suse.cz>
20827 * doc/invoke.texi: Document options that can't be combined with
20828 -fcheck-pointer-bounds.
20830 2017-03-14 Martin Liska <mliska@suse.cz>
20832 PR middle-end/79831
20833 * doc/invoke.texi (-Wchkp): Document the option.
20835 2017-03-14 Martin Liska <mliska@suse.cz>
20837 * Makefile.in: Install gcov-dump.
20839 2017-03-14 Martin Liska <mliska@suse.cz>
20841 * multiple_target.c (expand_target_clones): Bail out for
20842 an invalid attribute.
20844 2017-03-14 Richard Biener <rguenther@suse.de>
20846 * alias.c (struct alias_set_entry): Pack properly.
20847 * cfgloop.h (struct loop): Likewise.
20848 * cse.c (struct set): Likewise.
20849 * ipa-utils.c (struct searchc_env): Likewise.
20850 * loop-invariant.c (struct invariant): Likewise.
20851 * lra-remat.c (struct cand): Likewise.
20852 * recog.c (struct change_t): Likewise.
20853 * rtl.h (struct address_info): Likewise.
20854 * symbol-summary.h (function_summary): Likewise.
20855 * tree-loop-distribution.c (struct partition): Likewise.
20856 * tree-object-size.c (struct object_size_info): Likewise.
20857 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20858 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20859 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20860 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20861 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20862 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20863 (struct _stmt_vec_info): Likewise.
20865 2017-03-14 Martin Liska <mliska@suse.cz>
20868 * multiple_target.c (create_dispatcher_calls): Check that
20869 a target can create a function dispatcher.
20871 2017-03-14 Martin Liska <mliska@suse.cz>
20874 * multiple_target.c (expand_target_clones): Drop local.local
20875 flag for default implementation.
20877 2017-03-14 Richard Biener <rguenther@suse.de>
20879 PR tree-optimization/80030
20880 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20882 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20884 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20885 gcc_fallthrough() instead of __attribute__((fallthrough));
20887 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20889 * doc/gcc.texi: Remove "up" link to (DIR).
20890 * doc/gccint.texi: Ditto.
20892 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20894 * doc/install.texi (Specific) <avr>: Remove reference to
20897 2017-03-13 Jeff Law <law@redhat.com>
20899 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20900 attribute rather than comments.
20902 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20903 match_scratch operand is highest.
20905 2017-03-13 Martin Liska <mliska@suse.cz>
20907 PR middle-end/78339
20908 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20909 is a CHKP clone, use original declaration.
20911 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20913 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20914 (arc_conditional_register_usage): Use a different allocation order
20915 when optimizing for size.
20916 * common/config/arc/arc-common.c (arc_option_optimization_table):
20917 Section anchors default on when optimizing for size.
20919 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20921 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20923 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20925 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20926 * config/arc/arc.md (cpu_facility): Add cd variant.
20927 (*movqi_insn): Add code density variant.
20928 (*movhi_insn): Likewise.
20929 (*movqi_insn): Likewise.
20930 (*addsi3_mixed): Likewise.
20931 (subsi3_insn): Likewise.
20933 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20935 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20937 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20939 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20940 expressions with MINUS and UNARY ops.
20942 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20945 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20947 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20948 between vec_select and vector argument.
20949 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20950 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20951 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20952 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20953 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20954 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20956 2017-03-13 Richard Biener <rguenther@suse.de>
20959 * params.def (vect-max-peeling-for-alignment): Fix typo.
20961 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20963 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20964 issue that only occurred with binutils below 2.18.
20966 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20968 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20969 refer to binutils 2.11/2.12 minimum.
20971 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20973 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20974 ftp.kernel.org and simplify binutils requirement.
20976 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20978 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20980 (Optimize Options): Ditto. Also remove redundancy.
20982 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20984 PR translation/79848
20985 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20987 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20988 to G_ to avoid double translation.
20990 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20992 PR translation/79923
20993 * auto-profile.c (get_combined_location): Convert leading
20994 character of diagnostics to lower case and remove trailing period.
20995 (read_profile): Likewise for various diagnostics.
20996 * config/arm/arm.c (arm_option_override): Remove trailing period
20997 from various diagnostics.
20998 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20999 (msp430_expand_delay_cycles): Likewise.
21001 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21004 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21005 full command-line argument, rather than just "str".
21006 (aarch64_validate_march): Likewise.
21007 (aarch64_validate_mtune): Likewise.
21009 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21011 PR rtl-optimization/78911
21012 * lra-assigns.c (must_not_spill_p): New function.
21013 (spill_for): Use it.
21015 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21017 PR tree-optimization/79981
21018 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21019 ATOMIC_COMPARE_EXCHANGE ifn result.
21020 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21021 IFN_ATOMIC_COMPARE_EXCHANGE.
21023 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21026 * opts.c (parse_sanitizer_options): Add missing question mark to
21027 "did you mean" message.
21029 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21031 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21033 (VMULEUH_UNS): Likewise.
21034 (VMULOUB_UNS): Likewise.
21035 (VMULOUH_UNS): Likewise.
21036 * config/rs6000/rs6000.c (builtin_function_type): Remove
21037 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21039 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21042 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21043 x with result of extra_parsing_for_operand_code_0.
21044 (function_reader::extra_parsing_for_operand_code_0): Convert
21045 return type from void to rtx, returning x. When reading
21046 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21047 larger size containing struct block_symbol.
21049 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21052 -mfloat128-hardware without -m64.
21054 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21057 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21058 entries to the case statement that marks unsigned arguments to
21059 overloaded functions.
21061 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21063 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21064 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21066 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21069 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21070 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21072 2017-03-10 Martin Liska <mliska@suse.cz>
21076 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21077 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21080 2017-03-10 Olivier Hainque <hainque@adacore.com>
21082 * tree-switch-conversion (array_value_type): Start by resetting
21083 candidate type to it's main variant.
21085 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21087 PR rtl-optimization/79909
21088 * combine.c (try_combine): Use simplify_replace_rtx on individual
21089 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21090 of the whole CALL_INSN_FUNCTION_USAGE.
21092 PR tree-optimization/79972
21093 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21094 get_range_info on SSA_NAMEs. Formatting fixes.
21096 2017-03-10 Richard Biener <rguenther@suse.de>
21097 Jakub Jelinek <jakub@redhat.com>
21099 PR tree-optimization/77975
21100 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21101 edge to be constant.
21102 (get_val_for): For constant x return it. Formatting fix.
21103 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21104 would use the same bases as the current one.
21106 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21108 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21109 instead of vec_select for V1TImode.
21110 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21112 (VSX_LE_128): Add V1TI to this mode iterator.
21113 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21114 (*vsx_le_perm_store_<mode>): Likewise.
21115 (pre-reload splitter for VSX stores): Likewise.
21116 (post-reload splitter for VSX stores): Likewise.
21117 (*vsx_xxpermdi2_le_<mode>): Likewise.
21118 (*vsx_lxvd2x2_le_<mode>): Likewise.
21119 (*vsx_stxvd2x2_le_<mode>): Likewise.
21121 2017-03-09 Michael Eager <eager@eagercon.com>
21123 Correct failures with --enable-checking=yes,rtl.
21125 * config/microblaze/microblaze.c (microblaze_expand_shift):
21126 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21127 test for const0_rtx.
21128 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21129 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21131 2017-03-09 Richard Biener <rguenther@suse.de>
21133 PR tree-optimization/79977
21134 * graphite-scop-detection.c (scop_detection::merge_sese):
21135 Handle the case of extra exits to blocks dominating the entry.
21137 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21139 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21142 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21144 PR rtl-optimization/79949
21145 * lra-constraints.c (process_alt_operands): Check memory when
21146 trying to predict a cycle. Print about the overall increase.
21148 2017-03-09 Richard Biener <rguenther@suse.de>
21150 PR middle-end/79971
21151 * gimple-expr.c (useless_type_conversion_p): Preserve
21152 TYPE_SATURATING for fixed-point types.
21154 2017-03-09 Richard Biener <rguenther@suse.de>
21157 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21158 alignment of BLKmode params.
21160 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21163 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21164 (VALL_NO_V2Q): Likewise.
21166 * config/aarch64/aarch64-simd.md
21167 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21169 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21170 VALL_NO_V2Q mode iterator.
21171 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21173 2017-03-09 Martin Liska <mliska@suse.cz>
21175 PR tree-optimization/79631
21176 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21177 tree_int_cst_sign_bit just for INTEGER constants.
21179 2017-03-09 Martin Liska <mliska@suse.cz>
21183 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21186 2017-03-09 Marek Polacek <polacek@redhat.com>
21189 * tree.c (inchash::add_expr): Handle TREE_VEC.
21191 2017-03-09 Martin Liska <mliska@suse.cz>
21194 (chkp_narrow_size_and_offset): New function.
21195 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21196 (void chkp_parse_bit_field_ref): New function.
21197 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21198 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21200 2017-03-09 Martin Liska <mliska@suse.cz>
21203 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21204 (chkp_find_bounds_1): Remove gcc_unreachable.
21206 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21209 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21210 BUILT_IN_SYNC*, determine the access type from the size suffix and
21211 always build a MEM_REF with that type. Handle forgotten
21212 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21215 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21216 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21217 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21218 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21219 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21220 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21221 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21222 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21223 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21224 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21225 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21226 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21227 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21228 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21229 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21230 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21231 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21232 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21233 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21234 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21235 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21236 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21237 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21238 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21239 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21240 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21241 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21242 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21243 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21244 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21245 definitions outside of __OPTIMIZE__ guarded section.
21248 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21249 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21250 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21251 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21254 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21256 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21257 ("vfenez<mode>"): Add missing constraints.
21259 2017-03-08 Martin Sebor <msebor@redhat.com>
21262 * config/nds32/nds32.c (nds32_option_override):
21263 Fix misspelled diagnostic.
21265 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21268 * gimplify.c (gimplify_omp_for): Replace index var in outer
21269 taskloop statement with an artificial variable and add
21270 OMP_CLAUSE_PRIVATE clause for it.
21272 2017-03-08 Richard Biener <rguenther@suse.de>
21274 PR tree-optimization/79955
21275 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21276 for accesses that are completely outside of the variable.
21278 2017-03-08 Andrew Haley <aph@redhat.com>
21280 PR tree-optimization/79943
21281 * tree-ssa-loop-split.c (compute_new_first_bound): When
21282 calculating the new upper bound, (END-BEG) should be added, not
21285 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21287 * config/avr/avr.md (setmemhi): Make sure match_dup
21288 operand number comes before match_scratch.
21290 2017-03-08 Richard Biener <rguenther@suse.de>
21292 PR tree-optimization/79920
21293 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21294 with ncopies == 1 to ...
21295 (vect_transform_slp_perm_load): ... here. Properly compute
21296 all element loads by iterating VF times over the group. Do
21297 not handle ncopies (computed in a broken way) in
21298 vect_create_mask_and_perm.
21300 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21303 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21304 is a uniform vector, use uniform_vector_p return value instead of
21305 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21307 2017-03-07 Marek Polacek <polacek@redhat.com>
21309 PR middle-end/79809
21310 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21311 (alloca_call_type): Likewise.
21313 2017-03-07 Martin Liska <mliska@suse.cz>
21315 * gcov.c (process_args): Put comment to correct location.
21317 2017-03-07 Martin Liska <mliska@suse.cz>
21319 PR middle-end/68270
21320 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21321 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21322 (chkp_narrow_bounds_for_field): Likewise.
21323 (chkp_parse_array_and_component_ref): Pass one more argument to
21326 2017-03-07 Richard Biener <rguenther@suse.de>
21328 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21331 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21333 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21334 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21336 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21339 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21340 to end of description.
21341 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21343 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21345 PR rtl-optimization/79901
21346 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21348 (*avx512f_<code><mode>3<mask_name>): ... this.
21349 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21350 iterator instead of VI8_AVX2_AVX512BW.
21352 PR rtl-optimization/79901
21353 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21354 min/max expander, expand it using expand_vec_cond_expr.
21357 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21360 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21363 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21364 to void * for PCH reasons.
21365 * dwarf2out.c (output_loc_operands, output_die): Cast
21366 v.val_vec.array to unsigned char *.
21368 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21371 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21374 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21376 PR rtl-optimization/79571
21377 * lra-constraints.c (process_alt_operands): Calculate static
21378 reject and subtract it from overall when only addresses will be
21381 2017-03-06 Julia Koval <julia.koval@intel.com>
21384 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21385 incoming stack boundary to 128 for 64-bit targets.
21387 2017-03-06 Richard Biener <rguenther@suse.de>
21389 PR tree-optimization/79894
21390 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21391 to NULL after folding it.
21393 2017-03-06 Richard Biener <rguenther@suse.de>
21395 PR tree-optimization/79824
21396 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21397 check disabling peeling for gaps.
21399 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21401 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21402 attributes): Document gettimeofday.
21404 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21406 * config/s390/s390.c (s390_option_override_internal): Set
21407 PARAM_MIN_VECT_LOOP_BOUND
21409 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21411 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21412 * config/s390/s390.md: Likewise.
21414 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21417 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21418 (<avx2_avx512>_perm<mode>): Rename to ...
21419 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21421 (<avx512>_perm<mode>_mask): Rename to ...
21422 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21424 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21425 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21426 instead of VI8F_256_512.
21427 (avx512f_perm<mode>): New define_expand.
21428 (avx512f_perm<mode>_mask): Likewise.
21429 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21430 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21432 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21434 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21435 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21437 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21439 2017-03-06 Martin Liska <mliska@suse.cz>
21442 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21443 when having a SSA NAME w/o VAR_DECL assigned to it.
21445 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21447 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21448 msa_dpsub_<su>_d): Fix MODE for vec_select.
21450 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21452 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21454 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21456 2017-03-06 Richard Biener <rguenther@suse.de>
21458 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21459 * plugin.c (register_plugin_info): Likewise.
21460 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21462 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21464 * config/i386/sse.md (sse_storehps, sse_storelps,
21465 avx_<castmode><avxsizesuffix>_<castmode>,
21466 avx512f_<castmode><avxsizesuffix>_<castmode>,
21467 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21468 in condition that at least one operand is not a MEM.
21470 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21472 PR middle-end/79805
21473 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21474 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21476 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21477 gimple_call_nothrow_p flag based on whether original builtin can throw.
21478 If it can, emit following stmts on the fallthrough edge.
21479 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21480 don't create new bb if inserting just debug stmts on the edge, try to
21481 insert them on the fallthru bb or just reset debug stmts.
21483 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21486 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21487 restore recog_data (including the operand rtxes inside it) around
21488 the call to get_insn_template.
21490 2017-03-03 Martin Sebor <msebor@redhat.com>
21492 PR tree-optimization/79699
21493 * context.c (context::~context): Free MPFR caches to avoid
21494 a memory leak on program exit.
21496 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21498 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21499 Use wide_int::ulow () instead of .elt (0).
21501 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21503 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21504 (*pushxf): Limit oF constraint to 32bit targets and add oC
21505 constraint for 64bit targets.
21506 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21507 (*pushdf): Change rmF constraint to rmC.
21509 2017-03-03 Martin Liska <mliska@suse.cz>
21511 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21512 Remove unused variable.
21514 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21517 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21518 is a memory operand, increase num_memory.
21519 (ix86_expand_args_builtin): Likewise.
21521 2017-03-03 Jan Hubicka <jh@suse.cz>
21524 * ipa-devirt.c (maybe_record_node): Properly handle
21525 __cxa_pure_virtual visibility.
21527 2017-03-03 Martin Liska <mliska@suse.cz>
21529 PR tree-optimization/79803
21530 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21532 (pass_loop_prefetch::execute): Disabled optimization if an
21533 assumption about L1 cache size is not met.
21535 2017-03-03 Martin Liska <mliska@suse.cz>
21537 PR rtl-optimization/79574
21538 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21539 (hash_scan_set): Likewise.
21540 (dump_hash_table): Likewise.
21541 (hoist_code): Likewise.
21543 2017-03-03 Richard Biener <rguenther@suse.de>
21545 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21546 in place of uhigh ().
21547 (fixed_convert_from_real): Likewise.
21549 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21552 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21554 2017-03-03 Richard Biener <rguenther@suse.de>
21556 PR middle-end/79818
21557 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21558 TYPE_OVERFLOW_UNDEFINED check.
21560 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21562 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21564 (vector_ae_<mode>_p): Likewise.
21565 (vector_nez_<mode>_p): Likewise.
21566 (vector_ne_v2di_p): Likewise.
21567 (vector_ae_v2di_p): Likewise.
21568 (vector_ne_<mode>_p): Likewise.
21569 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21571 (vsx_tsqrt<mode>2_fe): Likewise.
21573 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21576 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21578 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21580 PR rtl-optimization/79780
21581 * cprop.c (one_cprop_pass): When second and further conditional trap
21582 in a single basic block is turned into an unconditional trap, turn it
21583 into a deleted note to avoid RTL verification failures.
21585 2017-03-02 Richard Biener <rguenther@suse.de>
21587 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21589 2017-03-02 Richard Biener <rguenther@suse.de>
21591 PR tree-optimization/79345
21593 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21594 param and abort the walk, returning -1 if it is hit.
21595 (walk_aliased_vdefs): Take a limit param and pass it on.
21596 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21597 defaulting to 0 and return a signed int.
21598 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21599 (check_defs): New helper.
21600 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21601 about uninitialized memory.
21602 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21603 bogus uninitialized warning.
21604 (fixed_convert_from_real): Likewise.
21606 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21608 PR tree-optimization/66768
21609 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21610 iv_use if base object can't be determined.
21612 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21614 PR tree-optimization/79345
21615 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21616 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21617 (get_pattern_stats): Initialize it.
21618 * genemit.c (gen_expand): Verify match_scratch numbers come after
21619 match_operand/match_dup numbers.
21620 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21621 match_scratch numbers.
21622 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21624 * config/s390/s390.md (trunctdsd2): Likewise.
21626 2017-03-02 Richard Biener <rguenther@suse.de>
21628 * wide-int.h (wide_int_storage::operator=): Implement in terms
21631 2017-03-02 Richard Biener <rguenther@suse.de>
21633 PR tree-optimization/79777
21634 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21635 the to insert expression to sth existing.
21637 2017-03-01 Martin Sebor <msebor@redhat.com>
21639 PR middle-end/79692
21640 * gimple-ssa-sprintf.c
21641 (directive::known_width_and_precision): New function.
21642 (format_integer): Use it.
21643 (get_mpfr_format_length): Consider the full range of precision
21644 when computing %g output with the # flag. Set the likely byte
21645 count to 3 rather than 1 when precision is indeterminate.
21646 (format_floating): Correct the lower bound of precision.
21648 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21650 * doc/invoke.texi: Document default code model for 64-bit Linux.
21652 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21655 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21656 udiv rather than div since input pattern is unsigned.
21658 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21660 * config/i386/i386.c (print_reg): Warn for values of
21661 unsupported size in integer register.
21663 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21666 * config/rs6000/predicates.md (current_file_function_operand): Do
21667 not allow self calls to be local if the function is replaceable.
21669 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21672 * config/rs6000/altivec.h (vec_ctz and others): Change the
21673 preprocessor macro that controls conditional compilation from
21674 _ARCH_PWR9 to __POWER9_VECTOR__.
