1 2017-09-05 Jeff Law <law@redhat.com>
3 PR tree-optimization/64910
4 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5 cases where we have 3 or more operands.
7 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10 * omp-low.c (lower_omp_for): Recompute tree invariant if
11 gimple_omp_for_initial/final is ADDR_EXPR.
14 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15 into gimple val before gimplification fo the COND_EXPR.
17 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
19 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
21 (thread_through_all_blocks): Remove unused argument to
22 duplicate_thread_path.
24 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
25 Alan Hayward <alan.hayward@arm.com>
26 David Sherwood <david.sherwood@arm.com>
28 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
29 Take a scalar_mode rather than a machine_mode.
30 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
31 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
32 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
33 (aarch64_gen_adjusted_ldpstp): Likewise.
34 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
36 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
37 Alan Hayward <alan.hayward@arm.com>
38 David Sherwood <david.sherwood@arm.com>
40 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
41 Take a scalar_int_mode instead of a machine_mode.
42 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
43 (aarch64_output_scalar_simd_mov_immediate): Likewise.
44 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
45 (aarch64_simd_attr_length_rglist): Delete.
46 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
47 a scalar_int_mode instead of a machine_mode.
48 (aarch64_add_offset): Likewise.
49 (aarch64_internal_mov_immediate): Likewise
50 (aarch64_add_constant_internal): Likewise.
51 (aarch64_add_constant): Likewise.
52 (aarch64_movw_imm): Likewise.
53 (aarch64_rtx_arith_op_extract_p): Likewise.
54 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
55 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
56 Remove assert that the mode isn't a vector.
57 (aarch64_output_scalar_simd_mov_immediate): Likewise.
58 (aarch64_expand_mov_immediate): Update calls after above changes.
59 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
60 (aarch64_and_bitmask_imm): Check for scalar integer modes.
61 (aarch64_move_imm): Likewise.
62 (aarch64_can_const_movi_rtx_p): Likewise.
63 (aarch64_strip_extend): Likewise.
64 (aarch64_extr_rtx_p): Likewise.
65 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
66 a CONST_INT when the mode parameter is VOIDmode.
67 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
69 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
71 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
72 * stor-layout.c (bitwise_mode_for_mode): Likewise.
73 (bitwise_type_for_mode): Update accordingly.
75 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
77 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
78 * stor-layout.c (mode_for_size_tree): Likewise.
79 (mode_for_array): Update accordingly.
80 (layout_decl): Likewise.
81 (compute_record_mode): Likewise. Only set the mode once.
83 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
85 * target.def (get_mask_mode): Change return type to opt_mode.
87 * doc/tm.texi: Regenerate.
88 * targhooks.h (default_get_mask_mode): Return an opt_mode.
89 * targhooks.c (default_get_mask_mode): Likewise.
90 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
91 * optabs-query.c (can_vec_mask_load_store_p): Update use of
92 targetm.get_mask_mode.
93 * tree.c (build_truth_vector_type): Likewise.
95 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
97 * machmode.h (mode_for_vector): Return an opt_mode.
98 * stor-layout.c (mode_for_vector): Likewise.
99 (mode_for_int_vector): Update accordingly.
100 (layout_type): Likewise.
101 * config/i386/i386.c (emit_memmov): Likewise.
102 (ix86_expand_set_or_movmem): Likewise.
103 (ix86_expand_vector_init): Likewise.
104 (ix86_get_mask_mode): Likewise.
105 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
107 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
108 * expmed.c (extract_bit_field_1): Likewise.
109 * expr.c (expand_expr_real_2): Likewise.
110 * optabs-query.c (can_vec_perm_p): Likewise.
111 (can_vec_mask_load_store_p): Likewise.
112 * optabs.c (expand_vec_perm): Likewise.
113 * targhooks.c (default_get_mask_mode): Likewise.
114 * tree-vect-stmts.c (vectorizable_store): Likewise.
115 (vectorizable_load): Likewise.
116 (get_vectype_for_scalar_type_and_size): Likewise.
118 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
120 * machmode.h (mode_for_int_vector): New function.
121 * stor-layout.c (mode_for_int_vector): Likewise.
122 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
123 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
124 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
125 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
126 (s390_expand_vcond): Likewise.
128 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
130 * machmode.h (opt_machine_mode): New type.
131 (opt_mode<T>): Allow construction from anything that can be
133 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
134 (mode_for_size): Return an opt_machine_mode.
135 * stor-layout.c (mode_for_size): Likewise.
136 (mode_for_size_tree): Update call accordingly.
137 (bitwise_mode_for_mode): Likewise.
138 (make_fract_type): Likewise.
139 (make_accum_type): Likewise.
140 * caller-save.c (replace_reg_with_saved_mem): Update call
142 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
143 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
144 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
145 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
146 * expmed.c (extract_bit_field_1): Likewise.
147 * reload.c (get_secondary_mem): Likewise.
148 * varasm.c (assemble_integer): Likewise.
149 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
152 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
154 * machmode.h (decimal_float_mode_for_size): New function.
155 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
156 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
157 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
158 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
159 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
160 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
162 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
164 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
165 (get_builtin_sync_mode): Likewise.
166 (expand_ifn_atomic_compare_exchange): Likewise.
167 (expand_builtin_atomic_clear): Likewise.
168 (expand_builtin_atomic_test_and_set): Likewise.
169 (fold_builtin_atomic_always_lock_free): Likewise.
170 * calls.c (compute_argument_addresses): Likewise.
171 (emit_library_call_value_1): Likewise.
172 (store_one_arg): Likewise.
173 * combine.c (combine_instructions): Likewise.
174 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
175 * config/arm/arm.c (arm_function_value): Likewise.
176 (aapcs_allocate_return_reg): Likewise.
177 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
178 * config/i386/i386.c (construct_container): Likewise.
179 (ix86_gimplify_va_arg): Likewise.
180 (ix86_expand_sse_cmp): Likewise.
181 (emit_memmov): Likewise.
182 (emit_memset): Likewise.
183 (expand_small_movmem_or_setmem): Likewise.
184 (ix86_expand_pextr): Likewise.
185 (ix86_expand_pinsr): Likewise.
186 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
187 * config/microblaze/microblaze.c (microblaze_block_move_straight):
189 * config/mips/mips.c (mips_function_value_1) Likewise.
190 (mips_block_move_straight): Likewise.
191 (mips_expand_ins_as_unaligned_store): Likewise.
192 * config/powerpcspe/powerpcspe.c
193 (rs6000_darwin64_record_arg_advance_flush): Likewise.
194 (rs6000_darwin64_record_arg_flush): Likewise.
195 * config/rs6000/rs6000.c
196 (rs6000_darwin64_record_arg_advance_flush): Likewise.
197 (rs6000_darwin64_record_arg_flush): Likewise.
198 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
199 (sparc_function_value_1): Likewise.
200 * config/spu/spu.c (adjust_operand): Likewise.
201 (spu_emit_branch_or_set): Likewise.
202 (arith_immediate_p): Likewise.
203 * emit-rtl.c (gen_lowpart_common): Likewise.
204 * expr.c (expand_expr_real_1): Likewise.
205 * function.c (assign_parm_setup_block): Likewise.
206 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
207 * reload1.c (alter_reg): Likewise.
208 * stor-layout.c (mode_for_vector): Likewise.
209 (layout_type): Likewise.
211 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
213 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
214 (spu_convert_move): Likewise.
215 * lower-subreg.c (resolve_simple_move): Likewise.
217 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
220 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
221 define_insn to a define_expand.
222 (altivec_vsum2sws_direct): New define_insn.
223 (altivec_vsumsws): Convert from a define_insn to a define_expand.
225 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
227 * config/arm/arm.c (arm_option_params_internal): Improve setting of
230 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
235 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
236 realign stack if stack alignment needed is less than incoming
239 2017-09-05 Marek Polacek <polacek@redhat.com>
242 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
245 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
247 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
249 2017-09-05 Richard Biener <rguenther@suse.de>
251 PR tree-optimization/82084
252 * fold-const.c (can_native_encode_string_p): Handle wide characters.
254 2017-09-05 Richard Biener <rguenther@suse.de>
256 PR tree-optimization/82102
257 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
259 2017-09-05 Martin Liska <mliska@suse.cz>
261 PR tree-optimization/82032
262 * tree-cfg.c (generate_range_test): New function.
263 * tree-cfg.h (generate_range_test): Declared here.
264 * tree-cfgcleanup.c (convert_single_case_switch): New function.
265 (cleanup_control_expr_graph): Use it.
266 * tree-switch-conversion.c (try_switch_expansion): Remove
268 (emit_case_nodes): Use generate_range_test.
270 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
273 * config/i386/i386.md (*<btsc><mode>_mask): Add
274 TARGET_USE_BT to insn constraint.
275 (*btr<mode>_mask): Ditto.
277 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
279 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
280 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
282 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
285 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
286 TARGET_NEON and TARGET_IWMMXT.
287 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
288 TARGET_NEON and TARGET_IWMMXT.
289 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
291 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
293 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
294 (ix86_rewrite_tls_address): Ditto.
295 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
296 (ix86_rewrite_tls_address_1): Ditto.
297 (ix86_rewrite_tls_address): Ditto.
298 * config/i386/predicates.md (tls_address_pattern): New predicate.
299 * config/i386/i386.md (TLS address splitter): New splitter.
301 2017-09-04 Richard Biener <rguenther@suse.de>
303 PR tree-optimization/82084
304 * fold-const.h (can_native_encode_string_p): Declare.
305 * fold-const.c (can_native_encode_string_p): Factor out from ...
306 (native_encode_string): ... here.
307 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
308 vectorizing stores from constants we later cannot handle.
310 2017-09-04 Marek Polacek <polacek@redhat.com>
313 * doc/invoke.texi: Update -Wtautological-compare documentation.
315 2017-09-04 Jeff Law <law@redhat.com>
317 PR tree-optimization/64910
318 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
319 swap the first and last operand if the last is a constant.
321 2017-09-04 Marek Polacek <polacek@redhat.com>
324 * convert.c (do_narrow): When sanitizing signed integer overflows,
325 bail out for signed types.
326 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
328 2017-09-04 Richard Biener <rguenther@suse.de>
330 PR tree-optimization/82060
331 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
332 Move devirtualization after stmt folding and before EH/AB/noreturn
333 cleanup to get the stmt refs canonicalized. Use a bool instead
334 of gimple_modified_p since that doesn't work for NOPs. Schedule
335 NOPs generated by folding for removal.
337 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
338 Alan Hayward <alan.hayward@arm.com>
339 David Sherwood <david.sherwood@arm.com>
341 * coretypes.h (pad_direction): New enum.
342 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
343 (FUNCTION_ARG_PADDING): Likewise.
344 * target.def (function_arg_padding): New hook.
345 * targhooks.h (default_function_arg_padding): Declare.
346 * targhooks.c (default_function_arg_padding): New function.
347 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
348 (TARGET_FUNCTION_ARG_PADDING): ...this.
349 * doc/tm.texi: Regenerate.
350 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
351 instead of direction.
352 (compute_argument_addresses): Likewise.
353 (load_register_parameters): Likewise.
354 (emit_library_call_value_1): Likewise.
355 (store_one_arg): Use targetm.calls.function_arg_padding instead
356 of FUNCTION_ARG_PADDING.
357 (must_pass_in_stack_var_size_or_pad): Likewise.
358 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
359 (emit_group_store): Likewise.
360 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
361 instead of FUNCTION_ARG_PADDING.
362 (emit_push_insn): Likewise, and propagate enum change throughout
364 * function.h (direction): Delete.
365 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
367 * function.c (assign_parm_find_stack_rtl): Likewise.
368 (assign_parm_setup_block_p): Likewise.
369 (assign_parm_setup_block): Likewise.
370 (gimplify_parameters): Likewise.
371 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
372 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
374 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
375 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
376 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
377 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
378 (aarch64_function_arg_padding): ...this new function.
379 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
380 (TARGET_FUNCTION_ARG_PADDING): Redefine.
381 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
382 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
383 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
384 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
385 (arm_pad_arg_upward): Replace with...
386 (arm_function_arg_padding): ...this new function.
387 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
389 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
390 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
391 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
392 (ia64_hpux_function_arg_padding): Replace with...
393 (ia64_function_arg_padding): ...this new function. Use pad_direction
394 instead of direction. Check for TARGET_HPUX.
395 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
396 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
397 (iq2000_function_arg_padding): New function.
398 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
399 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
400 (mips_function_arg_padding): ...this new function.
401 (mips_pad_reg_upward): Update accordingly.
402 (TARGET_FUNCTION_ARG_PADDING): Redefine.
403 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
404 targetm.calls.function_arg_padding.
405 (FUNCTION_ARG_PADDING): Delete.
406 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
407 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
408 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
409 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
410 (nios2_block_reg_padding): Return pad_direction instead of direction.
411 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
412 instead of direction.
413 (nios2_function_arg_padding): Likewise. Make static.
414 (TARGET_FUNCTION_ARG_PADDING): Redefine.
415 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
416 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
417 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
418 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
419 (pa_function_arg_padding): Make static. Return pad_direction instead
421 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
422 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
423 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
424 instead of direction. Use targetm.calls.function_arg_padding.
425 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
426 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
427 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
428 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
429 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
431 (function_arg_padding): Rename to...
432 (rs6000_function_arg_padding): ...this. Make static. Return
433 pad_direction instead of direction.
434 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
435 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
436 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
437 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
438 instead of direction. Use targetm.calls.function_arg_padding.
439 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
440 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
441 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
442 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
443 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
444 (function_arg_padding): Rename to...
445 (rs6000_function_arg_padding): ...this. Make static. Return
446 pad_direction instead of direction.
447 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
448 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
449 * config/s390/s390.c (s390_function_arg_padding): New function.
450 (TARGET_FUNCTION_ARG_PADDING): Redefine.
451 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
452 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
453 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
454 (function_arg_padding): Rename to...
455 (sparc_function_arg_padding): ...this. Make static. Return
456 pad_direction instead of direction.
457 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
458 * config/spu/spu.c (spu_function_arg_padding): New function.
459 (TARGET_FUNCTION_ARG_PADDING): Redefine.
460 * system.h (FUNCTION_ARG_PADDING): Poison.
462 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
463 Alan Hayward <alan.hayward@arm.com>
464 David Sherwood <david.sherwood@arm.com>
466 * target.def (modes_tieable_p): New hook.
467 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
468 (TARGET_MODES_TIEABLE_P): ...this.
469 * doc/tm.texi.in: Regenerate.
470 * hooks.h (hook_bool_mode_mode_true): Declare.
471 * hooks.c (hook_bool_mode_mode_true): New function.
472 * combine.c (subst): Use targetm.modes_tieable_p instead of
474 * dse.c (find_shift_sequence): Likewise.
475 * expmed.c (extract_low_bits): Likewise.
476 * lower-subreg.c: Include target.h.
477 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
479 * rtlanal.c (rtx_cost): Likewise.
480 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
481 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
482 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
483 (TARGET_MODES_TIEABLE_P): Redefine.
484 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
485 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
486 (TARGET_MODES_TIEABLE_P): Redefine.
487 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
488 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
489 (arc_modes_tieable_p): New function.
490 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
491 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
492 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
493 (arm_modes_tieable_p): Make static.
494 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
495 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
496 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
497 (TARGET_MODES_TIEABLE_P): Redefine.
498 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
499 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
500 (TARGET_MODES_TIEABLE_P): Redefine.
501 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
502 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
503 (cr16_modes_tieable_p): New function.
504 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
505 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
506 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
507 (TRULY_NOOP_TRUNCATION): Update comment.
508 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
509 (TRULY_NOOP_TRUNCATION): Update comment.
510 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
511 (frv_modes_tieable_p): New function.
512 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
513 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
514 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
515 (TARGET_MODES_TIEABLE_P): Redefine.
516 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
517 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
518 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
519 (TARGET_MODES_TIEABLE_P): Redefine.
520 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
521 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
522 (ia64_modes_tieable_p): New function.
523 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
524 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
525 (iq2000_modes_tieable_p): New function.
526 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
527 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
528 (lm32_modes_tieable_p): New function.
529 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
530 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
531 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
532 (TARGET_MODES_TIEABLE_P): Redefine.
533 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
534 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
535 (m32r_modes_tieable_p): New function.
536 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
537 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
538 (m68k_modes_tieable_p): New function.
539 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
540 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
541 (mcore_modes_tieable_p): New function.
542 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
543 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
545 (TARGET_MODES_TIEABLE_P): Redefine.
546 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
547 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
548 * config/mips/mips.c (mips_modes_tieable_p): Make static.
549 (TARGET_MODES_TIEABLE_P): Redefine.
550 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
551 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
552 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
553 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
554 (mn10300_modes_tieable_p): ...this and make static.
555 (TARGET_MODES_TIEABLE_P): Redefine.
556 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
557 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
558 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
559 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
560 (msp430_modes_tieable_p): Make static.
561 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
562 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
563 (TARGET_MODES_TIEABLE_P): Redefine.
564 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
565 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
566 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
567 (TARGET_MODES_TIEABLE_P): Redefine.
568 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
569 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
570 * config/pa/pa.c (pa_modes_tieable_p): Make static.
571 (TARGET_MODES_TIEABLE_P): Redefine.
572 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
573 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
574 (pdp11_modes_tieable_p): New function.
575 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
576 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
577 (rs6000_modes_tieable_p): New function.
578 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
579 * config/powerpcspe/powerpcspe.md: Update comment.
580 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
581 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
582 (TARGET_MODES_TIEABLE_P): Redefine.
583 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
584 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
585 (rl78_modes_tieable_p): New function.
586 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
587 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
588 (rs6000_modes_tieable_p): New function.
589 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
590 * config/rs6000/rs6000.md: Update comment.
591 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
592 * config/rx/rx.c (rx_modes_tieable_p): New function.
593 (TARGET_MODES_TIEABLE_P): Redefine.
594 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
595 * config/s390/s390.c (s390_modes_tieable_p): New function.
596 (TARGET_MODES_TIEABLE_P): Redefine.
597 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
598 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
599 (sh_modes_tieable_p): New function.