21675 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21676 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21677 control (instead of _ARCH_PWR9 control) so that template
21678 definition uses power9-specific function.
21679 (vec_any_eq): Likewise.
21680 (vec_all_ne): Change macro definition to use a power9-specific
21681 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21682 _ARCH_PWR9 control).
21683 (vec_any_eq) Likewise.
21684 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21685 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21686 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21687 support for xvcmpnedp instruction.
21688 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21689 macro expansion so that Power9 implementation of vec_all_ne does
21690 not use the AltiVec predicate framework.
21691 (VCMPNEH_P): Likewise.
21692 (VCMPNEW_P): Likewise.
21693 (VCMPNED_P): Likewise.
21694 (VCMPNEFP_P): Likewise.
21695 (VCMPNEDP_P): Likewise.
21696 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21697 implementation of vec_any_eq to not use AltiVec predicate
21699 (VCMPAEH_P): Likewise.
21700 (VCMPAEW_P): Likewise.
21701 (VCMPAED_P): Likewise.
21702 (VCMPAEFP_P): Likewise.
21703 (VCMPAEDP_P): Likewise.
21704 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21705 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21706 not use the AltiVec predicate framework.
21707 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21708 of vec_any_eq to not use AltiVec predicate framework.
21709 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21710 support for predefined __POWER9_VECTOR__ macro to indicate that
21711 Power9 instruction selection is enabled.
21712 (altivec_overloaded_builtins): Remove extraneous
21713 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21714 function argument types RS6000_BTI_bool_V16QI and
21715 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21716 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21717 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21718 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21719 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21720 Power9 for implementations of vec_cmpne. Change the signature for
21721 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21722 (representing vec_all_ne) to remove the previously described first
21723 argument of type RS6000_BTI_INTSI, as this was an artifact of
21724 reliance on the AltiVec predicate framework, which is no longer
21725 used in the implementation of these functions. Add
21726 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21727 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21728 since, unlike the AltiVec predicate framework implementation, we
21729 do not share function descriptors between vec_alle and vec_anyeq.
21730 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21731 set of modes that receive special treatment even when
21732 TARGET_P9_VECTOR is true. The special treatment emits code that
21733 does not depend on Power9 instructions.
21734 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21735 define_expand to not rely on AltiVec predicate framework.
21736 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21738 (vector_ne_v2di_p): Change this define_expand to not rely on
21739 AltiVec predicate framework.
21740 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21742 (vector_ne_<mode>_p): Change this define_expand to not rely on
21743 AltiVec predicate framework.
21744 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21746 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21747 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21748 define_insn pattern.
21749 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21750 define_insn pattern because the xvcmpne<VSs>. instruction is not
21752 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21753 instruction is not supported.
21755 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21757 * config/nvptx/nvptx.c: Include intl.h.
21759 2017-03-01 Martin Jambor <mjambor@suse.cz>
21762 * ipa-prop.h (ipa_bits): Removed field known.
21763 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21764 to pointers. Adjusted their comments to warn about their sharing.
21765 (ipcp_transformation_summary): Change bits to a vector of pointers.
21766 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21767 (ipa_get_ipa_bits_for_value): Declare.
21768 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21769 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21770 (ipa_bits_hash_table): Likewise.
21771 (ipa_vr_ggc_hash_traits): Likewise.
21772 (ipa_vr_hash_table): Likewise.
21773 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21774 being pointers and vr_known being removed.
21775 (ipa_set_jf_unknown): Likewise.
21776 (ipa_get_ipa_bits_for_value): New function.
21777 (ipa_set_jfunc_bits): Likewise.
21778 (ipa_get_value_range): New overloaded functions.
21779 (ipa_set_jfunc_vr): Likewise.
21780 (ipa_compute_jump_functions_for_edge): Use the above functions to
21781 construct bits and vr parts of jump functions.
21782 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21783 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21785 (ipcp_grow_transformations_if_necessary): Also allocate
21786 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21788 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21789 them. Fix too long lines.
21790 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21791 vr_known being removed.
21792 (ipa_read_jump_function): Use new setter functions to construct bits
21793 and vr parts of jump functions or set them to NULL.
21794 (write_ipcp_transformation_info): Adjust for bits being pointers.
21795 (read_ipcp_transformation_info): Likewise.
21796 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21798 Include gt-ipa-prop.h.
21799 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21801 (ipcp_store_bits_results): Likewise.
21802 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21803 Do not write to existing jump functions but use a temporary instead.
21805 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21808 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21809 attempt to use its first operand as BIT_FIELD_REF base.
21811 2017-03-01 Richard Biener <rguenther@suse.de>
21813 PR middle-end/79721
21814 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21815 interpolating formula in wrapping arithmetic.
21816 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21818 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21820 PR tree-optimization/79734
21821 * tree-vect-generic.c (expand_vector_condition): Optimize
21822 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21823 Handle VEC_COND_EXPR where comparison has different inner width from
21824 type's inner width.
21826 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21828 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21829 markup, and similar issues. Remove @opindex entries for things
21830 that aren't options. Add missing -mmpy-option entries.
21832 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21834 PR tree-optimization/79737
21835 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21836 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21837 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21838 instead of byte_size. Formatting fix.
21839 (shift_bytes_in_array_right): Formatting fix.
21841 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21844 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21845 condition on optimize for the leaf function test.
21847 2017-02-28 Martin Liska <mliska@suse.cz>
21850 * read-rtl-function.c (function_reader::handle_unknown_directive):
21851 Bail out when one uses -flto.
21853 2017-02-28 Martin Liska <mliska@suse.cz>
21855 * common.opt: Replace space with tabular for options of <number>
21857 * config/i386/i386.opt: Show <number> value for
21858 -mlarge-data-threshold.
21859 * opts.c (print_filtered_help): Do not display number in hexadecimal
21862 2017-02-28 Martin Liska <mliska@suse.cz>
21864 * common.opt: Fix --help=option -Q for options which are of
21867 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21869 * config/i386/i386.c (print_reg): Error out for values
21870 of 8-bit size in invalid integer register.
21872 2017-02-28 Martin Sebor <msebor@redhat.com>
21874 PR tree-optimization/79691
21875 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21877 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21880 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21881 gcc_unreachable with output_operand_lossage.
21883 2017-02-28 Richard Biener <rguenther@suse.de>
21885 PR tree-optimization/79740
21886 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21888 (visit_nary_op): Insert the nary into the hashtable if we
21889 pattern-matched sth.
21890 * tree-ssa-pre.c (eliminate_insert): Robustify.
21892 2017-02-28 Richard Biener <rguenther@suse.de>
21894 PR middle-end/79731
21895 * fold-const.c (decode_field_reference): Reject out-of-bound
21898 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21900 * config/i386/i386.c: Include intl.h.
21901 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21902 instead of just cond ? "..." : "...".
21903 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21904 * coverage.c (read_counts_file): Likewise.
21905 * omp-offload.c: Include intl.h.
21906 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21907 of just cond ? "..." : "...".
21908 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21909 of just cond ? "..." : "...".
21911 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21914 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21916 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21917 'tune for' CPU name.
21918 * config/arm/arm-cpu-data.h: Regenerated.
21920 2017-02-28 Richard Biener <rguenther@suse.de>
21922 PR tree-optimization/79732
21923 * tree-inline.c (expand_call_inline): Do not shadow var.
21925 2017-02-28 Richard Biener <rguenther@suse.de>
21927 PR tree-optimization/79723
21928 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21929 address-space properly.
21931 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21933 * doc/optinfo.texi (Optimization groups): Fix option used for
21935 * doc/invoke.texi (-fopt-info): Document "omp".
21936 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21937 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21938 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21939 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21940 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21942 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21944 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21945 OPTGROUP_OMP with "omp".
21947 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21950 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21951 for arithmetic shift of unsigned V2DI.
21953 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21955 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21956 arc/linux.h headers.
21957 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21958 (LINK_SPEC): Likewise.
21959 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21960 (EXTRA_SPECS): Likewise.
21961 (STARTFILE_SPEC): Likewise.
21962 (ENDFILE_SPEC): Likewise.
21963 (LIB_SPEC): Likewise.
21964 (TARGET_SDATA_DEFAULT): Likewise.
21965 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21966 (MULTILIB_DEFAULTS): Likewise.
21967 (DWARF2_UNWIND_INFO): Likewise.
21968 * config/arc/big.h: New file.
21969 * config/arc/elf.h: Likewise.
21970 * config/arc/linux.h: Likewise.
21971 * config/arc/t-uClibc: Remove.
21973 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21975 PR tree-optimization/77536
21976 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21977 (tree_transform_and_unroll_loop): Use above function to compute the
21978 estimated niter of unrolled loop and use it when scaling profile.
21979 Also use count info rather than frequency if it's non-zero.
21980 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21981 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21982 (vect_transform_loop): Call above function.
21984 2017-02-27 Richard Biener <rguenther@suse.de>
21986 PR tree-optimization/45397
21987 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21988 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21989 (visit_nary_op): Add pattern matching for CSEing sign-changed
21990 or truncated operations with wider ones.
21992 2017-02-27 Richard Biener <rguenther@suse.de>
21994 PR tree-optimization/79690
21995 * tree-vect-stmts.c (vectorizable_store): Use vector type
21996 built from the DR with address-space.
21998 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22000 * doc/invoke.texi (Optimize Options): Refine the description
22001 of asan-use-after-return.
22003 2017-02-25 Alan Modra <amodra@gmail.com>
22005 PR rtl-optimization/79584
22006 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22007 base, not ad->base_term, the reg within base. Remove assertion
22008 that ad->base == ad->base_term. Replace gen_int_mode using
22009 bogus mode with const0_rtx.
22011 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22013 PR middle-end/79396
22014 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22015 FMA_EXPR like tcc_binary or tcc_unary.
22017 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22020 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22022 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22023 (output_loc_operands): Handle DW_OP_call_ref and
22024 DW_OP_GNU_variable_value.
22025 (struct variable_value_struct): New type.
22026 (struct variable_value_hasher): Likewise.
22027 (variable_value_hash): New variable.
22028 (string_types): Remove.
22029 (copy_loc_descr): New function.
22030 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22031 (prepend_loc_descr_to_each): New function.
22032 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22033 instead of add_loc_descr_to_each if the first argument is single
22034 location list and the second has multiple.
22035 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22036 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22037 when looking for variable value which doesn't have other location info.
22038 (loc_list_from_tree): Formatting fix.
22039 (gen_array_type_die): Simplify DW_AT_string_length handling.
22040 (adjust_string_types): Remove.
22041 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22042 string_types. Call resolve_variable_values.
22043 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22044 (resolve_addr_in_expr): Likewise. Add A argument.
22045 (copy_deref_exprloc): Remove deref argument. Adjust for the
22046 original expression being DW_OP_GNU_variable_value with optionally
22047 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22048 optionally after it.
22049 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22050 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22051 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22052 (variable_value_hasher::hash, variable_value_hasher::equal): New
22054 (resolve_variable_value_in_expr, resolve_variable_value,
22055 resolve_variable_values, note_variable_value_in_expr,
22056 note_variable_value): New functions.
22057 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22060 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22063 * opts.h (handle_generated_option): Add GENERATED_P argument.
22064 * opts-common.c (handle_option): Adjust function comment.
22065 (handle_generated_option): Add GENERATED_P argument, pass it to
22067 (control_warning_option): Pass false to handle_generated_option
22069 * opts.c (maybe_default_option): Pass true to handle_generated_option
22071 * optc-gen.awk: Likewise.
22073 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22075 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22076 a REG, look at the REG it is a SUBREG of.
22077 (splitter for cmpeqsi_t): Ditto.
22079 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22081 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22082 the special USEs with the pattern of the insn, not the insn itself.
22084 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22087 * doc/invoke.texi: Document -mload-store-pairs.
22089 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22090 Sandra Loosemore <sandra@codesourcery.com>
22092 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22093 argument isn't a CONST_INT.
22094 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22095 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22096 (nios2_validate_compare): Bypass alternate compare logic if *op2
22097 is not a CONST_INT.
22098 (ldstwm_operation_p): Return false if first_base is not a REG or
22099 if first_offset is not a CONST_INT.
22101 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22103 * config/cris/cris.md: Use correct operand in a define_peephole2.
22105 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22107 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22109 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22111 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22112 this_insn if it is an INSN or JUMP_INSN.
22113 (force_offsettable): Look at base, not at addr.
22114 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22115 on things that aren't necessarily CONST_INTs.
22117 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22119 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22120 -mfpmath=sse is the default also for x86-32 targets with SSE2
22121 instruction set when @option{-ffast-math} is enabled
22123 2017-02-24 Jeff Law <law@redhat.com>
22125 PR rtl-optimizatoin/79286
22126 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22129 2017-02-24 Richard Biener <rguenther@suse.de>
22131 PR tree-optimization/79389
22132 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22135 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22137 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22138 function comment to reflect reality.
22139 (loop_exits_before_overflow): Fix typo in function description.
22141 2017-02-24 Richard Biener <rguenther@suse.de>
22143 PR tree-optimization/79389
22144 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22145 properly that a threading opportunity exists. Detect conditional
22146 copy/constant propagation opportunities.
22148 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22150 * config/visium/visium.md (type): Add trap.
22151 (b): New mode attribute.
22152 (*btst): Rename into...
22153 (*btst<mode>): ...this and adjust.
22154 (*cbranchsi4_btst_insn): Rename into...
22155 (*cbranch<mode>4_btst_insn): ...this and adjust.
22156 (trap): New define_insn.
22158 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22160 PR tree-optimization/79389
22161 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22162 (noce_reversed_cond_code): New function.
22163 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22164 reversed_comparison_code. Formatting fix.
22165 (noce_try_store_flag): Test rev_cond != NULL in addition to
22166 reversed_comparison_code.
22167 (noce_try_store_flag_constants): Likewise.
22168 (noce_try_store_flag_mask): Likewise.
22169 (noce_try_addcc): Use rev_cond if non-NULL instead of
22170 reversed_comparison_code.
22171 (noce_try_cmove_arith): Likewise. Formatting fixes.
22172 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22173 (noce_find_if_block): Initialize rev_cond.
22174 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22175 instead of false as last argument never attempt to reverse it
22178 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22180 PR tree-optimization/79663
22181 * tree-predcom.c (combine_chains): Process refs in reverse order
22182 only for ZERO length chains, and add explaining comment.
22184 2017-02-23 Jeff Law <law@redhat.com>
22186 PR tree-optimization/79578
22187 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22188 in call to operand_equal_p.
22190 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22193 * config/i386/cpuid.h: Fix another undefined behavior.
22195 2017-02-23 Richard Biener <rguenther@suse.de>
22197 PR tree-optimization/79683
22198 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22199 vector types for data-refs.
22201 2017-02-23 Martin Liska <mliska@suse.cz>
22203 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22205 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22207 PR middle-end/79665
22208 * internal-fn.c (get_range_pos_neg): Moved to ...
22209 * tree.c (get_range_pos_neg): ... here. No longer static.
22210 * tree.h (get_range_pos_neg): New prototype.
22211 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22212 are known to be in between 0 and signed maximum inclusive, try to
22213 expand both unsigned and signed divmod and use the cheaper one from
22216 2017-02-22 Jeff Law <law@redhat.com>
22218 PR tree-optimization/79578
22219 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22220 to compare base operands.
22222 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22225 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22226 gpc_reg_operand instead of fpr_reg_operand.
22228 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22230 * config/mips/mips.c (mips_return_in_memory): Force FP
22231 vector types to be returned in memory for o32 ABI.
22233 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22235 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22236 instead of DW_TAG_member for static data member declarations and don't
22237 set no_linkage_name for static inline data members.
22238 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22241 2017-02-22 Martin Liska <mliska@suse.cz>
22243 * doc/invoke.texi: Replace inequality signs with square brackets
22246 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22248 PR tree-optimization/68644
22249 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22251 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22254 * lra-constraints.c (simplify_operand_subreg): Handle
22255 WORD_REGISTER_OPERATIONS targets.
22257 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22260 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22261 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22262 elimination by swapping fld*.
22264 2017-02-22 Richard Biener <rguenther@suse.de>
22266 PR tree-optimization/79673
22267 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22268 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22269 irrelevant address-space qualifiers and avoiding a
22270 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22272 2017-02-22 Richard Biener <rguenther@suse.de>
22274 PR tree-optimization/79666
22275 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22276 to not symbolically negate if that may introduce undefined
22279 2017-02-22 Martin Liska <mliska@suse.cz>
22282 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22283 * data-streamer-out.c (streamer_write_gcov_count_stream):
22285 * value-prof.c (stream_out_histogram_value): Make assert more
22286 precise based on type of counter.
22288 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22291 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22292 Use nonimmediate_operand instead of memory_operand for operand 1.
22293 (float-extend standard_x87sse_constant_load splitter): Ditto.
22295 2017-02-21 Jeff Law <law@redhat.com>
22297 PR tree-optimization/79621
22298 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22299 blocks with edges to themselves.
22301 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22304 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22305 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22306 Use gimple_call_builtin_p.
22309 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22310 on temporarily removed DEBUG_INSNs.
22312 PR tree-optimization/79649
22313 * tree-loop-distribution.c (classify_partition): Give up on
22314 non-generic address space loads/stores.
22316 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22318 * doc/loop.texi (Loop manipulation): Remove nonexistent
22319 tree_ssa_loop_version from the documentation.
22320 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22322 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22325 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22326 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22327 * config/rs6000/rs6000.c: Include except.h.
22328 (rs6000_expand_split_stack_prologue): Call
22329 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22331 2017-02-21 Martin Jambor <mjambor@suse.cz>
22334 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22335 have been analyzed.
22337 2017-02-21 Martin Jambor <mjambor@suse.cz>
22339 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22340 for backward compatibility only.
22341 * doc/invoke.texi (Option Summary): Remove all references to
22342 -fipa-cp-alignment.
22344 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22348 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22350 * lra-constraints.c (curr_insn_transform): Handle
22351 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22353 2017-02-21 Martin Liska <mliska@suse.cz>
22355 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22357 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22359 PR translation/79638
22360 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22362 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22365 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22366 (arm_function_ok_for_sibcall): Return false for an indirect call by
22367 descriptor if all the argument registers are used.
22368 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22369 alignment of the function.
22371 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22373 PR tree-optimization/61441
22374 * simplify-rtx.c (simplify_const_unary_operation): For
22375 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22376 the sNaN unmodified.
22378 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22380 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22381 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22382 instead of SYSTEM_HEADER_DIR.
22384 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22385 Martin Liška <mliska@suse.cz>
22387 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22388 Fix typos and grammar, use active voice, and clarify.
22390 2017-02-20 Marek Polacek <polacek@redhat.com>
22392 PR middle-end/79537
22393 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22396 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22398 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22401 * config/i386/i386.c (ix86_expand_builtin): Handle
22402 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22403 ix86_builtins_isa[fcode].isa as a requirement of those
22404 flags and any other flag in the bitmask.
22405 (ix86_init_mmx_sse_builtins): Use 0 instead of
22406 ~OPTION_MASK_ISA_64BIT as mask.
22407 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22408 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22409 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22410 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22412 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22415 * lra-constraints.c (split_reg): Check requested split mode
22416 is supported by the register.