600 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
601 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
602 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
603 (sparc_modes_tieable_p): Make static.
604 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
605 * config/spu/spu.c (spu_modes_tieable_p): New function.
606 (TARGET_MODES_TIEABLE_P): Redefine.
607 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
608 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
609 (TARGET_MODES_TIEABLE_P): Redefine.
610 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
611 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
612 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
613 * config/v850/v850.c (v850_modes_tieable_p): New function.
614 (TARGET_MODES_TIEABLE_P): Redefine.
615 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
616 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
617 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
618 (visium_modes_tieable_p): New function.
619 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
620 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
621 (xtensa_modes_tieable_p): New function.
622 * system.h (MODES_TIEABLE_P): Poison.
624 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
625 Alan Hayward <alan.hayward@arm.com>
626 David Sherwood <david.sherwood@arm.com>
628 * target.def (hard_regno_mode_ok): New hook.
629 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
630 (TARGET_HARD_REGNO_MODE_OK): ...this.
631 * doc/tm.texi.in: Regenerate.
632 * hooks.h (hook_bool_uint_mode_true): Declare.
633 * hooks.c (hook_bool_uint_mode_true): New function.
634 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
636 * genpreds.c (write_insn_preds_c): Add an include of target.h.
637 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
638 instead of HARD_REGNO_MODE_OK.
639 * caller-save.c: Include target.h.
640 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
642 * combine.c (can_combine_p): Likewise.
643 (combinable_i3pat): Likewise.
644 (can_change_dest_mode): Likewise.
645 * expr.c (init_expr_target): Likewise.
646 (convert_move): Likewise.
647 (convert_modes): Likewise.
648 * ira.c (setup_prohibited_class_mode_regs): Likewise.
649 (setup_prohibited_mode_move_regs): Likewise.
650 * ira.h (target_ira): Likewise.
651 * lra-assigns.c (find_hard_regno_for_1): Likewise.
652 * lra-constraints.c (process_alt_operands): Likewise.
653 (split_reg): Likewise.
654 * recog.c (peep2_find_free_register): Likewise.
655 * ree.c (combine_reaching_defs): Likewise.
656 * regcprop.c (maybe_mode_change): Likewise.
657 * reginfo.c (init_reg_sets_1): Likewise.
658 (choose_hard_reg_mode): Likewise.
659 (simplifiable_subregs): Likewise.
660 * regrename.c (check_new_reg_p): Likewise.
661 * reload.c (find_valid_class): Likewise.
662 (find_valid_class_1): Likewise.
663 (reload_inner_reg_of_subreg): Likewise.
664 (push_reload): Likewise.
665 (combine_reloads): Likewise.
666 (find_dummy_reload): Likewise.
667 (find_reloads): Likewise.
668 * reload1.c (find_reg): Likewise.
669 (set_reload_reg): Likewise.
670 (allocate_reload_reg): Likewise.
671 (choose_reload_regs): Likewise.
672 (reload_adjust_reg_for_temp): Likewise.
673 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
674 (simplify_subreg_regno): Likewise.
675 * sel-sched.c (init_regs_for_mode): Likewise.
676 * varasm.c (make_decl_rtl): Likewise.
677 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
678 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
680 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
681 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
682 (TARGET_HARD_REGNO_MODE_OK): Redefine.
683 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
684 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
685 (TARGET_HARD_REGNO_MODE_OK): Redefine.
686 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
687 (arc_mode_class): Delete.
688 (HARD_REGNO_MODE_OK): Delete.
689 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
690 (arc_hard_regno_mode_ok): Rename old array to...
691 (arc_hard_regno_mode_ok_modes): ...this.
692 (arc_conditional_register_usage): Update accordingly.
693 (arc_mode_class): Make static.
694 (arc_hard_regno_mode_ok): New function.
695 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
696 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
697 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
698 (arm_hard_regno_mode_ok): Make static.
699 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
701 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
702 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
703 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
706 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
707 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
708 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
709 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
711 * config/bfin/predicates.md (valid_reg_operand): Use
712 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
713 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
714 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
716 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
717 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
718 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
719 (cr16_hard_regno_mode_ok): Make static and return a bool.
720 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
721 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
722 (cris_hard_regno_mode_ok): New function.
723 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
724 (epiphany_mode_class): Delete.
725 (HARD_REGNO_MODE_OK): Delete.
726 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
727 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
728 (hard_regno_mode_ok): Rename to...
729 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
730 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
731 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
733 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
734 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
735 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
736 (frv_hard_regno_mode_ok): Make static and return a bool.
737 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
739 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
740 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
741 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
742 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
745 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
746 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
747 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
750 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
751 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
752 (ia64_hard_regno_mode_ok): New function.
753 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
754 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
755 (iq2000_hard_regno_mode_ok): New function.
756 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
757 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
758 (lm32_hard_regno_mode_ok): New function.
759 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
760 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
761 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
762 instead of HARD_REGNO_MODE_OK.
763 (m32c_hard_regno_ok): Rename to...
764 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
765 (m32c_cannot_change_mode_class): Update accordingly.
766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
767 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
768 (m32r_mode_class): Delete.
769 (HARD_REGNO_MODE_OK): Delete.
770 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
771 (m32r_hard_regno_mode_ok): Rename to...
772 (m32r_hard_regno_modes): ...this.
773 (m32r_mode_class): Make static.
774 (m32r_hard_regno_mode_ok): New function.
775 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
776 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
777 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
778 (m68k_hard_regno_mode_ok): Make static.
779 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
780 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
781 (mcore_hard_regno_mode_ok): New function.
782 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
783 (HARD_REGNO_MODE_OK): Delete.
784 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
786 (microblaze_hard_regno_mode_ok_p): ...this and make static.
787 (microblaze_hard_regno_mode_ok): New function.
788 (TARGET_HARD_REGNO_MODE_OK): Redefine.
789 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
790 (mips_hard_regno_mode_ok): Delete.
791 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
792 (mips_hard_regno_mode_ok_p): ...this and make static.
793 (mips_hard_regno_mode_ok_p): Rename to...
794 (mips_hard_regno_mode_ok_uncached): ...this.
795 (mips_hard_regno_mode_ok): New function.
796 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
797 of HARD_REGNO_MODE_OK.
798 (mips_option_override): Update after above name changes.
799 (TARGET_HARD_REGNO_MODE_OK): Redefine.
800 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
801 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
802 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
803 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
804 (TARGET_HARD_REGNO_MODE_OK): Redefine.
805 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
806 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
807 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
808 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
809 (msp430_hard_regno_mode_ok): Make static and return a bool.
810 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
811 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
812 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
814 (TARGET_HARD_REGNO_MODE_OK): Redefine.
815 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
816 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
817 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
818 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
819 (PA_HARD_REGNO_MODE_OK): ...this
820 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
821 (PA_HARD_REGNO_MODE_OK): ...this.
822 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
823 (pa_hard_regno_mode_ok): New function.
824 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
825 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
826 (pdp11_hard_regno_mode_ok): New function.
827 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
828 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
830 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
832 (TARGET_HARD_REGNO_MODE_OK): Redefine.
833 (rs6000_hard_regno_mode_ok): Rename to...
834 (rs6000_hard_regno_mode_ok_uncached): ...this.
835 (rs6000_init_hard_regno_mode_ok): Update accordingly.
836 (rs6000_hard_regno_mode_ok): New function.
837 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
838 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
839 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
840 (riscv_hard_regno_mode_ok): ...this and make static.
841 (TARGET_HARD_REGNO_MODE_OK): Redefine.
842 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
843 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
844 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
845 (rl78_hard_regno_mode_ok): Make static and return bool.
846 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
847 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
849 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
851 (rs6000_hard_regno_mode_ok): Rename to...
852 (rs6000_hard_regno_mode_ok_uncached): ...this.
853 (rs6000_init_hard_regno_mode_ok): Update accordingly.
854 (rs6000_hard_regno_mode_ok): New function.
855 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
856 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
857 (TARGET_HARD_REGNO_MODE_OK): Redefine.
858 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
859 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
860 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
862 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
863 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
864 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
865 (sh_hard_regno_mode_ok): Make static.
866 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
867 instead of HARD_REGNO_MODE_OK.
868 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
869 (sparc_mode_class): Delete.
870 (HARD_REGNO_MODE_OK): Delete.
871 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
872 (hard_regno_mode_classes): Make static.
873 (sparc_mode_class): Likewise.
874 (sparc_hard_regno_mode_ok): New function.
875 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
876 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
877 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
880 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
881 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
882 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
883 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
885 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
886 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
887 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
888 (visium_hard_regno_mode_ok): New function.
889 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
890 instead of HARD_REGNO_MODE_OK.
891 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
892 (HARD_REGNO_MODE_OK): Delete.
893 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
894 (xtensa_hard_regno_mode_ok_p): ...this and make static.
895 (xtensa_option_override): Update accordingly.
896 (TARGET_HARD_REGNO_MODE_OK): Redefine.
897 (xtensa_hard_regno_mode_ok): New function.
898 * system.h (HARD_REGNO_MODE_OK): Poison.
900 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
901 Alan Hayward <alan.hayward@arm.com>
902 David Sherwood <david.sherwood@arm.com>
904 * target.def (hard_regno_call_part_clobbered): New hook.
905 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
906 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
907 * doc/tm.texi: Regenerate.
908 * hooks.h (hook_bool_uint_mode_false): Declare.
909 * hooks.c (hook_bool_uint_mode_false): New function.
910 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
911 * cselib.c (cselib_process_insn): Use
912 targetm.hard_regno_call_part_clobbered instead of
913 HARD_REGNO_CALL_PART_CLOBBERED.
914 * ira-conflicts.c (ira_build_conflicts): Likewise.
915 * ira-costs.c (ira_tune_allocno_costs): Likewise.
916 * lra-constraints.c (need_for_call_save_p): Likewise.
917 * lra-lives.c: Include target.h.
918 (check_pseudos_live_through_calls): Use
919 targetm.hard_regno_call_part_clobbered instead of
920 HARD_REGNO_CALL_PART_CLOBBERED.
921 * regcprop.c: Include target.h.
922 (copyprop_hardreg_forward_1): Use
923 targetm.hard_regno_call_part_clobbered instead of
924 HARD_REGNO_CALL_PART_CLOBBERED.
925 * reginfo.c (choose_hard_reg_mode): Likewise.
926 * regrename.c (check_new_reg_p): Likewise.
927 * reload.c (find_equiv_reg): Likewise.
928 * reload1.c (emit_reload_insns): Likewise.
929 * sched-deps.c (deps_analyze_insn): Likewise.
930 * sel-sched.c (init_regs_for_mode): Likewise.
931 (mark_unavailable_hard_regs): Likewise.
932 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
933 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
934 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
936 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
937 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
938 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
940 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
942 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
943 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
944 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
946 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
947 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
948 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
950 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
951 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
953 * config/powerpcspe/powerpcspe.c
954 (rs6000_hard_regno_call_part_clobbered): New function.
955 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
956 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
957 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
959 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
960 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
961 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
963 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
964 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
965 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
967 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
968 Alan Hayward <alan.hayward@arm.com>
969 David Sherwood <david.sherwood@arm.com>
971 * rtl.h (subreg_memory_offset): Declare.
972 * emit-rtl.c (subreg_memory_offset): New function.
973 * expmed.c (store_bit_field_1): Use it.
974 * expr.c (undefined_operand_subword_p): Likewise.
975 * simplify-rtx.c (simplify_subreg): Likewise.
977 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
979 PR rtl-optimization/57448
982 * optabs.c (expand_atomic_load): Place compiler memory barriers if
983 using atomic_load pattern.
984 (expand_atomic_store): Likewise.
986 2017-09-04 Jakub Jelinek <jakub@redhat.com>
989 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
990 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
991 handling. Use replace_call_with_value with NULL instead of
992 gsi_replace, unlink_stmt_vdef and release_defs.
994 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
997 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
999 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1002 * rtl.h (emit_library_call_value_1): Declare.
1003 (emit_library_call): Replace declaration with a series of overloads.
1004 Remove the parameter count argument.
1005 (emit_library_call_value): Likewise.
1006 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1007 with an "rtx_mode_t *".
1008 (emit_library_call_value): Delete.
1009 (emit_library_call): Likewise.
1010 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1011 (asan_emit_allocas_unpoison): Likewise.
1012 * builtins.c (expand_builtin_powi): Likewise.
1013 (expand_asan_emit_allocas_unpoison): Likewise.
1014 * cfgexpand.c (expand_main_function): Likewise.
1015 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1016 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1017 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1018 * config/arm/arm.c (arm_trampoline_init): Likewise.
1019 (arm_call_tls_get_addr): Likewise.
1020 (arm_expand_divmod_libfunc): Likewise.
1021 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1022 (smulsi3_highpart): Likewise.
1023 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1024 (c6x_expand_compare): Likewise.
1025 (c6x_expand_movmem): Likewise.
1026 * config/frv/frv.c (frv_trampoline_init): Likewise.
1027 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1028 (ix86_expand_divmod_libfunc): Likewise.
1029 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1030 (ia64_expand_compare): Likewise.
1031 (ia64_profile_hook): Likewise.
1032 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1033 (nonlocal_goto): Likewise.
1034 (restore_stack_nonlocal): Likewise.
1035 * config/m32r/m32r.c (block_move_call): Likewise.
1036 (m32r_trampoline_init): Likewise.
1037 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1038 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1039 (m68k_call_m68k_read_tp): Likewise.
1040 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1041 (microblaze_expand_divide): Likewise.
1042 * config/mips/mips.h (mips_args): Likewise.
1043 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1044 (MIPS_ICACHE_SYNC): Likewise.
1045 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1046 (nios2_trampoline_init): Likewise.
1047 * config/pa/pa.c (hppa_tls_call): Likewise.
1048 (pa_trampoline_init): Likewise.
1049 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1050 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1051 (expand_strn_compare): Likewise.
1052 (rs6000_generate_compare): Likewise.
1053 (rs6000_expand_float128_convert): Likewise.
1054 (output_profile_hook): Likewise.
1055 (rs6000_trampoline_init): Likewise.
1056 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1057 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1058 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1059 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1060 (rs6000_generate_compare): Likewise.
1061 (rs6000_expand_float128_convert): Likewise.
1062 (output_profile_hook): Likewise.
1063 (rs6000_trampoline_init): Likewise.
1064 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1065 * config/sh/sh.c (sh_trampoline_init): Likewise.
1066 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1067 (sparc_emit_float_lib_cmp): Likewise.
1068 (sparc32_initialize_trampoline): Likewise.
1069 (sparc64_initialize_trampoline): Likewise.
1070 (sparc_profile_hook): Likewise.
1071 * config/spu/spu.c (ea_load_store): Likewise.
1072 * config/spu/spu.md (floatunssidf2): Likewise.
1073 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1074 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1075 * config/visium/visium.c (expand_block_move_4): Likewise.
1076 (expand_block_move_2): Likewise.
1077 (expand_block_move_1): Likewise.
1078 (expand_block_set_4): Likewise.
1079 (expand_block_set_2): Likewise.
1080 (expand_block_set_1): Likewise.
1081 (visium_trampoline_init): Likewise.
1082 (visium_profile_hook): Likewise.
1083 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1084 (xtensa_setup_frame_addresses): Likewise.
1085 (xtensa_trampoline_init): Likewise.
1086 * except.c (sjlj_emit_function_enter): Likewise.
1087 (sjlj_emit_function_exit): Likewise.
1088 * explow.c (allocate_dynamic_stack_space): Likewise.
1089 (probe_stack_range): Likewise.
1090 * expr.c (convert_mode_scalar): Likewise.
1091 * optabs.c (expand_binop): Likewise.
1092 (expand_twoval_binop_libfunc): Likewise.
1093 (expand_unop): Likewise.
1094 (prepare_cmp_insn): Likewise.
1095 (prepare_float_lib_cmp): Likewise.
1096 (expand_float): Likewise.
1097 (expand_fix): Likewise.
1098 (expand_fixed_convert): Likewise.
1099 (maybe_emit_sync_lock_test_and_set): Likewise.
1100 (expand_atomic_compare_and_swap): Likewise.
1101 (expand_mem_thread_fence): Likewise.
1102 (expand_atomic_fetch_op): Likewise.
1104 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1106 * doc/generic.texi (OpenACC): Adjust URL.
1107 * doc/invoke.texi (C Dialect Options): Ditto.
1109 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1111 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1112 predicate for operand 1. Add (m,<S>) constraint.
1113 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1114 Prevent memory operand 1 with register operand 2.
1116 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1118 PR rtl-optimization/82024
1119 * combine.c (try_combine): If the combination result is a PARALLEL,
1120 and we only need to retain the SET in there that would be placed
1121 at I2, check that we can place that at I3 instead, before doing so.
1123 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1126 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1128 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1129 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1132 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1133 Jeff Law <law@redhat.com>
1135 * varasm.c (bss_initializer_p): Do not put constants into .bss
1136 (categorize_decl_for_section): Handle bss_initializer_p returning
1137 false when DECL_INITIAL is NULL.
1139 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1142 * config/s390/s390.c (s390_can_inline_p): New function.
1144 2017-09-01 Jeff Law <law@redhat.com>
1146 PR tree-optimization/82052
1147 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1148 Always initialize the returned slot after a hash table miss
1149 when INSERT is true.
1151 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1153 * config/s390/s390.md (mem_signal_fence): Remove.
1154 * doc/md.texi (mem_signal_fence): Remove.
1155 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1157 * target-insns.def (mem_signal_fence): Remove.
1159 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1162 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1165 * asan.c (create_odr_indicator): Strip name encoding from assembler
1166 name before appending it after __odr_asan_.
1168 2017-09-01 Martin Liska <mliska@suse.cz>
1170 PR tree-optimization/82059
1171 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1172 frequency only when an edge is redirected.
1174 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1176 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1177 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1178 (arc_conditional_register_usage): Remove ARC600 lp_count
1180 (arc_file_start): Emit Tag_ARC_CPU_variation.
1181 (arc_can_use_doloop_p): New conditions to use ZOLs.
1182 (hwloop_fail): New function.
1183 (hwloop_optimize): Likewise.
1184 (hwloop_pattern_reg): Likewise.