22418 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22420 * lra-constraints.c (simplify_operand_subreg): Remove early
22423 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22426 * lra-constraints.c (curr_insn_transform): Tighten condition
22427 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22429 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22432 * lra-constraints.c (curr_insn_transform): Handle
22433 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22435 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22438 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22440 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22442 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22445 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22448 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22450 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22453 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22455 * final.c (last_columnnum, override_columnnum): New variables.
22456 (final_start_function): Set last_columnnum, pass it to begin_prologue
22457 hook and pass 0 to dwarf2out_begin_prologue.
22458 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22459 to source_line debug hook.
22460 (notice_source_line): Compute last_columnnum and for debug_column_info
22461 return true on column changes.
22462 * debug.h (struct gcc_debug_hooks): Add column argument to
22463 source_line and begin_prologue hooks.
22464 (debug_nothing_int_charstar_int_bool): Remove prototype.
22465 (debug_nothing_int_int_charstar,
22466 debug_nothing_int_int_charstar_int_bool): New prototypes.
22467 (dwarf2out_begin_prologue): Add column argument.
22468 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22469 begin_prologue hooks.
22470 (debug_nothing_int_charstar_int_bool): Remove.
22471 (debug_nothing_int_int_charstar,
22472 debug_nothing_int_int_charstar_int_bool): New functions.
22473 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22474 through to dwarf2out_source_line.
22475 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22476 (dwarf2out_source_line): Add column argument, emit it if requested.
22477 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22479 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22480 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22481 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22482 through to dwarf2out_begin_prologue.
22483 (vmsdbgout_source_line): Add column argument, pass it through to
22484 dwarf2out_source_line.
22485 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22486 dbxout_source_line caller.
22487 (dbxout_source_line): Add column argument.
22489 * common.opt (gno-column-info, gcolumn-info): New options.
22490 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22491 (check_die): Also test for multiple DW_AT_decl_column attributes.
22492 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22493 DW_AT_decl_column if requested.
22494 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22496 (gen_variable_die): Likewise.
22497 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22498 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22501 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22502 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22503 (ix86_handle_option): Handle OPT_m3dnowa.
22504 * doc/invoke.texi (-m3dnowa): Document.
22505 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22506 -m3dnowa instead of -m3dnow -march=athlon.
22509 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22510 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22512 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22515 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22516 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22517 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22518 generator for vsx_xxpermdi_<mode>_be.
22519 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22520 force big-endian semantics.
22521 (vsx_xxpermdi_<mode>_be): New define_expand with same
22522 implementation as previous version of vsx_xxpermdi_<mode>.
22524 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22526 PR tree-optimization/79327
22527 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22528 variable, its initialization and use.
22530 2017-02-17 Julia Koval <julia.koval@intel.com>
22532 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22533 (OPTION_MASK_ISA_PKU_UNSET): New.
22534 (ix86_handle_option): Handle -mrdpid.
22535 * config/i386/cpuid.h (bit_RDPID): New.
22536 * config/i386/driver-i386.c (host_detect_local_cpu):
22537 Detect RDPID feature.
22538 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22539 * config/i386/i386-c.c (ix86_target_macros_internal):
22541 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22542 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22543 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22544 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22545 * config/i386/i386.md (define_insn "rdpid"): New.
22546 * config/i386/i386.opt Add -mrdpid.
22547 * config/i386/immintrin.h (_rdpid_u32): New.
22549 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22551 PR rtl-optimization/79541
22552 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22553 instead of transforming it into USE.
22555 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22557 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22558 If HONOR_SNANS (SFmode) force the input to a register.
22559 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22560 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22561 an frsp or similar insn.
22563 2017-02-17 Martin Liska <mliska@suse.cz>
22565 PR rtl-optimization/79577
22566 * params.def (selsched-max-sched-times): Increase minimum to 1.
22568 2017-02-17 Martin Liska <mliska@suse.cz>
22570 PR rtl-optimization/79574
22571 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22573 2017-02-17 Martin Liska <mliska@suse.cz>
22575 PR tree-optimization/79529
22576 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22577 ssa_defined_default_def_p to handle cases which are implicitly
22579 * tree-ssa.c (ssa_defined_default_def_p): New function.
22580 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22581 which are implicitly defined.
22582 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22584 2017-02-17 Richard Biener <rguenther@suse.de>
22586 PR middle-end/79576
22587 * params.def (max-ssa-name-query-depth): Limit to 10.
22589 2017-02-17 Richard Biener <rguenther@suse.de>
22591 PR tree-optimization/79552
22592 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22595 2017-02-17 Richard Biener <rguenther@suse.de>
22598 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22600 2017-02-17 Marek Polacek <polacek@redhat.com>
22602 PR middle-end/79536
22603 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22604 (fold_negate_expr): New wrapper.
22606 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22608 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22609 Correct terminology and de-emphasize pre-standard behavior.
22611 2017-02-16 Alan Modra <amodra@gmail.com>
22613 PR rtl-optimization/79286
22614 * ira.c (def_dominates_uses): New function.
22615 (update_equiv_regs): Don't create an equivalence for insns that
22616 may trap where the register def does not dominate the use.
22618 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22620 PR rtl-optimization/78127
22621 * lra.c (lra): Call lra_eliminate before finish the loop after
22624 2017-02-16 Richard Biener <rguenther@suse.de>
22626 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22628 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22629 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22630 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22631 (isl_val_int_from_wi): New function.
22632 (extract_affine_gmp): Rename to ...
22633 (extract_affine_wi): ... this, take a widest_int.
22634 (extract_affine_int): Just wrap extract_affine_wi.
22635 (add_param_constraints): Use isl_val_int_from_wi.
22636 (add_loop_constraints): Likewise, and extract_affine_wi.
22638 2017-02-15 Jeff Law <law@redhat.com>
22640 PR middle-end/79521
22641 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22642 ira_init_register_move_cost_if_necessary.
22644 2017-02-15 Martin Sebor <msebor@redhat.com>
22646 PR middle-end/32003
22647 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22648 removed in a prior commit.
22650 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22652 PR tree-optimization/79347
22653 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22654 counters during peeling.
22656 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22658 * Makefile.in (site.exp): Remove "set ISLVER".
22660 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22663 * real.c (real_from_integer): Call real_convert even for decimal.
22665 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22668 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22670 2017-02-14 Andrew Pinski <apinski@cavium.com>
22672 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22673 cores and change the partno/implementer to be correct.
22674 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22675 the 'B" as the implementer.
22676 * config/aarch64/aarch64-tune.md: Regenerate.
22678 2017-02-14 Carl Love <cel@us.ibm.com>
22680 * config/rs6000/rs6000.c: Add case statement entry to make the
22681 xvcvuxdsp built-in argument unsigned.
22682 * config/rs6000/vsx.md: Fix the source and return operand types so they
22683 match the instruction definitions from the ISA document. Fix typo
22684 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22687 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22690 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22691 member early_clobber_alts.
22692 * lra-lives.c (reg_early_clobber_p): New.
22693 (process_bb_lives): Use it.
22694 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22695 (debug_operand_data): Initialize early_clobber_alts.
22696 (setup_operand_alternative): Set up early_clobber_alts.
22697 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22698 alternatives to new_insn_reg.
22699 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22701 (lra_update_insn_regno_info): Pass the new arg.
22703 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22705 PR middle-end/79505
22706 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22707 (new_oacc_loop_raw): Don't clear already cleared fields.
22710 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22711 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22712 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22714 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22717 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22719 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22722 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22723 the extra instruction to the right place to store 128-bit constant
22726 2017-02-14 Martin Sebor <msebor@redhat.com>
22728 PR middle-end/79448
22729 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22730 warning for strings of unknown length.
22732 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22734 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22736 2017-02-14 Jeff Law <law@redhat.com>
22739 * ira-costs.c (scan_one_insn): Initialize register move costs
22740 for pseudos seen in USE/CLOBBER insns.
22742 PR tree-optimization/79095
22743 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22744 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22745 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22746 if the operands are known to be not equal, then the resulting range
22748 (intersect_ranges): If the new range is ~[0,0] and the old range is
22749 wide, then prefer ~[0,0].
22750 * tree-vrp.c (overflow_comparison_p_1): New function.
22751 (overflow_comparison_p): New function.
22752 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22753 if NAME is used in an overflow test.
22754 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22755 overflow check that can be expressed as an equality test, then adjust
22756 ops to be that equality test.
22758 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22760 * config/s390/s390-builtin-types.def: Remove flags argument.
22761 * config/s390/s390.c (s390_init_builtins): Likewise.
22763 2017-02-14 Martin Liska <mliska@suse.cz>
22765 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22766 vector. Fix trailing white spaces.
22768 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22770 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22773 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22775 PR rtl-optimization/68664
22776 * config/arm/arm.c (arm_sched_can_speculate_insn):
22777 New function. Declare prototype.
22778 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22780 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22782 PR rtl-optimization/68664
22783 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22785 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22787 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22789 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22790 max skip bytes for function, loop and jump.
22792 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22794 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22795 ABS_EXPR for gimple dump.
22797 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22800 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22802 PR tree-optimization/79408
22803 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22804 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22805 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22806 also if rhs1 is INTEGER_CST.
22808 2017-02-14 Richard Biener <rguenther@suse.de>
22810 PR middle-end/79432
22811 * tree-into-ssa.c (insert_phi_nodes): When the function can
22812 have abnormal edges rewrite SSA names with broken use-def
22813 dominance out of SSA and register them for PHI insertion.
22815 2017-02-13 Martin Sebor <msebor@redhat.com>
22817 PR middle-end/79496
22818 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22819 clearing info.nowrite flag when snprintf size argument is a range.
22821 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22823 * cprop.c (cprop_jump): Add missing space in string literal.
22824 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22825 (get_constraint_for_component_ref): Likewise.
22826 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22827 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22828 * lra-constraints.c (process_alt_operands): Likewise.
22829 * ipa-inline.c (inline_small_functions): Likewise.
22830 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22831 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22832 * trans-mem.c (diagnose_tm_1_op): Likewise.
22833 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22834 (grid_parallel_clauses_gridifiable): Likewise.
22836 * config/nvptx/mkoffload.c (process): Add space in between
22839 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22840 "MOD4_SSE_REGS" and "ALL_REGS".
22842 * spellcheck.c (test_data): Add , in between "foo" and "food".
22844 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22847 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22848 boundary crossing check and subsequent code generation agree.
22850 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22852 * config/aarch64/aarch64.c (has_memory_op): Delete.
22853 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22856 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22858 PR rtl-optimization/79388
22859 PR rtl-optimization/79450
22860 * combine.c (distribute_notes): When removing TEM_INSN for which
22861 corresponding dest has last value recorded, invalidate that last
22864 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22866 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22867 of explicit '@'. Add missing assembly comment marker on branch costs
22870 2017-02-13 Nathan Sidwell <nathan@acm.org>
22872 * gengtype-lex.l (<in_struct>): Add '/'.
22874 2017-02-13 Martin Liska <mliska@suse.cz>
22877 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22879 2017-02-13 Richard Biener <rguenther@suse.de>
22881 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22883 * configure: Re-generate.
22884 * config.in: Likewise.
22885 * graphite-dependences.c: Simplify as if
22886 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22887 * graphite-isl-ast-to-gimple.c: Likewise.
22888 * graphite-optimize-isl.c: Likewise.
22889 * graphite-poly.c: Likewise.
22890 * graphite-sese-to-poly.c: Likewise.
22891 * graphite.h: Likewise.
22892 * toplev.c: Include isl/version.h and use isl_version () for
22893 printing the ISL version.
22894 * doc/install.texi: Update ISL requirement.
22896 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22898 * doc/standards.texi (Standards): Update reference to
22901 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22903 * doc/extend.texi (Named Address Spaces): sourceware.org now
22905 * doc/install.texi (Binaries): Ditto.
22908 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22910 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22911 standard terminology "stringize"/"stringizing" throughout.
22912 * doc/cppinternals.texi: Likewise.
22914 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22916 * doc/extend.texi: Fix some spelling mistakes and typos.
22917 * doc/invoke.texi: Likewise.
22919 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22922 * params.def (inline-min-speedup) Change from 10 to 8.
22924 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22926 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22929 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22932 * ipa-inline-analysis.c (get_minimal_bb): New function.
22933 (record_modified): Use it.
22934 (remap_edge_change_prob): Handle also ancestor functions.
22936 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22938 * doc/contrib.texi (Contributors): Remove broken link into
22939 the Mauve CVS repository.
22941 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22943 PR middle-end/79454
22944 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22945 result computation whenever lhs doesn't have vector mode, not
22946 just when it has BLKmode.
22948 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22950 * doc/makefile.texi (profiledbootstrap): Refer to the
22951 installation instructions only in textual form.
22953 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22956 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22958 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22960 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22961 (Specific): Update mingw-w64 reference.
22963 (Specific): Remove broken link to Renesas RX processor.
22965 2017-02-10 Richard Biener <rguenther@suse.de>
22967 * toplev.c (process_options): Do not mention obsolete graphite
22968 options when printing sorry message about missing graphite support.
22969 Mention -floop-nest-optimize.
22971 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22973 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22974 (vtst_p16): Likewise.
22975 (vtstq_p8): Likewise.
22976 (vtstq_p16): Likewise.
22978 (vtstq_p64): Likewise.
22979 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22980 (vset_lane_p64): New.
22981 (vsetq_lane_p64): New.
22983 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22985 PR tree-optimization/79411
22986 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22987 stmt operands are SSA_NAMEs used in abnormal phis.
22988 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22991 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22994 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22997 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22999 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23001 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23003 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23007 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23008 not arbitrary TREE_CONSTANT.
23011 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23012 "omp declare target link" attribute unless is_global_var.
23013 * omp-offload.c (find_link_var_op): Likewise.
23015 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23016 Chung-Lin Tang <cltang@codesourcery.com>
23018 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23020 (gimplify_adjust_omp_clauses): Don't delete TILE.
23021 (gimplify_omp_for): Deal with TILE.
23022 * internal-fn.c (expand_GOACC_TILE): New function.
23023 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23025 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23026 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23028 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23029 avoid DIV for outermost collapse var.
23030 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23031 Remove out of date comments, fix whitespace.
23032 * omp-general.c (omp_extract_for_data): Deal with tiling.
23033 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23034 adjust OLF_DIM_BASE value.
23035 (struct omp_for_data): Add tiling field.
23036 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23037 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23038 for auto loops. Remove default auto determining, moved to
23039 oacc_loop_fixed_partitions.
23040 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23041 stmts, add e_mask field.
23042 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23043 (oacc_thread_numbers): Use oacc_dim_call.
23044 (oacc_xform_tile): New.
23045 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23046 (finish_oacc_loop): Adjust for ifns vector.
23047 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23048 add case for GOACC_TILE fns.
23049 (oacc_loop_xform_loop): Delete.
23050 (oacc_loop_process): Iterate over call list directly, and add
23051 handling for GOACC_TILE fns.
23052 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23054 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23055 vector partitioning to outer loops. Assign 2 partitions to loops
23056 when available. Add TILE handling.
23057 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23058 (execite_oacc_device_lower): Process GOACC_TILE fns,
23059 ignore unknown specs.
23060 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23061 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23062 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23064 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23066 * configure.ac (ACX_BUGURL): Update.
23067 * configure: Regenerate.
23069 2017-02-09 Richard Biener <rguenther@suse.de>
23071 PR tree-optimization/69823
23072 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23073 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23075 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23077 * config/arc/arc-c.def: Add __NPS400__ definition.
23078 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23079 (TARGET_NPS400): Define.
23081 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23083 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23085 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23086 pointer, arch_info.
23087 (arc_cpu_types): Fill the arch_info field with a pointer into the
23088 arc_arch_types table.
23089 (arc_selected_cpu): Declare.
23090 * config/arc/arc.c (arc_selected_cpu): Make global.
23091 (arc_selected_arch): Delete.
23092 (arc_base_cpu): Delete.
23093 (arc_override_options): Remove references to deleted variables,
23094 update access to arch information.
23095 (ARC_OPT): Update access to arch information.
23096 (ARC_OPTX): Likewise.
23097 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23098 (TARGET_ARC600): Update access to arch information.
23099 (TARGET_ARC601): Likewise.
23100 (TARGET_ARC700): Likewise.
23101 (TARGET_EM): Likewise.
23102 (TARGET_HS): Likewise.
23103 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23106 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23109 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23110 condition/operands for integer GE/LE/GEU/LEU operations.
23112 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23114 PR translation/79397
23115 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23118 2017-02-08 Martin Jambor <mjambor@suse.cz>
23121 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23122 whether allocation happened.
23123 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23124 nothing was allocated.
23126 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23128 PR tree-optimization/79408
23129 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23130 constant, but SSA_NAME with a known integer range, use the minimum
23131 of that range instead of op1 to determine if modulo can be replaced
23132 with its first operand.
23134 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23136 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23138 2017-02-08 Richard Biener <rguenther@suse.de>
23140 PR tree-optimization/71824
23141 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23142 Check all loops contained in the merged region.
23144 2017-02-07 Andrew Pinski <apinski@cavium.com>
23146 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23148 2017-02-07 Andrew Pinski <apinski@cavium.com>
23150 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23151 (thunderxt88): Likewise.
23152 (thunderxt81): Disable LSE and change v8.1 to v8.
23153 (thunderxt83): Likewise.
23155 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23156 Richard Biener <rguenther@suse.de>
23158 PR middle-end/79399
23159 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23160 type from int to size_t.
23161 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23163 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23165 PR rtl-optimization/79386
23166 * cprop.c (bypass_conditional_jumps): Initialize
23167 bypass_last_basic_block already before splitting bbs after
23168 unconditional traps...
23169 (bypass_conditional_jumps): ... rather than here.
23172 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23173 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23174 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23175 fix -masm=intel patterns.
23177 2017-02-07 Richard Biener <rguenther@suse.de>
23179 PR tree-optimization/79256
23180 PR middle-end/79278
23181 * builtins.c (get_object_alignment_2): Use min_align_of_type
23182 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23183 and ADJUST_FIELD_ALIGN.
23185 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23187 * doc/tm.texi: Regenerate.
23188 * stor-layout.c (layout_decl): Adjust.
23189 (update_alignment_for_field): Likewise.
23190 (place_field): Likewise.
23191 (min_align_of_type): Likewise.
23192 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23193 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23194 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23195 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23196 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23197 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23198 * config/i386/i386.c (x86_field_alignment): Likewise.
23199 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23200 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23201 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23202 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23203 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23204 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23208 2017-01-30 Richard Biener <rguenther@suse.de>
23210 PR tree-optimization/79256
23211 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23212 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23215 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23217 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23218 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23219 builtins to SImode and emit a zero-extend, if necessary.
23221 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23223 * docs/invoke.texi (RISC-V Options): Alphabetize.
23225 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23227 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23230 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23232 * config/riscv/riscv.c: New file.
23233 * common/config/riscv/riscv-common.c: Likewise.
23234 * config.gcc: Likewise.
23235 * config/riscv/constraints.md: Likewise.
23236 * config/riscv/elf.h: Likewise.
23237 * config/riscv/generic.md: Likewise.
23238 * config/riscv/linux.h: Likewise.
23239 * config/riscv/multilib-generator: Likewise.
23240 * config/riscv/peephole.md: Likewise.
23241 * config/riscv/pic.md: Likewise.
23242 * config/riscv/predicates.md: Likewise.
23243 * config/riscv/riscv-builtins.c: Likewise.
23244 * config/riscv/riscv-c.c: Likewise.