1185 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1186 (arc_reorg_loops): New function, calls reorg_loops.
1187 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1188 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1190 (arc_loop_hazard): Remove function, functionality moved into
1192 (arc_hazard): Remove arc_loop_hazard call.
1193 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1194 into hwloop_optimize.
1195 (arc_label_align): Remove ZOL handling.
1196 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1197 * config/arc/arc.md (doloop_begin): Remove pattern.
1198 (doloop_begin_i): Likewise.
1199 (doloop_end_i): Likewise.
1200 (doloop_fallback): Likewise.
1201 (doloop_fallback_m): Likewise.
1202 (doloop_end): Reimplement expand.
1203 (arc_lp): New pattern for LP instruction.
1204 (loop_end): New pattern.
1205 (loop_fail): Likewise.
1206 (decrement_and_branch_until_zero): Likewise.
1207 * config/arc/arc.opt (mlpc-width): New option.
1208 * doc/invoke.texi (mlpc-width): Document option.
1210 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1212 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1213 (arc_ccfsm_advance): Fix checking for delay slots.
1214 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1216 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1218 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1220 (movhi_insn): Update store instruction constraint which are saving
1221 6-bit short immediates.
1222 (movsi_insn): Consider also short scaled load operations.
1223 (zero_extendhisi2_i): Use Usd constraint instead of T.
1224 (extendhisi2_i): Add q constraint.
1225 (arc_clzsi2): Add type and length attributes.
1226 (arc_ctzsi2): Likewise.
1227 * config/arc/constraints.md (Usc): Update constraint, the
1228 assembler can parse two relocations for a single instruction.
1230 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1232 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1233 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1235 2017-08-31 Olivier Hainque <hainque@adacore.com>
1237 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1238 match as powerpc-wrs-vxworks*.
1240 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1242 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1243 register constraint for by-element operand.
1244 (aarch64_mls_elt_merge<mode>): Likewise.
1246 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1248 * config/arc/arc.c (arc_can_follow_jump): Check for short
1251 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1253 * config.gcc: Use g.opt for arc.
1254 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1255 functionality moved to ...
1256 (legitimate_scaled_address_p): New function, ...here.
1257 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1258 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1259 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1261 (arc_override_options): Handle G option.
1262 (arc_output_pic_addr_const): Correct function definition.
1263 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1264 (arc_decl_anon_ns_mem_p): Delete.
1265 (arc_in_small_data_p): Overhaul this function to take into
1266 consideration the value given via G option.
1267 (arc_rewrite_small_data_1): Renamed and corrected old
1268 arc_rewrite_small_data function.
1269 (arc_rewrite_small_data): New function.
1270 (small_data_pattern): Don't use pic_offset_table_rtx.
1271 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1272 * config/arc/simdext.md (movmisalignv2hi): Use
1273 prepare_move_operands function.
1275 (movmisalign*): Likewise.
1276 * doc/invoke.texi (ARC options): Document -G option.
1278 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1280 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1282 * config/arc/arc.c (arc_print_operand): Output scalled address for
1283 sdata whenever is possible.
1284 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1285 load/stores are available.
1286 (compact_sda_memory_operand): Check for the alignment required by
1287 code density instructions.
1288 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1290 * config/arc/constraints.md (Usd): Update constraint.
1291 (Us0): New constraint.
1292 (Usc): Update constraint.
1294 2017-08-31 Richard Biener <rguenther@suse.de>
1297 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1300 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1302 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1303 Resize type_signature.
1305 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1310 subregs whose inner modes can be stored in GPRs.
1311 (aarch64_classify_index): Likewise.
1313 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1314 Alan Hayward <alan.hayward@arm.com>
1315 David Sherwood <david.sherwood@arm.com>
1317 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1318 (V_INT_EQUIV): ...this.
1319 (v_cmp_result): Rename to...
1320 (v_int_equiv): ...this.
1321 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1322 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1323 (copysign<mode>3): Likewise.
1324 (aarch64_simd_bsl<mode>_internal): Likewise.
1325 (aarch64_simd_bsl<mode>): Likewise.
1326 (vec_cmp<mode><mode>): Likewise.
1327 (vcond<mode><mode>): Likewise.
1328 (vcond<v_cmp_mixed><mode>): Likewise.
1329 (vcondu<mode><v_cmp_mixed>): Likewise.
1330 (aarch64_cm<optab><mode>): Likewise.
1331 (aarch64_cmtst<mode>): Likewise.
1332 (aarch64_fac<optab><mode>): Likewise.
1333 (vec_perm_const<mode>): Likewise.
1334 (vcond_mask_<mode><v_cmp_result>): Rename to...
1335 (vcond_mask_<mode><v_int_equiv>): ...this.
1336 (vec_cmp<mode><v_cmp_result>): Rename to...
1337 (vec_cmp<mode><v_int_equiv>): ...this.
1339 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1340 Alan Hayward <alan.hayward@arm.com>
1341 David Sherwood <david.sherwood@arm.com>
1343 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1345 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1346 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1347 (UNSPEC_LD4_DREG): New unspecs.
1348 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1349 (aarch64_ld2<mode>_dreg_be): Replace with...
1350 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1352 (aarch64_ld3<mode>_dreg_le)
1353 (aarch64_ld3<mode>_dreg_be): Replace with...
1354 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1356 (aarch64_ld4<mode>_dreg_le)
1357 (aarch64_ld4<mode>_dreg_be): Replace with...
1358 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1361 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1363 PR tree-optimization/81987
1364 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1365 insert an initializer in a location not dominated by the stride
1368 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1370 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1371 on the entire header of the finally block in the fallthru case.
1373 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1375 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1377 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1379 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1380 rs6000_emit_move_from_cr and call renamed function.
1381 (rs6000_emit_prologue): Call renamed functions.
1382 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1383 movesi_from_cr, remove volatile CRs.
1385 2017-08-30 Jon Beniston <jon@beniston.com>
1386 Richard Biener <rguenther@suse.de>
1388 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1389 of VECTOR_MODE_P check.
1390 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1391 element vector types.
1393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1395 * df.h (df_read_modify_subreg_p): Remove in favor of...
1396 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1397 const_rtx instead of an rtx.
1398 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1399 * df-problems.c (df_word_lr_mark_ref): Likewise.
1400 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1401 (mark_pseudo_reg_dead): Likewise.
1402 (mark_ref_dead): Likewise.
1403 * reginfo.c (init_subregs_of_mode): Likewise.
1404 * sched-deps.c (sched_analyze_1): Likewise.
1405 * df-scan.c (df_def_record_1): Likewise.
1406 (df_uses_record): Likewise.
1407 (df_read_modify_subreg_p): Remove in favor of...
1408 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1409 const_rtx instead of an rtx.
1411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1412 Alan Hayward <alan.hayward@arm.com>
1413 David Sherwood <david.sherwood@arm.com>
1415 * rtl.h (partial_subreg_p): New function.
1416 * caller-save.c (save_call_clobbered_regs): Use it.
1417 * calls.c (expand_call): Likewise.
1418 * combine.c (combinable_i3pat): Likewise.
1419 (simplify_set): Likewise.
1420 (make_extraction): Likewise.
1421 (make_compound_operation_int): Likewise.
1422 (gen_lowpart_or_truncate): Likewise.
1423 (force_to_mode): Likewise.
1424 (make_field_assignment): Likewise.
1425 (reg_truncated_to_mode): Likewise.
1426 (record_truncated_value): Likewise.
1427 (move_deaths): Likewise.
1428 * cse.c (record_jump_cond): Likewise.
1429 (cse_insn): Likewise.
1430 * cselib.c (cselib_lookup_1): Likewise.
1431 * expmed.c (extract_bit_field_using_extv): Likewise.
1432 * function.c (assign_parm_setup_reg): Likewise.
1433 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1434 * ira-build.c (create_insn_allocnos): Likewise.
1435 * lra-coalesce.c (merge_pseudos): Likewise.
1436 * lra-constraints.c (match_reload): Likewise.
1437 (simplify_operand_subreg): Likewise.
1438 (curr_insn_transform): Likewise.
1439 * lra-lives.c (process_bb_lives): Likewise.
1440 * lra.c (new_insn_reg): Likewise.
1441 (lra_substitute_pseudo): Likewise.
1442 * regcprop.c (mode_change_ok): Likewise.
1443 (maybe_mode_change): Likewise.
1444 (copyprop_hardreg_forward_1): Likewise.
1445 * reload.c (push_reload): Likewise.
1446 (find_reloads): Likewise.
1447 (find_reloads_subreg_address): Likewise.
1448 * reload1.c (alter_reg): Likewise.
1449 (eliminate_regs_1): Likewise.
1450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1452 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1454 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1455 back to if statements, including unpack.
1457 2017-08-30 Martin Liska <mliska@suse.cz>
1460 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1462 (func_checker::compare_asm_inputs_outputs): ... this function.
1463 (func_checker::compare_gimple_asm): Use the function to compare
1464 also ASM constrains.
1465 * ipa-icf-gimple.h: Rename the function.
1467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * coretypes.h (complex_mode): New type.
1472 * gdbhooks.py (build_pretty_printer): Handle it.
1473 * machmode.h (complex_mode): New class.