23245 * config/riscv/riscv-ftypes.def: Likewise.
23246 * config/riscv/riscv-modes.def: Likewise.
23247 * config/riscv/riscv-opts.h: Likewise.
23248 * config/riscv/riscv-protos.h: Likewise.
23249 * config/riscv/riscv.h: Likewise.
23250 * config/riscv/riscv.md: Likewise.
23251 * config/riscv/riscv.opt: Likewise.
23252 * config/riscv/sync.md: Likewise.
23253 * config/riscv/t-elf-multilib: Likewise.
23254 * config/riscv/t-linux: Likewise.
23255 * config/riscv/t-linux-multilib: Likewise.
23256 * config/riscv/t-riscv: Likewise.
23257 * configure.ac: Likewise.
23258 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23259 Waterman as RISC-V maintainers.
23260 * doc/install.texi: Add RISC-V entries.
23261 * doc/invoke.texi: Add RISC-V options section.
23262 * doc/md.texi: Add RISC-V constraints section.
23263 * configure: Regenerated.
23265 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23268 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23269 false values to be constant vectors with all 0 or all 1 bits set.
23270 (vcondu<mode><mode>): Likewise.
23271 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23273 (fpmask_comparison_operator): Update comment.
23274 (vecint_comparison_operator): New predicate.
23275 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23276 vector conditionals when the true and false values are constant
23277 vectors with all 0 bits or all 1 bits set.
23279 2017-02-06 Martin Sebor <msebor@redhat.com>
23281 PR tree-optimization/79376
23282 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23284 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23286 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23287 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23288 to simplify split condition.
23290 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23292 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23293 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23296 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23298 PR rtl-optimization/68664
23299 * target.def (can_speculate_insn): New hook.
23300 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23301 * doc/tm.texi: Regenerate.
23302 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23303 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23304 (rs6000_sched_can_speculate_insn): New function.
23306 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23308 PR tree-optimization/79284
23309 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23310 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23311 vectorizable_mask_load_store, vectorizable_operation,
23312 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23313 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23314 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23315 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23316 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23317 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23318 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23319 is_gimple_assign (stmt). Replace another such test with
23320 is_gimple_assign (stmt).
23322 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23325 * config/avr/avr.c (rtl-iter.h): Include it.
23326 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23327 (avr_legitimate_combined_insn): ...and implementation.
23329 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23331 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23332 * config/s390/s390.c (s390_const_operand_ok)
23333 (s390_canonicalize_comparison, s390_extract_part)
23334 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23335 (s390_contiguous_bitmask_p, s390_rtx_costs)
23336 (legitimize_pic_address): Likewise.
23337 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23338 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23339 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23340 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23341 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23343 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23345 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23346 REGNO($0) == REGNO($1).
23348 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23350 * config/s390/linux.h(SIZE_TYPE): Add comment.
23352 2017-02-06 Julian Brown <julian@codesourcery.com>
23353 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23354 Virendra Pathak <virendra.pathak@broadcom.com>
23356 * config/aarch64/aarch64-cores.def: Change the scheduler
23358 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23359 * config/aarch64/thunderx2t99.md: New file.
23361 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23363 * doc/standards.texi (Go Language): Update link to language
23366 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23368 * tree-eh.c (lower_resx): Sanitize profile.
23369 (cleanup_empty_eh_move_lp): Likewise.
23371 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23374 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23376 * cfgloopmanip.h (loop_version): Update prototype.
23377 * modulo-sched.c (sms_schedule): Update call of loop_version.
23378 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23379 * tree-parloops.c (gen_parallel_loop): Likewise.
23380 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23381 * tree-ssa-loop-split.c (split_loop): Likewise.
23382 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23383 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23385 2017-02-05 Martin Liska <mliska@suse.cz>
23388 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23390 (print_operand_address): Initialize a struct to zero.
23392 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23394 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23395 garbage collector only in textual form.
23397 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23399 * doc/extend.texi (x86 specific memory model extensions for
23400 transactional memory): Simplify a phrase.
23402 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23405 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23406 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23407 (atomic_storedi_1): Likewise.
23409 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23411 PR tree-optimization/79338
23412 * tree-parloops.c (gather_scalar_reductions): Don't call
23413 vect_analyze_loop_form for loop->inner before destroying loop's
23416 2017-02-03 Martin Sebor <msebor@redhat.com>
23418 PR tree-optimization/79327
23419 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23420 when precision has resulted in leading zeros.
23421 (format_integer): Adjust the likely counter to assume an unknown
23422 argument that may be zero is non-zero.
23424 2017-02-03 Jason Merrill <jason@redhat.com>
23427 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23428 avoid copying non-taken branch.
23430 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23432 PR tree-optimization/79340
23433 * tree-vect-loop.c (vectorizable_reduction): Release
23434 vec_defs elements after safe_splicing them into other vectors.
23437 PR tree-optimization/79327
23438 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23439 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23441 (format_integer): Use wide_int_to_tree instead of build_int_cst
23442 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23443 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23444 of shortest and longest sequence.
23446 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23448 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23449 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23451 2017-02-03 Walter Lee <walt@tilera.com>
23454 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23455 after initial stackframe link reg save.
23456 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23458 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23461 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23462 wu for stxssp alternative.
23464 2017-02-03 Martin Sebor <msebor@redhat.com>
23466 PR tree-optimization/79352
23467 * gimple-fold.c (get_range_strlen): Add argument.
23468 (get_range_strlen): Change return type to bool.
23469 (get_maxval_strlen): Pass in a dummy argument.
23470 * gimple-fold.h (get_range_strlen): Change return type to bool.
23471 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23472 * tree.h (array_at_struct_end_p): Add argument.
23473 * tree.c (array_at_struct_end_p): Handle it.
23475 2017-02-03 Martin Liska <mliska@suse.cz>
23478 * multiple_target.c (create_dispatcher_calls): Redirect edge
23479 from a caller of a dispatcher.
23480 (expand_target_clones): Make the clones local.
23481 (ipa_target_clone): Do both target clones and resolvers.
23482 (ipa_dispatcher_calls): Remove the pass.
23483 (pass_dispatcher_calls::gate): Likewise.
23484 (make_pass_dispatcher_calls): Likewise.
23485 * passes.def (pass_target_clone): Put as very first IPA early
23488 2017-02-03 Martin Liska <mliska@suse.cz>
23490 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23491 in case of a function with ifunc attribute.
23493 2017-02-03 Martin Liska <mliska@suse.cz>
23495 * cgraph.c (cgraph_node::dump): Dump function version info.
23496 * symtab.c (symtab_node::dump_base): Add missing new line.
23498 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23500 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23501 (ifcombine_ifandif): Use it.
23503 2017-02-03 Martin Liska <mliska@suse.cz>
23505 * doc/invoke.texi: Document default value for
23506 use-after-scope-direct-emission-threshold.
23508 2017-02-03 Martin Liska <mliska@suse.cz>
23510 PR tree-optimization/79339
23511 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23512 (format_floating): Likewise.
23514 2017-02-03 Martin Liska <mliska@suse.cz>
23517 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23519 (ipa_node_params_t::remove): Likewise.
23520 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23521 initialization from removed ipa_node_params_t::insert.
23522 (ipa_node_params::~ipa_node_params): Move from removed
23523 ipa_node_params_t::release.
23524 * symbol-summary.h (symbol_summary::m_released): New member.
23525 Do not release a summary twice. Do not allow to call finalizer
23526 for types of a summary that live in GGC memory.
23528 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23530 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23533 2017-02-02 Martin Sebor <msebor@redhat.com>
23535 PR middle-end/79275
23536 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23537 (format_string): Tighten up the range of output for non-constant
23538 strings and correct the expected range for wide non-constant strings.
23540 2017-02-02 Martin Sebor <msebor@redhat.com>
23542 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23544 PR middle-end/32003
23545 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23547 (-fdump-tree-@var): Add to index and document how to come up
23548 with pass-specific option and dump file names.
23549 (-fdump-passes): Clarify where to look for output.
23551 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23553 PR middle-end/77445
23554 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23555 statistics of the analyzed path; allow threading for speed when
23556 any of BBs along the path are optimized for speed.
23558 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23560 PR middle-end/78468
23561 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23562 settings of the virtual registers.
23565 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23567 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23568 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23571 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23573 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23574 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23576 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23578 * config/s390/s390.md: Add missing comments with the expanded
23580 * config/s390/vector.md: Likewise.
23581 * config/s390/vx-builtins.md: Likewise.
23583 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23586 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23587 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23588 conditions on a single line.
23590 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23592 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23593 __S390_VX__ to __VX__.
23595 2017-02-01 Andrew Pinski <apinski@cavium.com>
23597 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23598 stmt_info to record_stmt_cost.
23599 (vect_get_known_peeling_cost): Pass stmt_info if known to
23601 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23602 cpu_vector_cost field into
23603 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23604 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23605 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23606 splitting of scalar_stmt_cost and vec_stmt_cost.
23607 (thunderx_vector_cost): Likewise.
23608 (cortexa57_vector_cost): LIkewise.
23609 (exynosm1_vector_cost): Likewise.
23610 (xgene1_vector_cost): Likewise.
23611 (thunderx2t99_vector_cost): Improve after the splitting of the two
23613 (aarch64_builtin_vectorization_cost): Update for the splitting of
23614 scalar_stmt_cost and vec_stmt_cost.
23616 2017-02-01 Torvald Riegel <triegel@redhat.com>
23617 Richard Henderson <rth@redhat.com>
23619 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23620 conditional on existance of a fast atomic load.
23621 * optabs-query.c (can_atomic_load_p): New function.
23622 * optabs-query.h (can_atomic_load_p): Declare it.
23623 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23624 no fast atomic load is available for the particular size of access.
23625 (expand_atomic_compare_and_swap): Likewise.
23626 (expand_atomic_load): Likewise.
23627 (expand_atomic_store): Likewise.
23628 (expand_atomic_fetch_op): Likewise.
23629 * testsuite/lib/target-supports.exp
23630 (check_effective_target_sync_int_128): Remove x86 because it provides
23631 no fast atomic load.
23632 (check_effective_target_sync_int_128_runtime): Likewise.
23634 2017-02-01 Richard Biener <rguenther@suse.de>
23636 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23637 (graphite_transform_loops): Provide opt-info for optimized nests.
23638 * tree-parloop.c (parallelize_loops): Provide opt-info for
23639 parallelized loops.
23641 2017-02-01 Richard Biener <rguenther@suse.de>
23643 PR middle-end/79315
23644 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23645 was not set before.
23647 2017-02-01 Richard Biener <rguenther@suse.de>
23649 PR tree-optimization/71824
23650 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23651 Verify the loops are valid in the merged SESE region.
23652 (scop_detection::can_represent_loop_1): Check analyzing the
23653 evolution of the number of iterations in the region succeeds.
23655 2017-01-31 Ian Lance Taylor <iant@golang.org>
23657 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23658 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23660 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23662 PR preprocessor/79210
23663 * input.c (get_substring_ranges_for_loc): Replace line_width
23664 assertion with error-handling.
23666 2017-01-31 Richard Biener <rguenther@suse.de>
23668 PR tree-optimization/77318
23669 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23670 (create_pw_aff_from_tree): Take loop parameter.
23671 (add_condition_to_pbb): Pass loop of the condition to
23672 create_pw_aff_from_tree.
23674 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23676 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23677 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23679 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23683 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23685 (convert_int_to_float128): Likewise.
23686 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23687 (convert_int_to_float128): Likewise.
23688 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23689 (UNSPEC_IEEE128_CONVERT): Likewise.
23690 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23691 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23692 Use local variables for IBM extended format.
23693 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23694 (fix_trunc<mode>si2_fprs): Likewise.
23695 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23696 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23697 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23698 to know that we can now have integers of all sizes in vector
23700 (fix<uns>_<mode>di2_hw): Likewise.
23701 (float<uns>_<mode>si2_hw): Likewise.
23702 (fix_<mode>si2_hw): Likewise.
23703 (fixuns_<mode>si2_hw): Likewise.
23704 (float<uns>_<mode>di2_hw): Likewise.
23705 (float_<mode>di2_hw): Likewise.
23706 (float_<mode>si2_hw): Likewise.
23707 (floatuns_<mode>di2_hw): Likewise.
23708 (floatuns_<mode>si2_hw): Likewise.
23709 (xscvqp<su>wz_<mode>): Delete, no longer used.
23710 (xscvqp<su>dz_<mode>): Likewise.
23711 (xscv<su>dqp_<mode>): Likewise.
23712 (ieee128_mfvsrd_64bit): Likewise.
23713 (ieee128_mfvsrd_32bit): Likewise.
23714 (ieee128_mfvsrwz): Likewise.
23715 (ieee128_mtvsrw): Likewise.
23716 (ieee128_mtvsrd_64bit): Likewise.
23717 (ieee128_mtvsrd_32bit): Likewise.
23719 2017-01-31 Martin Liska <mliska@suse.cz>
23722 * ipa-prop.c (ipa_free_all_node_params): Call release method
23723 instead of ~sumbol_summary to not to trigger double times
23726 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23728 PR tree-optimization/71691
23729 * bitmap.h (class auto_bitmap): New.
23730 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23731 is_maybe_undefined instead of ssa_undefined_value_p.
23733 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23735 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23736 __S390_ARCH_LEVEL__ to __ARCH__.
23738 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23740 PR tree-optimization/79267
23741 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23742 if should_remove_lhs_p is true.
23744 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23747 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23748 (add_alignment_attribute): New.
23749 (base_type_die): Add alignment attribute.
23750 (subrange_type_die): Likewise.
23751 (modified_type_die): Likewise.
23752 (gen_array_type_die): Likewise.
23753 (gen_descr_array_type_die: Likewise.
23754 (gen_enumeration_type_die): Likewise.
23755 (gen_subprogram_die): Likewise.
23756 (gen_variable_die): Likewise.
23757 (gen_field_die): Likewise.
23758 (gen_ptr_to_mbr_type_die): Likewise.
23759 (gen_struct_or_union_type_die): Likewise.
23760 (gen_subroutine_type_die): Likewise.
23761 (gen_typedef_die): Likewise.
23762 (base_type_cmp): Compare alignment attribute.
23764 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23767 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23768 (setb_unsigned) New pattern for setb with CCUNS.
23769 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23770 subfc./subfe sequence to avoid overflow problems. Generate a
23771 shorter sequence with cmpld/setb for power9.
23772 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23773 for generating subfc. instruction.
23774 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23775 now uses this instruction.
23777 2017-01-30 Ian Lance Taylor <iant@google.com>
23780 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23781 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23783 2017-01-30 Martin Sebor <msebor@redhat.com>
23785 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23786 Move constant to the right of a relational operator.
23787 (get_mpfr_format_length, format_character, format_string): Ditto.
23788 (should_warn_p, maybe_warn): Same.
23790 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23792 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23795 * asan.c (get_translation_unit_decl): Remove function.
23796 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23798 2017-01-30 Martin Liska <mliska@suse.cz>
23800 PR gcov-profile/79259
23801 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23802 -fprofile-generate.
23804 2017-01-30 Martin Liska <mliska@suse.cz>
23807 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23808 Initialize variables with NULL value.
23810 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23813 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23815 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23817 2017-01-30 Richard Biener <rguenther@suse.de>
23819 PR tree-optimization/79276
23820 * tree-vrp.c (process_assert_insertions): Properly adjust common
23821 when removing a duplicate.
23823 * gcc.dg/torture/pr79276.c: New testcase.
23825 2017-01-30 Richard Biener <rguenther@suse.de>
23827 PR tree-optimization/79256
23828 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23829 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23831 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23833 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23836 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23837 ("*r<noxa>sbg_<mode>_sll_bitmask")
23838 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23839 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23840 Use contiguous_bitmask_nowrap_operand.
23842 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23845 * config/rs6000/altivec.h (vec_xl): Revise #define.
23846 (vec_xst): Likewise.
23848 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23850 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23852 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23854 PR rtl-optimization/79194
23855 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23856 traps before call to bypass_conditional_jumps.
23858 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23860 PR tree-optimization/71374
23861 * lra-constraints.c (check_conflict_input_operands): New.
23862 (match_reload): Use it.
23864 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23867 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23868 account to calculate conflict_set.
23870 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23872 PR rtl-optimization/78559
23873 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23874 other_insn in combine.
23876 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23878 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23879 uint16_type_node for BT_UINT16.
23881 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23883 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23884 "RTL Tests" to menu.
23885 (GIMPLE Tests): New node.
23886 (RTL Tests): New node.
23888 2017-01-27 Richard Biener <rguenther@suse.de>
23890 PR tree-optimization/79245
23891 * tree-loop-distribution.c (distribute_loop): Apply cost
23892 modeling also to detected patterns.
23894 2017-01-27 Richard Biener <rguenther@suse.de>
23896 PR tree-optimization/71433
23897 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23898 (compare_assert_loc): New function.
23899 (process_assert_insertions): Sort and optimize assert locations
23900 to remove duplicates and push down identical assertions on
23901 edges to their destination block.
23903 2017-01-27 Richard Biener <rguenther@suse.de>
23905 PR tree-optimization/79244
23906 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23907 out SSA names even if abnormal.
23909 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23911 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23912 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23913 instead of MPFR_RNDN.
23915 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23918 * arm.c (arm_option_override): Don't call build_target_option_node
23919 until after doing all option overrides.
23920 (arm_valid_target_attribute_tree): Likewise.
23922 2017-01-27 Martin Liska <mliska@suse.cz>
23924 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23925 for {cd}tors and C++ {cd}tors.
23927 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23929 * config/s390/s390.md ("*setmem_long_and")
23930 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23932 2017-01-26 Martin Sebor <msebor@redhat.com>
23934 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23937 2017-01-26 Martin Sebor <msebor@redhat.com>
23939 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23940 HAVE_DFmode before using XFmode or DFmode.
23941 (parse_directive): Avoid using the z length modifier to avoid
23942 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23944 PR middle-end/78703
23945 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23946 to accept adjustment as an array.
23947 (get_int_range): New function.
23948 (struct directive): Make width and prec arrays.
23949 (directive::set_width, directive::set_precision): Call get_int_range.
23950 (format_integer, format_floating): Handle width and precision ranges.
23951 (format_string, parse_directive): Same.
23953 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23956 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23957 comdat_type_p set, just clone them, but keep the children in the
23961 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23962 which have direct callers with -fvar-tracking-assignments enabled
23964 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23965 inside of type units.
23967 2017-01-26 Martin Sebor <msebor@redhat.com>
23969 PR middle-end/78703
23970 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23972 (struct format_result): Replace number_chars, number_chars_min,
23973 and number_chars_max with a single member of struct result_range.
23975 (format_result::operator+=): Adjust.
23976 (struct fmtresult): Remove bounded. Handle likely and unlikely
23978 (fmtresult::adjust_for_width_or_precision): New function.
23979 (fmtresult:type_max_digits): New function.
23980 (bytes_remaining): Handle likely and unlikely counters.
23981 (min_bytes_remaining): Remove.
23982 (format_percent): Simplify.
23983 (format_integer, format_floating): Set likely and unlikely counters.
23984 (get_string_length, format_character, format_string): Same.
23985 (format_plain, should_warn_p): New function.
23986 (maybe_warn): Call should_warn_p. Update diagnostic messages
23987 and handle those for all directives, including plain strings.