1474 (complex_mode::includes_p): New function.
1475 (is_complex_int_mode): Likewise.
1476 (is_complex_float_mode): Likewise.
1477 * genmodes.c (get_mode_class): Handle complex mode classes.
1478 * function.c (expand_function_end): Use is_complex_int_mode.
1480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * coretypes.h (scalar_mode_pod): New typedef.
1485 * gdbhooks.py (build_pretty_printer): Handle it.
1486 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1487 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1488 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1489 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1493 Alan Hayward <alan.hayward@arm.com>
1494 David Sherwood <david.sherwood@arm.com>
1496 * machmode.h (mode_for_vector): Take a scalar_mode instead
1498 * stor-layout.c (mode_for_vector): Likewise.
1499 * explow.c (promote_mode): Use as_a <scalar_mode>.
1500 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1503 Alan Hayward <alan.hayward@arm.com>
1504 David Sherwood <david.sherwood@arm.com>
1506 * target.def (preferred_simd_mode): Take a scalar_mode
1507 instead of a machine_mode.
1508 * targhooks.h (default_preferred_simd_mode): Likewise.
1509 * targhooks.c (default_preferred_simd_mode): Likewise.
1510 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1511 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1512 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1513 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1514 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1515 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1516 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1517 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1519 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1520 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1521 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1522 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1523 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1524 * doc/tm.texi: Regenerate.
1525 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1529 Alan Hayward <alan.hayward@arm.com>
1530 David Sherwood <david.sherwood@arm.com>
1532 * target.def (scalar_mode_supported_p): Take a scalar_mode
1533 instead of a machine_mode.
1534 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1535 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1536 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1537 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1538 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1539 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1540 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1541 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1542 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1543 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1544 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1545 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1546 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1547 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1549 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1550 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1551 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1552 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1553 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1555 * doc/tm.texi: Regenerate.
1557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1558 Alan Hayward <alan.hayward@arm.com>
1559 David Sherwood <david.sherwood@arm.com>
1561 * coretypes.h (opt_scalar_mode): New typedef.
1562 * gdbhooks.py (build_pretty_printers): Handle it.
1563 * machmode.h (mode_iterator::get_2xwider): Add overload for
1565 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1567 * expr.c (convert_mode_scalar): Likewise.
1568 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1569 * optabs.c (expand_float): Likewise.
1570 (expand_fix): Likewise.
1571 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1574 Alan Hayward <alan.hayward@arm.com>
1575 David Sherwood <david.sherwood@arm.com>
1577 * optabs.c (expand_float): Explicitly check for scalars before
1578 using a branching expansion.
1579 (expand_fix): Likewise.
1581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1582 Alan Hayward <alan.hayward@arm.com>
1583 David Sherwood <david.sherwood@arm.com>
1585 * expr.c (convert_mode): Split scalar handling out into...
1586 (convert_mode_scalar): ...this new function. Treat the modes
1589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1590 Alan Hayward <alan.hayward@arm.com>
1591 David Sherwood <david.sherwood@arm.com>
1593 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1595 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1598 Alan Hayward <alan.hayward@arm.com>
1599 David Sherwood <david.sherwood@arm.com>
1601 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1602 rather than a machine_mode.
1603 (fixed_from_string): Likewise.
1604 (fixed_convert): Likewise.
1605 (fixed_convert_from_int): Likewise.
1606 (fixed_convert_from_real): Likewise.
1607 (real_convert_from_fixed): Likewise.
1608 * fixed-value.c (fixed_from_double_int): Likewise.
1609 (fixed_from_string): Likewise.
1610 (fixed_convert): Likewise.
1611 (fixed_convert_from_int): Likewise.
1612 (fixed_convert_from_real): Likewise.
1613 (real_convert_from_fixed): Likewise.
1614 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1617 Alan Hayward <alan.hayward@arm.com>
1618 David Sherwood <david.sherwood@arm.com>
1620 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1621 of separate mode class checks. Do not allow vector modes here.
1622 (immed_wide_int_const): Use as_a <scalar_mode>.
1623 * explow.c (trunc_int_for_mode): Likewise.
1624 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1625 (wi::shwi): Likewise.
1626 (wi::min_value): Likewise.
1627 (wi::max_value): Likewise.
1628 * dwarf2out.c (loc_descriptor): Likewise.
1629 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1633 Alan Hayward <alan.hayward@arm.com>
1634 David Sherwood <david.sherwood@arm.com>
1636 * tree.h (SCALAR_TYPE_MODE): New macro.
1637 * expr.c (expand_expr_addr_expr_1): Use it.
1638 (expand_expr_real_2): Likewise.
1639 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1640 (fold_convert_const_fixed_from_int): Likewise.
1641 (fold_convert_const_fixed_from_real): Likewise.
1642 (native_encode_fixed): Likewise
1643 (native_encode_complex): Likewise
1644 (native_encode_vector): Likewise.
1645 (native_interpret_fixed): Likewise.
1646 (native_interpret_real): Likewise.
1647 (native_interpret_complex): Likewise.
1648 (native_interpret_vector): Likewise.
1649 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1650 (simd_clone_adjust_argument_types): Likewise.
1651 (simd_clone_init_simd_arrays): Likewise.
1652 (simd_clone_adjust): Likewise.
1653 * stor-layout.c (layout_type): Likewise.
1654 * tree.c (build_minus_one_cst): Likewise.
1655 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1656 * tree-inline.c (estimate_move_cost): Likewise.
1657 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1658 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1659 (vectorizable_reduction): Likewise.
1660 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1661 (vect_recog_mixed_size_cond_pattern): Likewise.
1662 (check_bool_pattern): Likewise.
1663 (adjust_bool_pattern): Likewise.
1664 (search_type_for_mask_1): Likewise.
1665 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1666 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1667 (vectorizable_load): Likewise.
1668 (vectorizable_store): Likewise.
1669 * ubsan.c (ubsan_encode_value): Likewise.
1670 * varasm.c (output_constant): Likewise.
1672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1673 Alan Hayward <alan.hayward@arm.com>
1674 David Sherwood <david.sherwood@arm.com>
1676 * coretypes.h (scalar_mode): New class.
1677 * machmode.h (scalar_mode): Likewise.
1678 (scalar_mode::includes_p): New function.
1679 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1680 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1681 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1682 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1683 * expmed.c (store_bit_field_1): Likewise.
1684 (extract_bit_field_1): Likewise.
1685 * expr.c (write_complex_part): Likewise.
1686 (read_complex_part): Likewise.
1687 (emit_move_complex_push): Likewise.
1688 (expand_expr_real_2): Likewise.
1689 * function.c (assign_parm_setup_reg): Likewise.
1690 (assign_parms_unsplit_complex): Likewise.
1691 * optabs.c (expand_binop): Likewise.
1692 * rtlanal.c (subreg_get_info): Likewise.
1693 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1694 * varasm.c (output_constant_pool_2): Likewise.
1696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1697 Alan Hayward <alan.hayward@arm.com>
1698 David Sherwood <david.sherwood@arm.com>
1700 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1702 (expmed_mult_highpart_optab): Likewise.
1703 (expmed_mult_highpart): Likewise.
1705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1706 Alan Hayward <alan.hayward@arm.com>
1707 David Sherwood <david.sherwood@arm.com>
1709 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1710 instead of a machine_mode.
1711 (builtin_memset_read_str): Likewise.
1712 * builtins.c (c_readstr): Likewise.
1713 (builtin_memcpy_read_str): Likewise.
1714 (builtin_strncpy_read_str): Likewise.
1715 (builtin_memset_read_str): Likewise.
1716 (builtin_memset_gen_str): Likewise.
1717 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1718 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1719 instead of a machine_mode.
1720 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1722 (make_extraction): Likewise.
1723 (try_widen_shift_mode): Take and return scalar_int_modes instead
1725 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1726 a scalar_int_mode instead of a machine_mode.
1727 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1728 (avr_addr_space_pointer_mode): Likewise.
1729 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1730 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1731 (msp430_unwind_word_mode): Likewise.
1732 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1733 (spu_addr_space_pointer_mode): Likewise.
1734 (spu_addr_space_address_mode): Likewise.
1735 (spu_libgcc_cmp_return_mode): Likewise.
1736 (spu_libgcc_shift_count_mode): Likewise.
1737 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1738 (rl78_addr_space_pointer_mode): Likewise.
1739 (fl78_unwind_word_mode): Likewise.
1740 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1742 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1743 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1744 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1745 (mips_valid_pointer_mode): Likewise.
1746 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1747 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1748 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1750 (ft32_addr_space_address_mode): Likewise.
1751 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1752 scalar_int_mode instead of a machine_mode.
1753 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1755 (m32c_addr_space_address_mode): Likewise.
1756 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1757 (rs6000_eh_return_filter_mode): Likewise.
1758 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1759 (rs6000_eh_return_filter_mode): Likewise.
1760 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1761 (s390_libgcc_shift_count_mode): Likewise.
1762 (s390_unwind_word_mode): Likewise.
1763 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1765 * target.def (mode_rep_extended): Likewise.
1766 (valid_pointer_mode): Likewise.
1767 (addr_space.valid_pointer_mode): Likewise.
1768 (eh_return_filter_mode): Return a scalar_int_mode rather than
1770 (libgcc_cmp_return_mode): Likewise.
1771 (libgcc_shift_count_mode): Likewise.
1772 (unwind_word_mode): Likewise.
1773 (addr_space.pointer_mode): Likewise.
1774 (addr_space.address_mode): Likewise.
1775 * doc/tm.texi: Regenerate.
1776 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1778 (do_jump): Use scalar_int_mode for local variables.
1779 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1780 rather than a machine_mode.
1781 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1782 (scompare_loc_descriptor_wide): Likewise.
1783 (scompare_loc_descriptor_narrow): Likewise.
1784 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1786 * except.c (sjlj_emit_dispatch_table): Likewise.
1787 (expand_builtin_eh_copy_values): Likewise.
1788 * explow.c (convert_memory_address_addr_space_1): Likewise.
1789 Take a scalar_int_mode rather than a machine_mode.
1790 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1791 than a machine_mode.
1792 (memory_address_addr_space): Use scalar_int_mode for local variables.
1793 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1794 rather than a machine_mode.
1795 * expmed.c (mask_rtx): Likewise.
1796 (init_expmed_one_conv): Likewise.
1797 (expand_mult_highpart_adjust): Likewise.
1798 (extract_high_half): Likewise.
1799 (expmed_mult_highpart_optab): Likewise.
1800 (expmed_mult_highpart): Likewise.
1801 (expand_smod_pow2): Likewise.
1802 (expand_sdiv_pow2): Likewise.
1803 (emit_store_flag_int): Likewise.
1804 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1806 (extract_low_bits): Likewise.
1807 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1809 * expr.c (pieces_addr::adjust): Likewise.
1810 (can_store_by_pieces): Likewise.
1811 (store_by_pieces): Likewise.
1812 (clear_by_pieces_1): Likewise.
1813 (expand_expr_addr_expr_1): Likewise.
1814 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1815 (expand_expr_real_1): Likewise.
1816 (try_casesi): Likewise.
1817 * final.c (shorten_branches): Likewise.
1818 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1819 type of "mode" to machine_mode.
1820 * internal-fn.c (expand_arith_overflow_result_store): Take a
1821 scalar_int_mode rather than a machine_mode.
1822 (expand_mul_overflow): Use scalar_int_mode for local variables.
1823 * loop-doloop.c (doloop_modify): Likewise.
1824 (doloop_optimize): Likewise.
1825 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1826 than a machine_mode.
1827 (expand_doubleword_shift_condmove): Likewise.
1828 (expand_doubleword_shift): Likewise.
1829 (expand_doubleword_clz): Likewise.
1830 (expand_doubleword_popcount): Likewise.
1831 (expand_doubleword_parity): Likewise.
1832 (expand_absneg_bit): Use scalar_int_mode for local variables.
1833 (prepare_float_lib_cmp): Likewise.
1834 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1835 rather than a machine_mode.
1836 (convert_memory_address_addr_space): Likewise.
1837 (get_mode_bounds): Likewise.
1838 (get_address_mode): Return a scalar_int_mode rather than a
1840 * rtlanal.c (get_address_mode): Likewise.
1841 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1842 than a machine_mode.
1843 * targhooks.c (default_mode_rep_extended): Likewise.
1844 (default_valid_pointer_mode): Likewise.
1845 (default_addr_space_valid_pointer_mode): Likewise.
1846 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1847 than a machine_mode.
1848 (default_libgcc_cmp_return_mode): Likewise.
1849 (default_libgcc_shift_count_mode): Likewise.
1850 (default_unwind_word_mode): Likewise.
1851 (default_addr_space_pointer_mode): Likewise.
1852 (default_addr_space_address_mode): Likewise.
1853 * targhooks.h (default_eh_return_filter_mode): Likewise.
1854 (default_libgcc_cmp_return_mode): Likewise.
1855 (default_libgcc_shift_count_mode): Likewise.
1856 (default_unwind_word_mode): Likewise.
1857 (default_addr_space_pointer_mode): Likewise.
1858 (default_addr_space_address_mode): Likewise.
1859 (default_mode_rep_extended): Take a scalar_int_mode rather than
1861 (default_valid_pointer_mode): Likewise.
1862 (default_addr_space_valid_pointer_mode): Likewise.
1863 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1865 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
1866 rather than a machine_mode.
1867 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
1868 for local variables.
1869 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1870 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
1871 than a machine_mode.
1873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1874 Alan Hayward <alan.hayward@arm.com>
1875 David Sherwood <david.sherwood@arm.com>
1877 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
1878 the mode argument to scalar_int_mode.
1879 (do_jump_by_parts_zero_rtx): Likewise.
1880 (do_jump_by_parts_equality_rtx): Likewise.
1881 (do_jump_by_parts_greater): Take a mode argument.
1882 (do_jump_by_parts_equality): Likewise.
1883 (do_jump_1): Update calls accordingly.
1885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1886 Alan Hayward <alan.hayward@arm.com>
1887 David Sherwood <david.sherwood@arm.com>
1889 * is-a.h (safe_dyn_cast): New function.
1890 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
1891 (jump_table_for_label): Likewise.
1892 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
1893 instead of an rtx_insn *.
1894 (shorten_branches): Use dyn_cast instead of LABEL_P and
1895 JUMP_TABLE_DATA_P. Use jump_table_for_label and
1896 rtx_jump_table_data::get_data_mode.
1897 (final_scan_insn): Likewise.
1899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1900 Alan Hayward <alan.hayward@arm.com>
1901 David Sherwood <david.sherwood@arm.com>
1903 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1904 trying to combine a full-register integer set with a subreg
1907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1908 Alan Hayward <alan.hayward@arm.com>
1909 David Sherwood <david.sherwood@arm.com>
1911 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1912 that is always either address_mode or pointer_mode.
1914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1915 Alan Hayward <alan.hayward@arm.com>
1916 David Sherwood <david.sherwood@arm.com>
1918 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1919 when the two are known to be equal.
1921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1922 Alan Hayward <alan.hayward@arm.com>
1923 David Sherwood <david.sherwood@arm.com>
1925 * simplify-rtx.c (simplify_const_unary_operation): Use
1926 is_a <scalar_int_mode> instead of checking for a nonzero
1927 precision. Forcibly convert op_mode to a scalar_int_mode
1928 in that case. More clearly differentiate the operand and
1929 result modes and use the former when deciding what the value
1930 of a count-bits operation should be. Use is_int_mode instead
1931 of checking for a MODE_INT. Remove redundant check for whether
1932 this mode has a zero precision.
1934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1935 Alan Hayward <alan.hayward@arm.com>
1936 David Sherwood <david.sherwood@arm.com>
1938 * optabs.c (widen_leading): Change the type of the mode argument
1939 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1940 (widen_bswap): Likewise.
1941 (expand_parity): Likewise.
1942 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1943 (expand_ffs): Likewise.
1944 (epand_unop): Check for scalar integer modes before calling the
1947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1948 Alan Hayward <alan.hayward@arm.com>
1949 David Sherwood <david.sherwood@arm.com>
1951 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1953 (expand_expr_real_1): Update call accordingly.
1955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1956 Alan Hayward <alan.hayward@arm.com>
1957 David Sherwood <david.sherwood@arm.com>
1959 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1960 value_mode arguments. Use scalar_int_mode internally.
1961 (store_bit_field_1): Rename the new integer mode from imode
1962 to op0_mode and use it instead of GET_MODE (op0). Update calls
1963 to store_split_bit_field, store_bit_field_using_insv and
1964 store_fixed_bit_field.
1965 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1966 Use scalar_int_mode internally. Use a bit count rather than a mode
1967 when calculating the largest bit size for get_best_mode.
1968 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1969 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1970 Remove assertion that OP0 has a scalar integer mode.
1971 (store_split_bit_field): Add op0_mode and value_mode arguments.
1972 Update calls to extract_fixed_bit_field.
1973 (extract_bit_field_using_extv): Add an op0_mode argument.
1974 Use scalar_int_mode internally.
1975 (extract_bit_field_1): Rename the new integer mode from imode to
1976 op0_mode and use it instead of GET_MODE (op0). Update calls to
1977 extract_split_bit_field, extract_bit_field_using_extv and
1978 extract_fixed_bit_field.
1979 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1980 to extract_split_bit_field and extract_fixed_bit_field_1.
1981 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1982 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1984 (extract_split_bit_field): Add an op0_mode argument. Update call
1985 to extract_fixed_bit_field.
1987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1988 Alan Hayward <alan.hayward@arm.com>
1989 David Sherwood <david.sherwood@arm.com>
1991 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1992 * explow.c (hard_function_value): Likewise.
1993 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1994 convert_to_mode call outside the loop.
1995 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1996 for the mode iterator. Require the mode specified by max_pieces
1998 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2000 (copy_blkmode_to_reg): Likewise.
2001 (set_storage_via_setmem): Likewise.
2002 * optabs.c (prepare_cmp_insn): Likewise.
2003 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2004 * stor-layout.c (finish_bitfield_representative): Likewise.
2006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2007 Alan Hayward <alan.hayward@arm.com>
2008 David Sherwood <david.sherwood@arm.com>
2010 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2011 * expr.c (convert_move): Use them.