23988 (format_directive): Handle likely and unlikely counters.
23989 Remove unnecessary quoting from diagnostics. Add an informational
23991 (add_bytes): Remove.
23992 (pass_sprintf_length::compute_format_length): Simplify.
23993 (try_substitute_return_value): Handle likely and unlikely counters.
23995 2017-01-26 Carl Love <cel@us.ibm.com>
23997 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23998 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24000 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24003 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24004 endianess for subregs into account.
24005 * lra-constraints.c (lra_constraints): Do risky transformations
24006 always on the first iteration.
24007 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24008 last_call_used_reg_set.
24009 (process_bb_lives): Define and use last_call_used_reg_set.
24010 * lra.c (lra): Always continue after lra_constraints on the first
24013 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24015 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24017 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24019 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24021 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24022 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24023 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24024 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24025 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24026 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24027 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24028 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24029 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24031 2017-01-26 Marek Polacek <polacek@redhat.com>
24034 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24035 for the third operand.
24037 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24039 PR middle-end/79236
24040 * omp-low.c (struct omp_context): Add simt_stmt field.
24041 (scan_omp_for): Return omp_context *.
24042 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24043 context to the _simt_ SIMD stmt.
24044 (lower_omp_for): For combined SIMD with sibling _simt_
24045 SIMD, make sure to use the same decls in _looptemp_
24046 clauses as in the sibling.
24048 2017-01-26 David Sherwood <david.sherwood@arm.com>
24050 PR middle-end/79212
24051 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24054 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24057 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24058 emit fld b; fld a; if possible.
24060 * brig-builtins.def: Update copyright years.
24061 * config/arm/arm_acle_builtins.def: Update copyright years.
24063 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24066 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24067 constraint instead of o for the stxsd instruction.
24069 2017-01-25 Carl Love <cel@us.ibm.com>
24071 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24072 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24074 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24076 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24078 2017-01-25 Richard Biener <rguenther@suse.de>
24080 PR tree-optimization/69264
24081 * target.def (vector_alignment_reachable): Improve documentation.
24082 * doc/tm.texi: Regenerate.
24083 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24085 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24086 earlier changes with respect to TYPE_USER_ALIGN.
24087 (vector_alignment_reachable_p): Likewise. Improve dumping.
24089 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24092 * config/arm/arm.md (xordi3): Force constant operand into a register
24095 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24097 * doc/invoke.texi (-fstore-merging): Correct default optimization
24098 levels at which it is enabled.
24099 (-O): Move -fstore-merging from list to...
24102 2017-01-25 Richard Biener <rguenther@suse.de>
24105 * omp-expand.c: Include debug.h.
24106 (expand_omp_taskreg): Make sure to generate early debug before
24107 outlining anything from a function.
24108 (expand_omp_target): Likewise.
24109 (grid_expand_target_grid_body): Likewise.
24111 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24114 * asan.c (get_translation_unit_decl): New function.
24115 (asan_add_global): Extract modules file name from globals
24116 TRANSLATION_UNIT_DECL name.
24118 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24121 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24122 for long calls with APCS frame and VFP.
24124 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24126 * cfg.c (original_copy_tables_initialized_p): New function.
24127 * cfg.h (original_copy_tables_initialized_p): New decl.
24128 * cfgrtl.c (relink_block_chain): Guard the call to
24129 free_original_copy_tables with a call to
24130 original_copy_tables_initialized_p.
24131 * cgraph.h (symtab_node::native_rtl_p): New decl.
24132 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24133 (symtab_node::needed_p): Don't assert for early assembly output
24134 for __RTL functions.
24135 (cgraph_node::finalize_function): Set "force_output" for __RTL
24137 (cgraph_node::analyze): Bail out early for __RTL functions.
24138 (analyze_functions): Update assertion to support __RTL functions.
24139 (cgraph_node::expand): Bail out early for __RTL functions.
24140 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24142 * function.h (struct function): Update comment for field
24144 * gimple-expr.c: Include "tree-pass.h".
24145 (gimple_has_body_p): Return false for __RTL functions.
24146 * Makefile.in (OBJS): Add run-rtl-passes.o.
24147 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24149 (gcc::pass_manager::get_clean_slate): New accessor.
24150 * passes.c: Include "insn-addr.h".
24151 (should_skip_pass_p): Add logging. Update logic for running
24152 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24153 property-provider override so it is only done for gimple passes.
24155 (skip_pass): New function.
24156 (execute_one_pass): Call skip_pass when skipping passes.
24157 * read-md.c (md_reader::read_char): Support filtering
24158 the input to a subset of line numbers.
24159 (md_reader::md_reader): Initialize fields
24160 m_first_line and m_last_line.
24161 (md_reader::read_file_fragment): New function.
24162 * read-md.h (md_reader::read_file_fragment): New decl.
24163 (md_reader::m_first_line): New field.
24164 (md_reader::m_last_line): New field.
24165 * read-rtl-function.c (function_reader::create_function): Only
24166 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24167 curr_properties. Set DECL_INITIAL to a dummy block.
24168 (read_rtl_function_body_from_file_range): New function.
24169 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24171 * run-rtl-passes.c: New file.
24172 * run-rtl-passes.h: New file.
24174 2017-01-24 Jeff Law <law@redhat.com>
24176 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24179 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24181 PR tree-optimization/79159
24182 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24183 (record_nonwrapping_iv): Improve boundary using above function if no
24184 value range information.
24186 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24187 Martin Jambor <mjambor@suse.cz>
24189 * brig-builtins.def: New file.
24190 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24191 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24192 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24193 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24194 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24195 * builtin-types.def (BT_INT8): New.
24196 (BT_INT16): Likewise.
24197 (BT_UINT8): Likewise.
24198 (BT_UINT16): Likewise.
24199 (BT_FN_ULONG): Likewise.
24200 (BT_FN_UINT_INT): Likewise.
24201 (BT_FN_UINT_ULONG): Likewise.
24202 (BT_FN_UINT_LONG): Likewise.
24203 (BT_FN_UINT_PTR): Likewise.
24204 (BT_FN_ULONG_PTR): Likewise.
24205 (BT_FN_INT8_FLOAT): Likewise.
24206 (BT_FN_INT16_FLOAT): Likewise.
24207 (BT_FN_UINT32_FLOAT): Likewise.
24208 (BT_FN_UINT16_FLOAT): Likewise.
24209 (BT_FN_UINT8_FLOAT): Likewise.
24210 (BT_FN_UINT64_FLOAT): Likewise.
24211 (BT_FN_UINT16_UINT32): Likewise.
24212 (BT_FN_UINT32_UINT16): Likewise.
24213 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24214 (BT_FN_INT_PTR_INT): Likewise.
24215 (BT_FN_UINT_PTR_UINT): Likewise.
24216 (BT_FN_LONG_PTR_LONG): Likewise.
24217 (BT_FN_ULONG_PTR_ULONG): Likewise.
24218 (BT_FN_VOID_UINT64_UINT64): Likewise.
24219 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24220 (BT_FN_INT8_INT8_INT8): Likewise.
24221 (BT_FN_INT16_INT16_INT16): Likewise.
24222 (BT_FN_INT_INT_INT): Likewise.
24223 (BT_FN_UINT_FLOAT_UINT): Likewise.
24224 (BT_FN_FLOAT_UINT_UINT): Likewise.
24225 (BT_FN_ULONG_UINT_UINT): Likewise.
24226 (BT_FN_ULONG_UINT_PTR): Likewise.
24227 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24228 (BT_FN_UINT_UINT_UINT): Likewise.
24229 (BT_FN_VOID_UINT_PTR): Likewise.
24230 (BT_FN_UINT_UINT_PTR: Likewise.
24231 (BT_FN_UINT32_UINT64_PTR): Likewise.
24232 (BT_FN_INT_INT_UINT_UINT): Likewise.
24233 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24234 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24235 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24236 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24237 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24238 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24239 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24240 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24241 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24242 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24243 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24244 * doc/frontends.texi: List BRIG FE.
24245 * doc/install.texi (Testing): Add BRIG tesring requirements.
24246 * doc/invoke.texi (Overall Options): Mention BRIG.
24247 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24249 2017-01-24 Richard Biener <rguenther@suse.de>
24251 PR translation/79208
24252 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24254 2017-01-24 Martin Jambor <mjambor@suse.cz>
24257 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24258 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24259 and known_contexts.
24261 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24263 PR middle-end/79123
24264 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24265 casts from signed to unsigned really don't have a range.
24267 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24269 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24270 GMP_RNDx for compatiblity.
24272 2017-01-24 Martin Liska <mliska@suse.cz>
24275 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24276 that would prevent us to call alloca with -1 as argument.
24278 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24280 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24281 Avoid -Wformat-security warning.
24283 2017-01-23 Andrew Pinski <apinski@cavium.com>
24285 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24288 2017-01-23 Martin Sebor <msebor@redhat.com>
24290 PR middle-end/78703
24291 * gimple-ssa-sprintf.c (warn_level): New global.
24292 (format_integer): Use it here and throughout the rest of the file.
24293 Use the same switch to compute sign as base.
24294 (maybe_warn): New function.
24295 (format_directive): Factor out warnings into maybe_warn.
24296 Add debugging output. Use warn_level.
24297 (add_bytes): Use warn_level.
24298 (pass_sprintf_length::compute_format_length): Add debugging output.
24299 (try_substitute_return_value): Same.
24300 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24302 PR middle-end/78703
24303 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24304 (struct fmtresult, format_integer, format_floating): Adjust.
24305 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24306 (get_string_length, format_string,format_directive): Same.
24307 (pass_sprintf_length::compute_format_length): Same.
24308 (try_substitute_return_value): Simplify slightly.
24310 PR middle-end/78703
24311 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24312 (fmtresult::operator+=): Outlined.
24313 (struct fmtresult): Add ctors.
24314 (struct conversion_spec): Rename...
24315 (struct directive): ...to this. Add and remove data members.
24316 (directive::set_width, directive::set_precision): New functions.
24317 (format_percent): Use fmtresult ctor.
24318 (get_width_and_precision): Remove.
24319 (format_integer): Make naming changes. Avoid computing width and
24321 (format_floating): Same. Adjust indentation.
24322 (format_character, format_none): New functions.
24323 (format_string): Moved character handling to format_character.
24324 (format_directive): Remove arguments, change return type.
24325 (parse_directive): New function.
24326 (pass_sprintf_length::compute_format_length): Move directive
24327 parsing to parse_directive.
24329 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24331 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24332 (assign_assembler_name_if_needed): ... this.
24333 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24334 (assign_assembler_name_if_needed): ... this.
24335 (free_lang_data_in_cgraph): Adjust callers.
24336 * cgraphunit.c (cgraph_node::analyze): Likewise.
24337 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24340 2017-01-23 Richard Biener <rguenther@suse.de>
24342 PR tree-optimization/79088
24343 PR tree-optimization/79188
24344 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24345 resetting loop bounds after last path deletion. Reset loop
24346 bounds of the target loop, make code match the comments.
24347 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24348 Make sure loops need no fixups.
24350 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24352 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24353 exponent support with double type for first argument.
24354 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24355 type returned by __builtin_vec_extract_sig,
24356 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24357 functions from "vector int" to "vector unsigned int" or from
24358 "vector long long int" to "vector unsigned long long int".
24359 Changed type returned by __builtin_vec_extract_exp,
24360 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24361 functions from "vector int" to "vector unsigned int" or from
24362 "vector long long int" to "vector unsigned long long int".
24363 Changed return type of __builtin_vec_test_data_class,
24364 __builtin_vec_test_data_class_sp, and
24365 __builtin_vec_test_data_class_dp from "vector int" to
24366 "vector bool int" or from "vector long long int" to "vector bool
24367 long long int" and changed second argument type from "unsigned
24368 int" to "int". Added new overloaded function forms "vector float
24369 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24370 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24371 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24372 double, vector unsigned long long int)" and "vector double
24373 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24374 long int)". Changed return type of
24375 __builtin_scalar_test_data_class and
24376 __builtin_scalar_test_data_class_sp and
24377 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24378 int" and changed second argument from "unsigned int" to "int".
24379 Changed type returned by __builtin_scalar_test_neg,
24380 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24381 from "int" to "bool int". Added new overloaded function form
24382 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24383 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24384 exponent double-precision with floating point first argument.
24385 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24386 documentation of scalar_test_data_class, scalar_test_neg,
24387 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24388 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24389 vec_test_data_class built-in functions to reflect refinements in
24390 their type signatures.
24392 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24394 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24396 (aarch64_elf_asm_destructor): Likewise.
24398 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24400 PR rtl-optimization/78634
24401 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24402 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24403 * ifcvt.c (noce_try_cmove): Add missing cost check.
24405 PR rtl-optimization/71724
24406 * combine.c (if_then_else_cond): Look for situations where it is
24407 beneficial to undo the work of one of the recursive calls.
24409 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24411 PR tree-optimization/70754
24412 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24413 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24414 combined stmt before it if not NULL.
24415 (combine_chains): Process refs reversely and compute dominance point
24418 2017-01-23 Martin Liska <mliska@suse.cz>
24420 PR tree-optimization/79196
24421 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24422 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24424 (strlen_optimize_stmt): Call the renamed function.
24426 2017-01-23 Michael Matz <matz@suse.de>
24428 PR tree-optimization/78384
24429 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24431 2017-01-23 Richard Biener <rguenther@suse.de>
24433 PR tree-optimization/79186
24434 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24435 both incoming edges before moving an assert.
24437 2017-01-23 Martin Jambor <mjambor@suse.cz>
24439 * ipa-prop.c (load_from_param_1): Removed.
24440 (load_from_unmodified_param): Bits from load_from_param_1 put back
24442 (load_from_param): Removed.
24443 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24444 with stmt. Reverted back to use of load_from_unmodified_param.
24446 2017-01-23 Martin Jambor <mjambor@suse.cz>
24449 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24450 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24451 field a pointer to garbage collected vector, mark lattices and
24452 ipcp_orig_node with GTY((skip)).
24453 (ipa_get_param_count): Adjust to descriptors being a pointer.
24454 (ipa_get_param): Likewise.
24455 (ipa_get_type): Likewise.
24456 (ipa_get_param_move_cost): Likewise.
24457 (ipa_set_param_used): Likewise.
24458 (ipa_get_controlled_uses): Likewise.
24459 (ipa_set_controlled_uses): Likewise.
24460 (ipa_is_param_used): Likewise.
24461 (ipa_node_params_t): Move into garbage collector. New methods insert
24463 (ipa_node_params_sum): Annotate wth GTY(()).
24464 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24466 (ipa_load_from_parm_agg): Adjust declaration.
24467 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24468 * ipa-profile.c (ipa_profile): Likewise.
24469 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24470 (ipa_populate_param_decls): Make descriptors parameter garbage
24472 (ipa_dump_param): Adjust to descriptors being a pointer.
24473 (ipa_alloc_node_params): Likewise.
24474 (ipa_initialize_node_params): Likewise.
24475 (load_from_param_1): Make descriptors parameter garbage collected.
24476 (load_from_unmodified_param): Likewise.
24477 (load_from_param): Likewise.
24478 (ipa_load_from_parm_agg): Likewise.
24479 (ipa_node_params::~ipa_node_params): Removed.
24480 (ipa_free_all_node_params): Remove call to delete operator.
24481 (ipa_node_params_t::insert): New.
24482 (ipa_node_params_t::remove): Likewise.
24483 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24484 copy known_csts and known_contexts vectors.
24485 (ipa_read_node_info): Adjust to descriptors being a pointer.
24486 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24488 (ipcp_transform_function): Make descriptors variable garbage
24491 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24493 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24494 * config/i386/avx512dqintrin.h: Ditto.
24495 * config/i386/avx512fintrin.h: Ditto.
24496 * config/i386/i386.c: Handle new builtins.
24497 * config/i386/i386-builtin.def: Add new builtins.
24498 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24499 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24501 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24502 Martin Liska <mliska@suse.cz>
24504 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24505 * asan.c (asan_expand_poison_ifn): Support stores and use
24506 appropriate ASAN report function.
24507 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24508 * internal-fn.def (ASAN_POISON_USE): Declare.
24509 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24510 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24511 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24512 ASAN_POISON calls w/o LHS.
24513 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24514 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24515 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24516 * gimplify.c (asan_poison_variables): Add attribute
24517 use_after_scope_memory to variables that really needs to live
24519 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24520 having the attribute.
24522 2017-01-23 Martin Liska <mliska@suse.cz>
24524 * asan.c (create_asan_shadow_var): New function.
24525 (asan_expand_poison_ifn): Likewise.
24526 * asan.h (asan_expand_poison_ifn): New declaration.
24527 * internal-fn.c (expand_ASAN_POISON): Likewise.
24528 * internal-fn.def (ASAN_POISON): New builtin.
24529 * sanopt.c (pass_sanopt::execute): Expand
24530 asan_expand_poison_ifn.
24531 * tree-inline.c (copy_decl_for_dup_finish): Make function
24533 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24534 * tree-ssa.c (is_asan_mark_p): New function.
24535 (execute_update_addresses_taken): Rewrite local variables
24536 (identified just by use-after-scope as addressable) into SSA.
24538 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24540 * doc/install.texi (Specific): opensource.apple.com uses https
24541 now. Remove trailing slash.
24543 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24545 * README.Portability: Remove note on an Irix compatibility issue.
24547 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24549 * gcov.c (INCLUDE_ALGORITHM): Define.
24550 (INCLUDE_VECTOR): Define.
24551 No longer include <vector> and <algorithm> directly.
24553 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24555 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24557 * doc/invoke.texi (Code Gen Options): Ditto.
24559 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24562 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24564 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24566 rtl-optimization/79125
24567 * cprop.c (local_cprop_pass): Handle cases where we make an
24568 unconditional trap.
24570 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24574 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24575 read from, for big endian.
24577 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24579 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24580 register pauth builtins for LP64 only.
24582 2017-01-20 Marek Polacek <polacek@redhat.com>
24585 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24588 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24590 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24592 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24593 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24594 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24596 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24599 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24600 in big-endian mode when they are not a single duplicated value.
24602 2017-01-20 Richard Biener <rguenther@suse.de>
24604 * BASE-VER: Bump to 7.0.1.
24606 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24608 * omp-low.c (omplow_simd_context): New struct. Use it...
24609 (lower_rec_simd_input_clauses): ...here and...
24610 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24611 references to idx, lane, max_vf, is_simt.
24613 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24615 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24618 2017-01-20 Martin Jambor <mjambor@suse.cz>
24620 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24621 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24623 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24624 (GTFILES): Rename hsa.c to hsa-common.c.
24625 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24626 * hsa-dump.c: Likewise.
24627 * hsa-gen.c: Likewise.
24628 * hsa-regalloc.c: Likewise.
24629 * ipa-hsa.c: Likewise.
24630 * omp-expand.c: Likewise.
24631 * omp-low.c: Likewise.
24632 * toplev.c: Likewise.
24634 2017-01-20 Marek Polacek <polacek@redhat.com>
24637 * doc/invoke.texi: Document -Wduplicated-branches.
24638 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24639 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24640 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24641 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24642 return 0 only when not OEP_LEXICOGRAPHIC.
24643 (fold_build_cleanup_point_expr): Use the expression
24644 location when building CLEANUP_POINT_EXPR.
24645 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24646 * tree.c (add_expr): Handle error_mark_node.
24648 2017-01-20 Martin Liska <mliska@suse.cz>
24651 * tree-profile.c (init_ic_make_global_vars): Do not call
24653 (gimple_init_gcov_profiler): Likewise.
24655 2017-01-20 Martin Liska <mliska@suse.cz>
24658 * cgraph.h (maybe_create_reference): Remove argument and
24660 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24662 * ipa-cp.c (create_specialized_node): Likewise.
24663 * symtab.c (symtab_node::maybe_create_reference): Handle
24664 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24666 2017-01-20 Martin Liska <mliska@suse.cz>
24668 * read-rtl-function.c (function_reader::create_function): Use
24669 build_decl instread of build_decl_stat.
24671 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24673 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24674 * config/i386/avx512dqintrin.h: Ditto.
24675 * config/i386/avx512fintrin.h: Ditto.
24676 * config/i386/i386-builtin-types.def: Add new types.
24677 * config/i386/i386.c: Handle new types.
24678 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24679 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24680 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24681 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24682 (__builtin_ia32_kshiftridi): New.
24683 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24685 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24689 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24690 define to rs6000_init_stack_protect_guard.
24691 (rs6000_init_stack_protect_guard): New function.
24693 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24694 Yunqiang Su <yunqiang.su@imgtec.com>
24696 * config.gcc (supported_defaults): Add madd4.
24697 (with_madd4): Add validation.
24698 (all_defaults): Add madd4.
24699 * config/mips/mips.opt (mmadd4): New option.
24700 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24702 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24704 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24705 (ISA_HAS_FUSED_MADD4): Likewise.
24706 * doc/invoke.texi (-mmadd4): Document the new option.
24707 * doc/install.texi (--with-madd4): Document the new option.
24709 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24711 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24712 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24713 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24714 (aarch64_init_pauth_hint_builtins): New.
24715 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24716 (aarch64_expand_builtin): Expand new builtins.
24718 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24720 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24721 * combine-stack-adj.c (no_unhandled_cfa): Handle
24722 REG_CFA_TOGGLE_RA_MANGLE.
24723 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24724 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24725 info for return address signing.
24726 (aarch64_expand_epilogue): Likewise.
24728 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24730 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24731 * config/aarch64/aarch64-protos.h
24732 (aarch64_return_address_signing_enabled): New declaration.
24733 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24735 (aarch64_expand_prologue): Sign return address before it's pushed onto
24737 (aarch64_expand_epilogue): Authenticate return address fetched from
24739 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24740 (aarch64_attributes): New function attributes for "sign-return-address".
24741 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24742 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24743 ("*do_return"): Generate combined instructions according to key index.
24744 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24745 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24747 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24748 * config/aarch64/aarch64.opt (msign-return-address=): New.
24749 * doc/extend.texi (AArch64 Function Attributes): Documents
24750 "sign-return-address=".
24751 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24753 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24755 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24756 overall option summary.
24758 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24760 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24761 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24762 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24763 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24765 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24767 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24768 -mpower9-minmax by default for -mcpu=power9.
24769 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24770 128-bit floating point.
24772 2017-01-20 Alan Modra <amodra@gmail.com>
24774 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24775 optimizing for size.
24777 2017-01-20 Alan Modra <amodra@gmail.com>
24780 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24781 for strcmp and strncmp from corresponding builtin decl.
24783 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24785 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24786 instead of i386/rtems-64.h.
24787 * config/i386/rtems-64.h: Remove.
24789 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24793 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24795 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24797 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24799 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24800 Change int to HOST_WIDE_INT.
24801 * config/aarch64/aarch64-protos.h
24802 (aarch64_simd_gen_const_vector_dup): Likewise.
24803 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24805 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24807 * langhooks-def.h (lhd_type_for_size): New decl.
24808 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24809 * langhooks.c (lhd_type_for_size): New function, taken from
24812 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24814 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24815 define_bypass for CR latency.
24816 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24817 (power9-alu2): Add define_bypass for CR latency.
24819 (power9-mul): Update insn latency.
24820 (power9-mul-compare): Update insn latency, bypass latency and remove
24823 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24825 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24827 * config/aarch64/aarch64.md
24828 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24829 aarch64_nopcrelative_literal_loads.
24830 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24832 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24834 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24835 TARGET_LOONGSON_3A.
24836 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24838 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24841 * config.gcc (supported_defaults): Add lxc1-sxc1.
24842 (with_lxc1_sxc1): Add validation.
24843 (all_defaults): Add lxc1-sxc1.
24844 * config/mips/mips.opt (mlxc1-sxc1): New option.
24845 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24847 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24848 __mips_no_lxc1_sxc1.
24849 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24850 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24851 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24853 2017-01-19 Richard Biener <rguenther@suse.de>
24855 PR tree-optimization/72488
24856 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24857 sure to restore SSA info.
24858 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24860 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24862 PR rtl-optimization/79121
24863 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24864 of the inner type when shifting an extended value.
24866 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24869 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24870 interposable aliases.
24872 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24875 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24876 Use the evmergelohi instruction.
24877 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24878 (mov_sitf_e500_subreg8_2_be): Likewise.
24879 (mov_sitf_e500_subreg12_2_le): Likewise.
24880 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24881 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24882 (mov_sitf_e500_subreg8_2_le): Likewise.
24883 (mov_sitf_e500_subreg12_2_be): Likewise.
24885 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24887 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24888 attribute from vecsimple to vecperm.
24889 (altivec_vbpermq2): Likewise.
24891 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24894 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24896 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24897 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24898 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24899 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24900 case where N arg is SIZE_MAX.
24901 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24902 (cmpstrsi): Add pattern.
24904 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24906 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24907 __builtin_vec_revb builtins.
24908 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24909 built-in functions to support generation of the ISA 3.0 XXBR<x>
24910 vector byte reverse instructions.
24911 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24912 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24913 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24914 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24915 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24916 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24917 (P9V_BUILTIN_VEC_REVB): Likewise.
24918 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24919 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24920 (p9_xxbrq_v16qi): Likewise.
24921 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24922 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24923 (p9_xxbrh_v8hi): Likewise.
24924 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24925 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24926 vec_revb built-in functions.
24928 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24930 PR rtl-optimization/78952
24931 * config/i386/i386.md (any_extract): New code iterator.
24932 (*insvqi_2): Use any_extract for source operand.
24933 (*insvqi_3): Use any_shiftrt for source operand.
24935 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24937 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24939 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24941 2017-01-18 Matthias Klose <doko@ubuntu.com>
24943 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24945 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24947 * config/rs6000/altivec.h (vec_bperm): Change #define.
24948 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24949 (altivec_vbpermq2): New define_insn.
24950 (altivec_vbpermd): Likewise.
24951 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24952 function interface.
24953 (VBPERMD): Likewise.
24954 (VBPERM): New polymorphic function interface.
24955 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24956 Add entries for P9V_BUILTIN_VEC_VBPERM.
24957 * doc/extend.texi: Add interfaces for vec_bperm.
24959 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24961 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24962 first letter of error messages.
24963 (s390_resolve_overloaded_builtin): Likewise.
24964 * config/s390/s390.c (s390_expand_builtin): Likewise.
24965 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24966 (s390_valid_target_attribute_inner_p): Likewise.
24967 * config/s390/s390.md ("tabort"): Likewise.
24969 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24971 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24972 (ISA_AVOID_DIV_HILO): New macro.
24973 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24974 (ISA_HAS_DDIV): Likewise.
24976 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24978 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24980 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24982 * doc/invoke.texi (fabi-version): Spelling fix.
24984 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24987 * doc/invoke.texi (fabi-version): Mention mangling fix for
24990 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24993 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24995 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24998 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24999 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25001 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25003 (stack_protect_test): Ditto.
25004 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25005 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25007 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25008 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25009 -mstack-protector-guard-offset=.
25010 (RS/6000 and PowerPC Options): Ditto.
25012 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25014 * config/i386/i386.h (MASK_CLASS_P): New define.
25015 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25016 there are no registers from different register sets also when
25017 mask registers are used. Update function comment.
25018 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25019 to (*k/*r) and (*k/*km) alternatives.
25021 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25023 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25024 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25025 (EH_RETURN_HANDLER_RTX): New define.
25026 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25027 Force frame pointer in EH return functions.
25028 (aarch64_expand_epilogue): Add barrier for eh_return.
25029 (aarch64_final_eh_return_addr): Remove.
25030 (aarch64_eh_return_handler_rtx): New function.
25031 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25033 (aarch64_eh_return_handler_rtx): New prototype.
25035 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25037 * config/rs6000/altivec.h (vec_rlmi): New #define.
25038 (vec_vrlnm): Likewise.
25039 (vec_rlnm): Likewise.
25040 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25041 (UNSPEC_VRLNM): Likewise.
25042 (VIlong): New mode iterator.
25043 (altivec_vrl<VI_char>mi): New define_insn.
25044 (altivec_vrl<VI_char>nm): Likewise.
25045 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25047 (VRLDNM): Likewise.
25048 (RLNM): New polymorphic function entry.
25049 (VRLWMI): New monomorphic function entry.
25050 (VRLDMI): Likewise.
25051 (RLMI): New polymorphic function entry.
25052 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25053 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25054 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25057 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25060 * dwarf2out.c (field_byte_offset): Restore the
25061 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25062 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25063 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25066 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25069 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25071 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25074 * dwarf2out.c (add_data_member_location_attribute): For constant
25075 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25076 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25077 DW_AT_byte_size attributes.
25079 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25081 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25082 after forcing to constant memory when the code model is medium.
25084 2017-01-17 Julia Koval <julia.koval@intel.com>
25087 * config/i386/avx512fintrin.h
25088 (_mm512_i32gather_ps): Change __addr type to void const*.
25089 (_mm512_mask_i32gather_ps): Ditto.
25090 (_mm512_i32gather_pd): Ditto.
25091 (_mm512_mask_i32gather_pd): Ditto.
25092 (_mm512_i64gather_ps): Ditto.
25093 (_mm512_mask_i64gather_ps): Ditto.
25094 (_mm512_i64gather_pd): Ditto.
25095 (_mm512_mask_i64gather_pd): Ditto.
25096 (_mm512_i32gather_epi32): Ditto.
25097 (_mm512_mask_i32gather_epi32): Ditto.
25098 (_mm512_i32gather_epi64): Ditto.
25099 (_mm512_mask_i32gather_epi64): Ditto.
25100 (_mm512_i64gather_epi32): Ditto.
25101 (_mm512_mask_i64gather_epi32): Ditto.
25102 (_mm512_i64gather_epi64): Ditto.
25103 (_mm512_mask_i64gather_epi64): Ditto.
25104 (_mm512_i32scatter_ps): Change __addr type to void*.
25105 (_mm512_mask_i32scatter_ps): Ditto.
25106 (_mm512_i32scatter_pd): Ditto.
25107 (_mm512_mask_i32scatter_pd): Ditto.
25108 (_mm512_i64scatter_ps): Ditto.
25109 (_mm512_mask_i64scatter_ps): Ditto.
25110 (_mm512_i64scatter_pd): Ditto.
25111 (_mm512_mask_i64scatter_pd): Ditto.
25112 (_mm512_i32scatter_epi32): Ditto.
25113 (_mm512_mask_i32scatter_epi32): Ditto.
25114 (_mm512_i32scatter_epi64): Ditto.
25115 (_mm512_mask_i32scatter_epi64): Ditto.
25116 (_mm512_i64scatter_epi32): Ditto.
25117 (_mm512_mask_i64scatter_epi32): Ditto.
25118 (_mm512_i64scatter_epi64): Ditto.
25119 (_mm512_mask_i64scatter_epi64): Ditto.
25120 * config/i386/avx512pfintrin.h
25121 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25122 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25123 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25124 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25125 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25126 (_mm512_prefetch_i32scatter_ps): Ditto.
25127 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25128 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25129 (_mm512_prefetch_i64scatter_pd): Ditto.
25130 (_mm512_prefetch_i64scatter_ps): Ditto.
25131 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25132 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25133 * config/i386/avx512vlintrin.h
25134 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25135 (_mm_mmask_i32gather_ps): Ditto.
25136 (_mm256_mmask_i32gather_pd): Ditto.
25137 (_mm_mmask_i32gather_pd): Ditto.
25138 (_mm256_mmask_i64gather_ps): Ditto.
25139 (_mm_mmask_i64gather_ps): Ditto.
25140 (_mm256_mmask_i64gather_pd): Ditto.
25141 (_mm_mmask_i64gather_pd): Ditto.
25142 (_mm256_mmask_i32gather_epi32): Ditto.
25143 (_mm_mmask_i32gather_epi32): Ditto.
25144 (_mm256_mmask_i32gather_epi64): Ditto.
25145 (_mm_mmask_i32gather_epi64): Ditto.
25146 (_mm256_mmask_i64gather_epi32): Ditto.
25147 (_mm_mmask_i64gather_epi32): Ditto.
25148 (_mm256_mmask_i64gather_epi64): Ditto.
25149 (_mm_mmask_i64gather_epi64): Ditto.
25150 (_mm256_i32scatter_ps): Change __addr type to void*.
25151 (_mm256_mask_i32scatter_ps): Ditto.
25152 (_mm_i32scatter_ps): Ditto.
25153 (_mm_mask_i32scatter_ps): Ditto.
25154 (_mm256_i32scatter_pd): Ditto.
25155 (_mm256_mask_i32scatter_pd): Ditto.
25156 (_mm_i32scatter_pd): Ditto.
25157 (_mm_mask_i32scatter_pd): Ditto.
25158 (_mm256_i64scatter_ps): Ditto.
25159 (_mm256_mask_i64scatter_ps): Ditto.
25160 (_mm_i64scatter_ps): Ditto.
25161 (_mm_mask_i64scatter_ps): Ditto.
25162 (_mm256_i64scatter_pd): Ditto.
25163 (_mm256_mask_i64scatter_pd): Ditto.
25164 (_mm_i64scatter_pd): Ditto.
25165 (_mm_mask_i64scatter_pd): Ditto.
25166 (_mm256_i32scatter_epi32): Ditto.
25167 (_mm256_mask_i32scatter_epi32): Ditto.
25168 (_mm_i32scatter_epi32): Ditto.
25169 (_mm_mask_i32scatter_epi32): Ditto.
25170 (_mm256_i32scatter_epi64): Ditto.
25171 (_mm256_mask_i32scatter_epi64): Ditto.
25172 (_mm_i32scatter_epi64): Ditto.
25173 (_mm_mask_i32scatter_epi64): Ditto.
25174 (_mm256_i64scatter_epi32): Ditto.
25175 (_mm256_mask_i64scatter_epi32): Ditto.
25176 (_mm_i64scatter_epi32): Ditto.
25177 (_mm_mask_i64scatter_epi32): Ditto.
25178 (_mm256_i64scatter_epi64): Ditto.
25179 (_mm256_mask_i64scatter_epi64): Ditto.
25180 (_mm_i64scatter_epi64): Ditto.
25181 (_mm_mask_i64scatter_epi64): Ditto.
25182 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25183 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25184 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25185 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25186 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25187 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25188 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25189 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25190 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25191 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25192 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25193 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25194 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25195 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25196 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25197 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25198 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25199 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25200 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25201 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25202 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25203 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25204 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25205 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25206 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25207 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25208 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25209 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25210 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25211 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25212 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25213 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25214 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25215 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25216 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25217 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25218 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25219 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25220 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25221 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25222 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25223 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25224 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25225 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25226 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25227 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25228 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25229 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25230 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25231 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25232 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25233 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25234 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25235 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25236 definitions accordingly.
25238 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25239 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25242 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25245 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25248 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25249 hard regs for inner regno.
25251 2017-01-17 Martin Liska <mliska@suse.cz>
25254 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25255 assumption and add comment.
25257 2017-01-17 Nathan Sidwell <nathan@acm.org>
25259 * ipa-visibility.c (localize_node): New function, broken out of ...
25260 (function_and_variable_visibility): ... here. Call it.
25262 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25264 PR middle-end/77445
25265 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25266 correctly set frequency of oudgoing edge.
25267 (duplicate_thread_path): Fix profile updating.
25269 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25272 * configure.ac: Add GCC_BASE_VER.
25273 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25274 version from BASE-VER file.
25275 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25276 (gcc.o): Depend on $(BASEVER).
25277 * common.opt (dumpfullversion): New option.
25278 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25279 * doc/invoke.texi: Document -dumpfullversion.
25280 * doc/install.texi: Document --with-gcc-major-version-only.
25281 * configure: Regenerated.
25283 2017-01-17 Richard Biener <rguenther@suse.de>
25285 PR tree-optimization/71433
25286 * tree-vrp.c (register_new_assert_for): Merge same asserts
25287 on all incoming edges.
25288 (process_assert_insertions_for): Handle insertions at the
25291 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25293 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25294 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25296 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25299 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25302 2017-01-17 Alan Modra <amodra@gmail.com>
25305 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25306 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25307 symbolic stack limit when pic.
25309 2017-01-16 Martin Sebor <msebor@redhat.com>
25311 PR tree-optimization/78608
25312 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25314 2017-01-16 Jeff Law <law@redhat.com>
25317 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25318 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25319 for several include directories that may be relative to sysroot.
25320 * config/i386/x-mingw32 (gplus_includedir): Define.
25321 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25322 (native_system_includedir): Likewise.
25323 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25324 override if TARGET_SYSTEM_ROOT is defined.
25325 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25327 PR tree-optimization/79090
25328 PR tree-optimization/33562
25329 PR tree-optimization/61912
25330 PR tree-optimization/77485
25331 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25332 and computed trims into the dump file.
25334 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25336 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25338 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25341 * gimplify.c (gimplify_init_constructor): If want_value and
25342 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25346 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25347 sequence. Formatting fixes.
25348 (doloop_optimize): Formatting fixes.
25351 * gcc.c (debug_level_greater_than_spec_func): New function.
25352 (static_spec_functions): Add debug-level-gt spec function.
25353 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25355 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25356 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25357 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25358 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25359 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25360 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25362 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25364 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25365 QImode fixups to general and mask registers only.
25367 2017-01-16 Carl Love <cel@us.ibm.com>
25369 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25370 for built-in functions
25371 vector signed char vec_nabs (vector signed char)
25372 vector signed short vec_nabs (vector signed short)
25373 vector signed int vec_nabs (vector signed int)
25374 vector signed long long vec_nabs (vector signed long long)
25375 vector float vec_nabs (vector float)
25376 vector double vec_nabs (vector double)
25377 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25379 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25380 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25381 * doc/extend.texi: Update the documentation file for the new built-in
25384 2017-01-16 Martin Sebor <msebor@redhat.com>
25386 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25389 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25391 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25392 UNSPEC_VSX__XXSPLTD to require special splat handling.
25394 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25397 * system.h: Poison strndup.
25399 2017-01-16 Alan Modra <amodra@gmail.com>
25402 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25405 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25407 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25409 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25411 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25412 call recog here. Assert that INSN_CODE (insn) is non-negative.
25414 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25417 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25419 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25420 in the currently scheduled RTL fragment.
25422 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25424 PR rtl-optimization/78751
25425 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25428 2017-01-14 Jeff Law <law@redhat.com>
25430 PR tree-optimization/79090
25431 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25432 variable length stores.
25433 (compute_trims): Delete dead assignment to *trim_tail.
25434 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25437 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25439 PR rtl-optimization/78626
25440 PR rtl-optimization/78727
25441 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25442 of a block, and split such blocks after everything else is finished.
25444 2017-01-14 Alan Modra <amodra@gmail.com>
25447 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25448 target legitimate_combined_insn.
25449 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25450 (rs6000_legitimate_combined_insn): New function.
25451 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25453 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25454 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25455 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25457 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25459 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25461 2017-01-13 Jeff Law <law@redhat.com>
25463 PR tree-optimization/33562
25464 PR tree-optimization/61912
25465 PR tree-optimization/77485
25466 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25468 (delete_dead_assignment): Likewise.
25469 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25470 statement to delete_dead_call and delete_dead_assignment.
25472 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25475 * substring-locations.c (format_warning_va): Strengthen case 1 so
25476 that both endpoints of the substring must be within the format
25477 range for just the substring to be printed.
25479 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25481 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25482 * config/i386/i386.c (ix86_target_string): Add missing options
25483 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25484 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25485 flags_other and ix86_target_other to flags2_other. Display unknown
25487 (ix86_valid_target_attribute_inner_p): Add missing options and
25488 reorder options by implied ISAs, as in ix86_target_string.
25490 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25492 * hash-table.h (hash_table::too_empty_p): New function.
25493 (hash_table::expand): Use it.
25494 (hash_table::traverse): Likewise.
25495 (hash_table::empty_slot): Use sizeof (value_type) instead of
25496 sizeof (PTR) to convert bytes to elements. Shrink the table
25497 if the current size is excessive for the current number of
25500 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25502 * ira-costs.c (record_reg_classes): Break from the inner loop
25503 early once alt_fail is known to be true. Update outer loop
25504 handling accordingly.
25506 2017-01-13 Jeff Law <law@redhat.com>
25508 * tree-ssa-dse.c (decrement_count): New function.
25509 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25510 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25511 when we know the partially dead statement is a mem* function.
25513 PR tree-optimization/61912
25514 PR tree-optimization/77485
25515 * tree-ssa-dse.c: Include expr.h.
25516 (maybe_trim_constructor_store): New function.
25517 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25519 PR tree-optimization/33562
25520 PR tree-optimization/61912
25521 PR tree-optimization/77485
25522 * doc/invoke.texi: Document new dse-max-object-size param.
25523 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25524 * tree-ssa-dse.c: Include params.h.
25525 (dse_store_status): New enum.
25526 (initialize_ao_ref_for_dse): New, partially extracted from
25528 (valid_ao_ref_for_dse, normalize_ref): New.
25529 (setup_live_bytes_from_ref, compute_trims): Likewise.
25530 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25531 (maybe_trim_partially_dead_store): Likewise.
25532 (maybe_trim_complex_store): Likewise.
25533 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25534 Track what bytes live from the original store. Return tri-state
25535 for dead, partially dead or live.
25536 (dse_dom_walker): Add constructor, destructor and new private members.
25537 (delete_dead_call, delete_dead_assignment): New extracted from
25539 (dse_optimize_stmt): Make a member of dse_dom_walker.
25540 Use initialize_ao_ref_for_dse.
25542 PR tree-optimization/33562
25543 PR tree-optimization/61912
25544 PR tree-optimization/77485
25545 * sbitmap.h (bitmap_count_bits): Prototype.
25546 (bitmap_clear_range, bitmap_set_range): Likewise.
25547 * sbitmap.c (bitmap_clear_range): New function.
25548 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25550 2017-01-13 Martin Liska <mliska@suse.cz>
25553 * function.c (set_cfun): Add new argument force.
25554 * function.h (set_cfun): Likewise.
25555 * ipa-inline-transform.c (inline_call): Use the function when
25556 strict alising from is dropped for function we inline to.
25558 2017-01-13 Richard Biener <rguenther@suse.de>
25560 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25561 for dumping GIMPLE INTEGER_CSTs.
25563 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25565 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25566 to 201112L since C++17.
25568 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25571 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25572 if -fsanitize=kernel-address is present.
25574 2017-01-13 Richard Biener <rguenther@suse.de>
25576 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25577 as _Literal ( type ) number in case usual suffixes do not
25578 preserve all information.
25580 2017-01-13 Richard Biener <rguenther@suse.de>
25582 PR tree-optimization/77283
25583 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25584 and ssa-iterators.h.
25585 (is_feasible_trace): Implement a cost model based on joiner
25588 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25591 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25592 char or short to __float128/_Float128 directly.
25594 2017-01-12 Martin Sebor <msebor@redhat.com>
25596 to -Wformat-overflow.
25597 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25598 (min_bytes_remaining): Same.
25599 (get_string_length): Same.
25600 (format_string): Same.
25601 (format_directive): Same.
25603 (pass_sprintf_length::handle_gimple_call): Same.
25605 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25607 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25608 info.nowrite calls with no lhs that can't throw. Return bool
25609 whether gsi_remove has been called or not.
25610 (pass_sprintf_length::handle_gimple_call): Return bool whether
25611 try_substitute_return_value called gsi_remove. Formatting fix.
25612 (pass_sprintf_length::execute): Don't use gsi_remove if
25613 handle_gimple_call returned true.
25616 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25617 be removed due to side-effects, don't remove following barrier nor
25618 turn the successor edge into fallthru edge.
25620 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25623 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25624 element-reversing loads and stores as not swappable.
25626 2017-01-12 Nathan Sidwell <nathan@acm.org>
25627 Nicolai Stange <nicstange@gmail.com>
25629 * combine.c (try_combine): Don't ignore result of overlap checking
25630 loop. Combine overlap & asm check into single loop.
25632 2017-01-12 Richard Biener <rguenther@suse.de>
25634 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25635 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25637 2017-01-12 Richard Biener <rguenther@suse.de>
25639 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25640 and TS_TARGET_OPTION directly derive from TS_BASE.
25641 * tree-core.h (tree_optimization_option): Derive from tree_base.
25642 (tree_target_option): Likewise.
25644 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25646 * config/i386/i386.c (memory_address_length): Increase len
25647 only when rip_relative_addr_p returns false.
25649 2017-01-11 Julia Koval <julia.koval@intel.com>
25651 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25652 (OPTION_MASK_ISA_SGX_SET): New.
25653 (ix86_handle_option): Handle OPT_msgx.
25654 * config.gcc: Added sgxintrin.h.
25655 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25656 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25657 * config/i386/i386.c (ix86_target_string): Add -msgx.
25659 (ix86_option_override_internal): Handle new options.
25660 (ix86_valid_target_attribute_inner_p): Add sgx.
25661 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25662 * config/i386/i386.opt: Add msgx.
25663 * config/i386/sgxintrin.h: New file.
25664 * config/i386/x86intrin.h: Add sgxintrin.h.
25666 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25669 * fold-const.c (maybe_nonzero_address): Return 1 for function
25671 (tree_single_nonzero_warnv_p): Don't handle function local objects
25675 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25678 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25681 * opts.c: Include "spellcheck.h"
25682 (struct string_fragment): New struct.
25683 (struct edit_distance_traits<const string_fragment &>): New
25685 (get_closest_sanitizer_option): New function.
25686 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25688 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25690 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25692 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25693 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25694 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25695 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25696 for initial die_offset if dwarf_split_debug_info.
25697 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25698 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25700 (output_skeleton_debug_sections): Formatting fix. Use
25701 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25702 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25704 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25706 * config/arm/cortex-a53.md: Add bypasses for
25707 cortex_a53_r2f_cvt.
25708 (cortex_a53_r2f): Only use for transfers.
25709 (cortex_a53_f2r): Likewise.
25710 (cortex_a53_r2f_cvt): Add reservation for conversions.
25711 (cortex_a53_f2r_cvt): Likewise.
25713 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25715 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25716 to all inlined functions, change static to extern.
25718 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25721 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25723 (arm_assemble_integer): Likewise.
25725 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25727 * config.gcc: Use new awk script to check CPU, FPU and architecture
25728 parameters for --with-... options.
25729 * config/arm/parsecpu.awk: New file
25730 * config/arm/arm-cpus.in: New file.
25731 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25733 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25735 * config/arm/t-arm: Update dependency rules.
25736 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25737 of processing .def files.
25738 * config/arm/genopt.sh: Deleted.
25739 * config/arm/gentune.sh: Deleted.
25740 * config/arm/arm-cores.def: Deleted.
25741 * config/arm/arm-arches.def: Deleted.
25742 * config/arm/arm-fpus.def: Deleted.
25743 * config/arm/arm-tune.md: Regenerated.
25744 * config/arm/arm-tables.opt: Regenerated.
25745 * config/arm/arm-cpu.h: New generated file.
25746 * config/arm/arm-cpu-data.h: New generated file.
25747 * config/arm/arm-cpu-cdata.h: New generated file.
25749 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25752 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25754 (input_varpool_node): Unpack dynamically_initialized bit.
25756 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25758 PR rtl-optimization/79032
25759 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25760 the alignment of the adjusted memory reference against that of MODE,
25761 instead of the alignment of the original memory reference.
25763 2017-01-11 Martin Jambor <mjambor@suse.cz>
25765 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25767 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25768 decorated functions.
25770 2017-01-11 Richard Biener <rguenther@suse.de>
25772 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25773 set range/nonnull info for PHI results. Do not set it on
25774 stmts marked for removal.
25776 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25778 * expr.c (store_field): In the bitfield case, fetch the return value
25779 from the registers before applying a single big-endian adjustment.
25780 Always do a final load for a BLKmode value not larger than a word.
25782 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25785 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25786 that we correctly handle column numbers greater than
25787 LINE_MAP_MAX_COLUMN_NUMBER.
25789 2017-01-10 Martin Sebor <msebor@redhat.com>
25791 PR middle-end/78245
25792 * gimple-ssa-sprintf.c (get_destination_size): Call
25793 {init,fini}object_sizes.
25794 * tree-object-size.c (addr_object_size): Adjust.
25795 (pass_through_call): Adjust.
25796 (pass_object_sizes::execute): Adjust.
25797 * tree-object-size.h (fini_object_sizes): Declare.
25799 2017-01-10 Martin Sebor <msebor@redhat.com>
25801 PR tree-optimization/78775
25802 * builtins.c (get_size_range): Move...
25803 * calls.c: ...to here.
25804 (alloc_max_size): Accept zero argument.
25805 (operand_signed_p): Remove.
25806 (maybe_warn_alloc_args_overflow): Call get_size_range.
25807 * calls.h (get_size_range): Declare.
25809 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25811 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25812 from TI's devices.csv file as of September 2016.
25813 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25815 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25817 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25818 * doc/invoke.texi: Likewise.
25819 * doc/md.texi: Likewise.
25820 * doc/objc.texi: Likewise.
25822 2017-01-10 Joshua Conner <joshconner@google.com>
25824 * config/arm/fuchsia-elf.h: New file.
25825 * config/fuchsia.h: New file.
25826 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25827 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25829 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25831 2016-01-10 Richard Biener <rguenther@suse.de>
25833 PR tree-optimization/79034
25834 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25835 Propagate out degenerate PHIs in the joiner.
25837 2017-01-10 Martin Liska <mliska@suse.cz>
25839 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25840 (sort_congruence_classes_by_decl_uid): Likewise.
25841 (sort_congruence_class_groups_by_decl_uid): Likewise.
25842 (sem_item_optimizer::merge_classes): Sort class, groups in these
25843 classes and members in the groups by DECL_UID of declarations.
25844 This would make merge operations stable.
25846 2017-01-10 Martin Liska <mliska@suse.cz>
25848 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25849 usage of m_classes_vec.
25850 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25851 (sem_item_optimizer::get_group_by_hash): Likewise.
25852 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25853 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25854 (sem_item_optimizer::verify_classes): Likewise.
25855 (sem_item_optimizer::process_cong_reduction): Likewise.
25856 (sem_item_optimizer::dump_cong_classes): Likewise.
25857 (sem_item_optimizer::merge_classes): Likewise.
25858 * ipa-icf.h (congruence_class_hash): Rename from
25859 congruence_class_group_hash. Remove declaration of m_classes_vec.
25861 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25863 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25864 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25865 * config.gcc: Add avx512vpopcntdqintrin.h.
25866 * config/i386/avx512vpopcntdqintrin.h: New.
25867 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25868 * config/i386/i386-builtin-types.def: Add new types.
25869 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25870 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25871 __builtin_ia32_vpopcountq_v8di_mask): New.
25872 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25873 __AVX512VPOPCNTDQ__.
25874 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25875 (PTA_AVX512VPOPCNTDQ): Define.
25876 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25877 TARGET_AVX512VPOPCNTDQ_P): Define.
25878 * config/i386/i386.opt: Add mavx512vpopcntdq.
25879 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25880 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25882 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25884 PR middle-end/77484
25885 * predict.def (PRED_CALL): Set to 67.
25887 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25889 * expr.c (store_field): In the bitfield case, if the value comes from
25890 a function call and is of an aggregate type returned in registers, do
25891 not modify the field mode; extract the value in all cases if the mode
25892 is BLKmode and the size is not larger than a word.
25894 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25897 * config/i386/cpuid.h: Fix undefined behavior.
25899 2017-01-04 Jeff Law <law@redhat.com>
25901 PR tree-optimization/79007
25902 PR tree-optimization/67955
25903 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25904 conservative for pt.null when flag_non_call_exceptions is on.
25906 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25908 PR translation/79019
25909 PR translation/79020
25910 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25911 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25912 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25914 * config/avr/avr.opt (maccumulate-args): Likewise.
25915 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25916 * common.opt (freport-bug): Likewise.
25917 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25918 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25919 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25920 translatable string.
25921 * config/i386/i386.c (function_value_32): Likewise.
25922 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25923 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25925 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25926 * common/config/msp430/msp430-common.c (msp430_handle_option):
25928 * symtab.c (symtab_node::verify_base): Likewise.
25929 * opts.c (set_debug_level): Likewise.
25930 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25931 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25932 missing whitespace to translatable strings.
25933 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25934 * config/sh/superh.h: Likewise.
25935 * config/i386/xopintrin.h: Likewise.
25936 * config/i386/znver1.md: Likewise.
25937 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25938 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25939 * double-int.h (struct double_int): Likewise.
25940 * double-int.c (div_and_round_double): Likewise.
25941 * wide-int.cc: Likewise.
25942 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25943 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25944 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25945 (crossjumps_occurred): ... this.
25946 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25949 PR tree-optimization/78899
25950 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25951 returning bool return struct loop *, NULL for failure and the new
25953 (versionable_outer_loop_p): Don't version outer loop if it has
25954 dont_vectorized bit set.
25955 (tree_if_conversion): When versioning outer loop, ensure
25956 tree_if_conversion is performed also on the inner loop of the
25957 non-vectorizable outer loop copy.
25958 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25959 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25960 prevent vectorization of it.
25961 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25962 the outer loop vectorization of the non-scalar version is attempted
25963 before vectorization of the inner loop in scalar version. If
25964 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25965 vectorization of its inner loop.
25966 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25967 has 2 inner loops, rename also on edges from bb whose single pred
25968 is outer_loop->header. Fix typo in function comment.
25970 2017-01-09 Martin Sebor <msebor@redhat.com>
25973 * asan.c (asan_emit_stack_protection): Increase local buffer size
25974 to avoid snprintf truncation warning.
25976 2017-01-09 Andrew Pinski <apinski@cavium.com>
25978 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25979 to reference thunderx2t99 for the tuning structure
25980 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25982 (thunderx2t99_extra_costs): This.
25983 * config/aarch64/aarch64-tune.md: Regenerate.
25984 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25985 (vulcan_addrcost_table): This.
25986 (vulcan_regmove_cost): Rename to ...
25987 (thunderx2t99_regmove_cost): This.
25988 (vulcan_vector_cost): Rename to ...
25989 (thunderx2t99_vector_cost): this.
25990 (vulcan_branch_cost): Rename to ...
25991 (thunderx2t99_branch_cost): This.
25992 (vulcan_tunings): Rename to ...
25993 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25994 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25996 2017-01-09 Martin Jambor <mjambor@suse.cz>
26000 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26001 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26002 (propagate_vr_accross_jump_function): Use the above function for all
26003 value range computations for pass-through jump functions and type
26004 converasion from explicit value range values.
26005 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26006 parameters from TYPE_ARG_TYPES.
26007 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26008 (ipa_write_node_info): Stream type of the actual argument.
26009 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26011 2017-01-09 Martin Liska <mliska@suse.cz>
26014 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26015 (lookup_compiler): Do not show error message with have_E.
26017 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26019 PR tree-optimization/78938
26020 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26021 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26022 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26023 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26026 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26028 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26031 2017-01-09 Richard Biener <rguenther@suse.de>
26033 PR tree-optimization/78997
26034 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26035 name condition properly.
26037 2017-01-09 Richard Biener <rguenther@suse.de>
26040 * dwarf2out.c (is_cxx): New overload with context.
26041 (is_naming_typedef_decl): Use it.
26043 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26045 * invoke.texi (Option Summary): Correct spacing in option lists
26046 and add line breaks to fix over-long lines.
26048 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26050 PR middle-end/17660
26052 * extend.texi (Common Variable Attributes): Add xref to GCC
26053 Internals manual to explain mode attribute keywords.
26055 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26058 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26059 and Preprocessor Options.
26060 (Options for Linking): Document -pthread here....
26061 (RS/6000 and PowerPC Options): ...not here.
26062 (Solaris 2 Options): ...or here.
26063 * doc/cppopts.texi: Document -pthread.
26065 2017-01-08 Martin Sebor <msebor@redhat.com>
26067 PR middle-end/77708
26068 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26069 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26070 New member functions.
26071 (format_directive): Used them.
26073 (pass_sprintf_length::handle_gimple_call): Same.
26074 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26075 to avoid truncation for any argument.
26076 (extract_affine_mul): Same.
26077 * tree.c (get_file_function_name): Same.
26079 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26081 PR middle-end/77484
26082 * predict.def (PRED_INDIR_CALL): Set to 86.
26084 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26086 PR preprocessor/54124
26087 * doc/cppopts.texi: Reformat -d subtable to list the full name
26088 of the options. Add cross-reference to the docs for the general
26089 compiler -d options.
26090 * doc/invoke.texi (Developer Options): Add cross-reference to the
26091 preprocessor-specific -d option documentation.
26093 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26095 PR preprocessor/13498
26096 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26097 redudant material, and reflect new command-line options.
26098 (System Headers): Likewise.
26100 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26102 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26103 -isystem, and -idirafter. Copy-edit.
26104 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26105 default for -ftrack-macro-expansion. Delete obsolete and
26106 badly-formatted implementation details about -fdebug-cpp output.
26107 * doc/cppwarnopts.texi: Copy-edit.
26109 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26112 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26113 that the transition from a max line width >= 1<<10 to narrower
26114 lines works correctly.
26116 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26118 * doc/options.texi (PerFunction): New.
26119 * opt-functions.awk (switch_flags): Map both Optimization and
26120 PerFunction to CL_OPTIMIZATION.
26121 * opth-gen.awk: Test for PerFunction flag along with
26123 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26124 it only when the latter is present. Skip those that don't in
26125 the hash function generator.
26126 * common.opt (fvar-tracking): Mark as PerFunction instead of
26128 (fvar-tracking-assignments): Likewise.
26129 (fvar-tracking-assignments-toggle): Likewise.
26130 (fvar-tracking-uninit): Likewise.
26132 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26134 PR translation/79018
26135 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26138 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26141 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26142 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26143 TARGET_LONG_JUMP_TABLE_OFFSETS.
26144 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26145 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26146 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26147 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26148 * config/m68k/m68k.md (tablejump expander): Likewise.
26149 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26150 TARGET_LONG_JUMP_TABLE_OFFSETS.
26151 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26152 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26154 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26155 David Holsgrove <david.holsgrove@xilinx.com>
26157 * common/config/microblaze/microblaze-common.c
26158 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26159 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26161 * config/microblaze/microblaze.c (microblaze_must_save_register)
26162 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26164 (microblaze_eh_return): New function.
26165 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26166 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26167 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26168 * config/microblaze/microblaze.md (eh_return): New pattern.
26170 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26172 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26173 GCC_DIAGNOSTIC_STRINGIFY): Define.
26175 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26177 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26179 * config/arm/arm.md (<mcrr>): New.
26181 * config/arm/arm.c (arm_arch5te): New.
26182 (arm_option_override): Set arm_arch5te.
26183 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26185 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26186 (arm_mcrr_qualifiers): ... this. New.
26187 (MRRC_QUALIFIERS): Define to...
26188 (arm_mrrc_qualifiers): ... this. New.
26189 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26191 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26192 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26193 (MRRCI, mrrc, MRRC): New.
26194 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26195 VUNSPEC_MRRC2): New.
26197 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26199 * config/arm/arm.md (<mcr>): New.
26201 * config/arm/arm.c (arm_coproc_builtin_available): Add
26202 support for mcr, mrc, mcr2 and mrc2.
26203 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26204 (arm_mcr_qualifiers): ... this. New.
26205 (MRC_QUALIFIERS): Define to ...
26206 (arm_mrc_qualifiers): ... this. New.
26207 (MCR_QUALIFIERS): Define to ...
26208 (arm_mcr_qualifiers): ... this. New.
26209 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26211 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26212 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26213 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26214 VUNSPEC_MRC2): New.
26216 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26218 * config/arm/arm.md (*ldc): New.
26222 * config/arm/arm.c (arm_coproc_builtin_available): Add
26223 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26224 (arm_coproc_ldc_stc_legitimate_address): New.
26225 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26226 'qualifier_const_pointer'.
26227 (LDC_QUALIFIERS): Define to...
26228 (arm_ldc_qualifiers): ... this. New.
26229 (STC_QUALIFIERS): Define to...
26230 (arm_stc_qualifiers): ... this. New.
26231 * config/arm/arm-protos.h
26232 (arm_coproc_ldc_stc_legitimate_address): New.
26233 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26234 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26235 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26236 stc2, stcl, stc2l): New.
26237 * config/arm/constraints.md (Uz): New.
26238 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26239 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26240 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26241 VUNSPEC_STC2L): New.
26243 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26245 * config/arm/arm.md (<cdp>): New.
26246 * config/arm/arm.c (neon_const_bounds): Rename this ...
26247 (arm_const_bounds): ... this.
26248 (arm_coproc_builtin_available): New.
26249 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26250 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26251 (CDP_QUALIFIERS): Define to...
26252 (arm_cdp_qualifiers): ... this. New.
26254 (arm_expand_builtin_args): Add case for 6 arguments.
26255 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26256 (arm_const_bounds): ... this.
26257 (arm_coproc_builtin_available): New.
26258 * config/arm/arm_acle.h (__arm_cdp): New.
26260 * config/arm/arm_acle_builtins.def (cdp): New.
26262 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26263 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26264 'arm_const_bounds'.
26265 * config/arm/types.md (coproc): New.
26266 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26267 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26268 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26269 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26271 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26273 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26274 (UBINOP_QUALIFIERS): New.
26276 (acle_builtin_data): New. Change comment.
26277 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26278 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26279 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26280 arm_acle_builtins.def.
26281 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26282 (arm_init_acle_builtins): New.
26283 (CRC32_BUILTIN): Remove.
26284 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26285 crc32cb, crc32ch and crc32cw.
26286 (arm_init_crc32_builtins): Remove.
26287 (arm_init_builtins): Use arm_init_acle_builtins rather
26288 than arm_init_crc32_builtins.
26289 (arm_expand_acle_builtin): New.
26290 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26291 * config/arm/arm_acle_builtins.def: New.
26293 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26295 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26296 (arm_builtin_datum): ... this.
26297 (arm_init_neon_builtin): Rename to ...
26298 (arm_init_builtin): ... this. Add a new parameters PREFIX
26299 and USE_SIG_IN_NAME.
26300 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26301 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26302 'arm_builtin_datum'.
26303 (arm_init_vfp_builtins): Likewise.
26304 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26305 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26306 (arm_expand_neon_args): Rename to ...
26307 (arm_expand_builtin_args): ... this. Rename builtin_arg
26308 enum values and differentiate between ARG_BUILTIN_MEMORY
26309 and ARG_BUILTIN_NEON_MEMORY.
26310 (arm_expand_neon_builtin_1): Rename to ...
26311 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26312 values, arm_expand_builtin_args and add bool parameter NEON.
26313 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26314 (arm_expand_vfp_builtin): Likewise.
26315 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26317 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26319 PR middle-end/77484
26320 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26321 * predict.c (tree_estimate_probability_bb): Reverse direction of
26322 polymorphic call predictor.
26324 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26326 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26327 (determine_pass_name_match): ...this new function and...
26328 (should_skip_pass_p): ...this new function.
26330 2017-01-06 Nathan Sidwell <nathan@acm.org>
26332 * ipa-visibility.c (function_and_variable_visibility): Reformat
26333 comments and long lines. Remove extrneous if.
26334 * symtab.c (symtab_node::make_decl_local): Fix code format.
26335 (symtab_node::set_section_for_node): Fix comment typo.
26337 2017-01-06 Martin Liska <mliska@suse.cz>
26340 * lra-constraints.c: Rename invariant to lra_invariant.
26341 * predict.c (set_even_probabilities): Initialize e to NULL.
26343 2017-01-05 Martin Sebor <msebor@redhat.com>
26345 PR tree-optimization/78910
26346 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26347 (format_integer): Correct off-by-one error in the handling
26348 of precision with negative numbers in signed conversions..
26350 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26352 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26354 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26356 PR tree-optimization/71016
26357 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26358 factor_out_conditional_conversion. Formatting fix.
26359 (factor_out_conditional_conversion): Add cond_stmt argument.
26360 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26361 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26364 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26366 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26367 read-rtl-function.o, and selftest-rtl.o.
26368 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26369 (selftest::aarch64_test_loading_full_dump): New function.
26370 (selftest::aarch64_run_selftests): New function.
26371 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26372 selftest::aarch64_run_selftests.
26373 * config/i386/i386.c
26374 (selftest::ix86_test_loading_dump_fragment_1): New function.
26375 (selftest::ix86_test_loading_call_insn): New function.
26376 (selftest::ix86_test_loading_full_dump): New function.
26377 (selftest::ix86_test_loading_unspec): New function.
26378 (selftest::ix86_run_selftests): Call the new functions.
26379 * emit-rtl.c (maybe_set_max_label_num): New function.
26380 * emit-rtl.h (maybe_set_max_label_num): New decl.
26381 * function.c (instantiate_decls): Guard call to
26382 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26383 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26385 * gensupport.c (gen_reader::gen_reader): Pass "false"
26386 for new "compact" param of rtx_reader.
26387 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26388 rather than an empty string for NULL strings.
26389 * read-md.c: Potentially include config.h rather than bconfig.h.
26390 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26391 (have_error): New global, copied from errors.c.
26392 (md_reader::read_name): Rename to...
26393 (md_reader::read_name_1): ...this, adding "out_loc" param,
26394 and converting "missing name or number" to returning false, rather
26396 (md_reader::read_name): Reimplement in terms of read_name_1.
26397 (md_reader::read_name_or_nil): New function.
26398 (md_reader::read_string): Handle "(nil)" by returning NULL.
26399 (md_reader::md_reader): Add new param "compact".
26400 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26401 (md_reader::read_file): New method.
26402 * read-md.h (md_reader::md_reader): Add new param "compact".
26403 (md_reader::read_file): New method.
26404 (md_reader::is_compact): New accessor.
26405 (md_reader::read_name): Convert return type from void to file_location.
26406 (md_reader::read_name_or_nil): New decl.
26407 (md_reader::read_name_1): New decl.
26408 (md_reader::m_compact): New field.
26409 (noop_reader::noop_reader): Pass "false" for new "compact" param
26411 (rtx_reader::rtx_reader): Add new "compact" param.
26412 (rtx_reader::read_rtx_operand): Make virtual and convert return
26413 type from void to rtx.
26414 (rtx_reader::read_until): New decl.
26415 (rtx_reader::handle_any_trailing_information): New virtual function.
26416 (rtx_reader::postprocess): New virtual function.
26417 (rtx_reader::finalize_string): New virtual function.
26418 (rtx_reader::m_in_call_function_usage): New field.
26419 (rtx_reader::m_reuse_rtx_by_id): New field.
26420 * read-rtl-function.c: New file.
26421 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26422 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26423 (selftest::verify_three_block_rtl_cfg): New decl.
26424 * read-rtl-function.h: New file.
26425 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26426 For host, include function.h, memmodel.h, and emit-rtl.h.
26427 (one_time_initialization): New function.
26428 (struct compact_insn_name): New struct.
26429 (compact_insn_names): New array.
26430 (find_code): Handle insn codes in compact dumps.
26431 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26432 (bind_subst_iter_and_attr): Likewise.
26433 (add_condition_to_string): Likewise.
26434 (add_condition_to_rtx): Likewise.
26435 (apply_attribute_uses): Likewise.
26436 (add_current_iterators): Likewise.
26437 (apply_iterators): Likewise.
26438 (initialize_iterators): Guard usage of apply_subst_iterator with
26439 #ifdef GENERATOR_FILE.
26440 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26441 (md_reader::read_mapping): Likewise.
26442 (add_define_attr_for_define_subst): Likewise.
26443 (add_define_subst_attr): Likewise.
26444 (read_subst_mapping): Likewise.
26445 (check_code_iterator): Likewise.
26446 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26448 (one_time_initialization): New function.
26449 (rtx_reader::read_until): New method.
26450 (read_flags): New function.
26451 (parse_reg_note_name): New function.
26452 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26453 Handle reuse_rtx ids.
26454 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26455 Add parsing support for RTL dumps, mirroring the special-cases in
26456 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26457 values, and calling handle_any_trailing_information.
26458 (rtx_reader::read_rtx_operand): Convert return type from void
26459 to rtx, returning return_rtx. Handle case 'e'. Call
26460 finalize_string on XSTR and XTMPL fields.
26461 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26462 "(nil)" values were omitted. Call the postprocess vfunc on the
26464 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26465 class ctor. Initialize m_in_call_function_usage. Call
26466 one_time_initialization.
26467 * rtl-tests.c (selftest::test_uncond_jump): Call
26468 set_new_first_and_last_insn.
26469 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26470 * selftest-rtl.c: New file.
26471 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26472 (selftest::get_insn_by_uid): New decl.
26473 * selftest-run-tests.c (selftest::run_tests): Call
26474 read_rtl_function_c_tests.
26475 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26476 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26479 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26481 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26482 operands in a special way. Assert that pos+len <= mode precision.
26484 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26486 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26487 3 argument Alias with unlimited for the negative form.
26488 (fno-vect-cost-model): Removed.
26490 2017-01-05 Martin Liska <mliska@suse.cz>
26492 * hsa-gen.c (gen_hsa_divmod): New function.
26493 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26495 2017-01-05 Martin Liska <mliska@suse.cz>
26498 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26501 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26503 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26504 small constant length operands.
26506 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26508 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26509 between loop iterations.
26511 2017-01-05 Martin Liska <mliska@suse.cz>
26514 * gimplify.c (gimplify_decl_expr): Compare to
26515 asan_poisoned_variables instread of checking flags.
26516 (gimplify_target_expr): Likewise.
26517 (gimplify_expr): Likewise.
26518 (gimplify_function_tree): Conditionally initialize
26519 asan_poisoned_variables.
26521 2017-01-04 Jeff Law <law@redhat.com>
26523 PR tree-optimizatin/78812
26524 * rtl.h (contains_mem_rtx_p): Prototype.
26525 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26526 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26527 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26528 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26529 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26531 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26533 * input.c (assert_char_at_range): Default-initialize actual_range.
26535 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26537 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26538 to match the caller.
26540 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26542 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26543 insns after final jump in test to emit dummy move.
26545 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26547 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26548 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26550 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26552 * multiple_target.c (create_dispatcher_calls): Init e_next.
26553 * tree-ssa-loop-split.c (split_loop): Init border.
26554 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26557 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26562 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26563 (altivec_register_operand): Do not return true if the operand
26564 contains a SUBREG mixing SImode and SFmode.
26565 (vsx_register_operand): Likewise.
26566 (vsx_reg_sfsubreg_ok): New predicate.
26567 (vfloat_operand): Do not return true if the operand contains a
26568 SUBREG mixing SImode and SFmode.
26569 (vint_operand): Likewise.
26570 (vlogical_operand): Likewise.
26571 (gpc_reg_operand): Likewise.
26572 (int_reg_operand): Likewise.
26573 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26574 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26575 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26577 (rs6000_emit_move_si_sf_subreg): New helper function.
26578 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26579 fixup SUBREGs involving SImode and SFmode.
26580 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26581 numbers for the new peephole2 optimization.
26582 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26583 the GLIBC math library that do AND/IOR/XOR operations on single
26584 precision floating point.
26585 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26586 target macros to say whether we need to avoid SUBREGs mixing
26588 (TARGET_ALLOW_SF_SUBREG): Likewise.
26589 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26590 (UNSPEC_SI_FROM_SF): Likewise.
26591 (iorxor): Change spacing.
26592 (and_ior_xor): New iterator for AND, IOR, and XOR.
26593 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26594 (movdi_from_sf_zero_ext): Likewise.
26595 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26596 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26597 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26598 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26599 (fms<mode>4): Likewise.
26600 (fnma<mode>4): Likewise.
26601 (fnms<mode>4): Likewise.
26602 (nfma<mode>4): Likewise.
26603 (nfms<mode>4): Likewise.
26605 2017-01-04 Marek Polacek <polacek@redhat.com>
26608 * doc/invoke.texi: Document -Wpointer-compare.
26610 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26612 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26615 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26616 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26619 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26622 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26623 documentation of the powerpc_popcntb_ok attribute.
26624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26625 code to issue warning messages if a requested CPU configuration is
26626 not supported by the binary (assembler and loader) toolchain.
26627 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26628 made to define a built-in function that has been disabled.
26629 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26630 made to define a built-in function that has been disabled.
26631 (altivec_init_builtins): Add comment explaining why definition
26632 of the DST built-in functions is not preceded by an assertion
26633 check. Add assertions to prevent ICE if attempts are made to
26634 define an altivec predicate or an abs* built-in function that has
26636 (htm_init_builtins): Add comment explaining why definition of the
26637 htm built-in functions is not preceded by an assertion check.
26639 2017-01-04 Jeff Law <law@redhat.com>
26641 PR tree-optimizatin/67955
26642 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26643 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26644 the points-to solution does not include pt_null. Use DECL_PT_UID
26647 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26649 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26650 Use gen_int_mode instead of gen_lopwart for const_int operands.
26652 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26654 PR tree-optimization/71563
26655 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26656 out of range value - has low bits known to be zero.
26658 2017-01-04 Alan Modra <amodra@gmail.com>
26660 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26661 * configure: Regenerate.
26662 * config.in: Regenerate.
26664 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26667 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26668 a substring of the message, but strcmp with the whole message. Ifdef
26669 ENABLE_NLS, translate the message first using dgettext.
26671 2017-01-03 Jeff Law <law@redhat.com>
26673 PR tree-optimizatin/78856
26674 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26675 (mark_threaded_blocks): Remove code to truncate thread paths that
26676 cross multiple loop headers. Instead invalidate the cached loop
26677 iteration information and handle case of a thread path walking
26678 into an irreducible region.
26680 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26683 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26684 assertions. Add support for doing the signbit if the IEEE 128-bit
26685 floating point value is in a GPR.
26686 * config/rs6000/rs6000.md (Fsignbit): Delete.
26687 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26688 Update the length attribute if the value is in a GPR.
26689 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26690 the sign or zero extension instruction, since the value is always 0/1.
26691 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26694 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26695 extracting SImode to a GPR register so that we can generate a
26696 store, limit the vector to be in a traditional Altivec register
26697 for the vextuwrx instruction.
26699 2017-01-03 Ian Lance Taylor <iant@google.com>
26701 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26703 2017-01-03 Martin Sebor <msebor@redhat.com>
26705 PR tree-optimization/78696
26706 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26707 precision. Use MPFR for %f for greater fidelity. Correct handling
26709 (pass_sprintf_length::compute_format_length): Set width and precision
26710 specified by asrerisk to void_node for vararg functions.
26711 (try_substitute_return_value): Adjust dump output.
26713 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26715 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26717 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26719 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26720 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26721 -mlra/-mno-lra was passed to the compiler.
26723 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26725 PR rtl-optimization/65618
26726 * emit-rtl.c (try_split): Move initialization of "before" and
26727 "after" to just before the call to emit_insn_after_setloc.
26729 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26731 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26733 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26735 * dwarf2out.c (gen_enumeration_type_die): When
26736 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26738 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26740 PR tree-optimization/78965
26741 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26742 Change first argument from const call_info & to call_info &. For %n
26743 set info.nowrite to false.
26745 PR middle-end/78901
26746 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26747 possibly throwing calls.
26749 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26750 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26751 and fns handling, rather than in a separate case SSA_NAME.
26753 2017-01-02 Jeff Law <law@redhat.com>
26755 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26756 fixes for first_period and second_period variables.
26758 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26761 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26762 (*insvqi_1): New insn pattern.
26763 (*insvqi_1_mem_rex64): Ditto.
26764 (*insvqi_2): Ditto.
26765 (*insvqi_3): Rename from *insvqi.
26767 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26769 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26771 * doc/cfg.texi (Edges): Remove reference to Java.
26772 (Maintaining the CFG): Ditto.
26774 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26776 PR middle-end/77674
26777 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26778 transparent aliases.
26780 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26782 PR middle-end/77484
26783 * predict.def (PRED_CALL): Update hitrate.
26784 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26785 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26786 into direct/indirect/polymorphic variants.
26788 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26790 Update copyright years.
26792 * gcc.c (process_command): Update copyright notice dates.
26793 * gcov-dump.c (print_version): Ditto.
26794 * gcov.c (print_version): Ditto.
26795 * gcov-tool.c (print_version): Ditto.
26796 * gengtype.c (create_file): Ditto.
26797 * doc/cpp.texi: Bump @copying's copyright year.
26798 * doc/cppinternals.texi: Ditto.
26799 * doc/gcc.texi: Ditto.
26800 * doc/gccint.texi: Ditto.
26801 * doc/gcov.texi: Ditto.
26802 * doc/install.texi: Ditto.
26803 * doc/invoke.texi: Ditto.
26805 Copyright (C) 2017 Free Software Foundation, Inc.
26807 Copying and distribution of this file, with or without modification,
26808 are permitted in any medium without royalty provided the copyright
26809 notice and this notice are preserved.