2012 (convert_modes): Likewise.
2013 (store_expr_with_bounds): Likewise.
2015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2016 Alan Hayward <alan.hayward@arm.com>
2017 David Sherwood <david.sherwood@arm.com>
2019 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2020 parameter for the mode of "x". Remove the "known_x", "known_mode"
2021 and "known_ret" arguments. Change the type of the mode argument
2023 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2024 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2025 (reg_num_sign_bit_copies_for_combine): Likewise.
2026 * rtlanal.c (nonzero_bits1): Likewise.
2027 (num_sign_bit_copies1): Likewise.
2028 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2029 (reg_num_sign_bit_copies_general): Likewise.
2030 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2031 (reg_nonzero_bits_general): Likewise.
2033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2034 Alan Hayward <alan.hayward@arm.com>
2035 David Sherwood <david.sherwood@arm.com>
2037 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2038 than in subroutines. Return 1 for non-integer modes.
2039 (cached_num_sign_bit_copies): Change the type of the mode parameter
2041 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2042 classes. Handle CONST_INT_P first and then check whether X also
2043 has a scalar integer mode. Check the same thing for inner registers
2044 of a SUBREG and for values that are being extended or truncated.
2046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2047 Alan Hayward <alan.hayward@arm.com>
2048 David Sherwood <david.sherwood@arm.com>
2050 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2051 in subroutines. Return the mode mask for non-integer modes.
2052 (cached_nonzero_bits): Change the type of the mode parameter
2054 (nonzero_bits1): Likewise. Remove early exit for other mode
2055 classes. Handle CONST_INT_P first and then check whether X
2056 also has a scalar integer mode.
2058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2059 Alan Hayward <alan.hayward@arm.com>
2060 David Sherwood <david.sherwood@arm.com>
2062 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2063 Return a scalar_int_mode and assert that the size is greater than
2065 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2066 (op_by_pieces_d::op_by_pieces_d): Likewise.
2067 (op_by_pieces_d::run): Likewise.
2068 (can_store_by_pieces): Likewise.
2070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2071 Alan Hayward <alan.hayward@arm.com>
2072 David Sherwood <david.sherwood@arm.com>
2074 * combine.c (extract_left_shift): Add a mode argument and update
2076 (make_compound_operation_int): Change the type of the mode parameter
2077 to scalar_int_mode and update the call to extract_left_shift.
2079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2080 Alan Hayward <alan.hayward@arm.com>
2081 David Sherwood <david.sherwood@arm.com>
2083 * combine.c (simplify_and_const_int): Change the type of the mode
2084 parameter to scalar_int_mode.
2085 (simplify_and_const_int_1): Likewise. Update recursive call.
2087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2088 Alan Hayward <alan.hayward@arm.com>
2089 David Sherwood <david.sherwood@arm.com>
2091 * combine.c (simplify_compare_const): Check that the mode is a
2092 scalar_int_mode (rather than VOIDmode) before testing its
2094 (simplify_comparison): Move COMPARISON_P handling out of the
2095 loop and restrict the latter part of the loop to scalar_int_modes.
2096 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2097 and when considering SUBREG_REGs. Use is_int_mode instead of
2098 checking GET_MODE_CLASS against MODE_INT.
2100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2101 Alan Hayward <alan.hayward@arm.com>
2102 David Sherwood <david.sherwood@arm.com>
2104 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2105 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2106 shift_unit_mode and for modes involved in scalar shifts.
2108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2109 Alan Hayward <alan.hayward@arm.com>
2110 David Sherwood <david.sherwood@arm.com>
2112 * combine.c (force_int_to_mode): New function, split out from...
2113 (force_to_mode): ...here. Keep xmode up-to-date and use it
2114 instead of GET_MODE (x).
2116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2117 Alan Hayward <alan.hayward@arm.com>
2118 David Sherwood <david.sherwood@arm.com>
2120 * optabs-query.h (extraction_insn::struct_mode): Change type to
2121 opt_scalar_int_mode and update comment.
2122 (extraction_insn::field_mode): Change type to scalar_int_mode.
2123 (extraction_insn::pos_mode): Likewise.
2124 * combine.c (make_extraction): Update accordingly.
2125 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2126 (get_optab_extraction_insn): Likewise.
2127 * recog.c (simplify_while_replacing): Likewise.
2128 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2129 parameter to opt_scalar_int_mode.
2131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2132 Alan Hayward <alan.hayward@arm.com>
2133 David Sherwood <david.sherwood@arm.com>
2135 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2136 to a scalar_int_mode instead of a machine_mode.
2137 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2138 (get_best_mode): Return a boolean and use a pointer argument to store
2139 the selected mode. Replace the limit mode parameter with a bit limit.
2140 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2141 for the values returned by bit_field_mode_iterator::next_mode.
2142 (store_bit_field): Update call to get_best_mode.
2143 (store_fixed_bit_field): Likewise.
2144 (extract_fixed_bit_field): Likewise.
2145 * expr.c (optimize_bitfield_assignment_op): Likewise.
2146 * fold-const.c (optimize_bit_field_compare): Likewise.
2147 (fold_truth_andor_1): Likewise.
2148 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2149 Update for new type of m_mode.
2150 (get_best_mode): As above.
2152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2153 Alan Hayward <alan.hayward@arm.com>
2154 David Sherwood <david.sherwood@arm.com>
2156 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2157 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2158 (store_bit_field): Check is_a <scalar_int_mode> before calling
2159 strict_volatile_bitfield_p.
2160 (extract_bit_field): Likewise.
2162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2163 Alan Hayward <alan.hayward@arm.com>
2164 David Sherwood <david.sherwood@arm.com>
2166 * target.def (cstore_mode): Return a scalar_int_mode.
2167 * doc/tm.texi: Regenerate.
2168 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2169 * targhooks.h (default_cstore_mode): Likewise.
2170 * targhooks.c (default_cstore_mode): Likewise, using a forced
2172 * expmed.c (emit_cstore): Expect the target of the cstore to be
2175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2176 Alan Hayward <alan.hayward@arm.com>
2177 David Sherwood <david.sherwood@arm.com>
2179 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2181 (niter_desc): Likewise mode.
2182 (iv_analyze): Add a mode parameter.
2184 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2185 and change its type to scalar_int_mode.
2186 * loop-iv.c: Update commentary at head of file.
2187 (iv_constant): Pass the mode paraeter before the rtx it describes
2188 and change its type to scalar_int_mode. Remove VOIDmode handling.
2189 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2190 (iv_extend): Likewise.
2191 (shorten_into_mode): Likewise.
2192 (iv_add): Use scalar_int_mode.
2193 (iv_mult): Likewise.
2194 (iv_shift): Likewise.
2195 (canonicalize_iv_subregs): Likewise.
2196 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2197 it describes and change its mode to scalar_int_mode. Also change
2198 the type of the returned inner_mode to scalar_int_mode.
2199 (get_biv_step): Likewise, turning outer_mode from a pointer
2200 into a direct parameter. Update call to get_biv_step_1.
2201 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2202 iv_constant and get_biv_step.
2203 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2204 and change its type to scalar_int_mode. Don't initialise iv->mode
2205 to VOIDmode and remove later checks for its still being VOIDmode.
2206 Update calls to iv_analyze_op and iv_analyze_expr. Check
2207 is_a <scalar_int_mode> when changing the mode under consideration.
2208 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2209 Update call to iv_analyze_expr.
2210 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2211 inner register is not also a scalar_int_mode. Update call to
2213 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2214 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2215 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2216 separate mode class checks. Update calls to iv_analyze. Remove
2217 fix-up of VOIDmodes after iv_analyze_biv.
2218 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2219 don't have a scalar_int_mode. Update call to biv_p.
2221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2222 Alan Hayward <alan.hayward@arm.com>
2223 David Sherwood <david.sherwood@arm.com>
2225 * cfgexpand.c (convert_debug_memory_address): Use
2226 as_a <scalar_int_mode>.
2227 * combine.c (expand_compound_operation): Likewise.
2228 (make_extraction): Likewise.
2229 (change_zero_ext): Likewise.
2230 (simplify_comparison): Likewise.
2231 * cse.c (cse_insn): Likewise.
2232 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2233 (mem_loc_descriptor): Likewise.
2234 (loc_descriptor): Likewise.
2235 * expmed.c (init_expmed_one_mode): Likewise.
2236 (synth_mult): Likewise.
2237 (emit_store_flag_1): Likewise.
2238 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2239 of a comparison with size.
2240 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2241 (reduce_to_bit_field_precision): Likewise.
2242 * function.c (expand_function_end): Likewise.
2243 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2244 * loop-doloop.c (doloop_modify): Likewise.
2245 * optabs.c (expand_binop): Likewise.
2246 (expand_unop): Likewise.
2247 (expand_copysign_absneg): Likewise.
2248 (prepare_cmp_insn): Likewise.
2249 (maybe_legitimize_operand): Likewise.
2250 * recog.c (const_scalar_int_operand): Likewise.
2251 * rtlanal.c (get_address_mode): Likewise.
2252 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2253 (simplify_cond_clz_ctz): Likewise.
2254 * tree-nested.c (get_nl_goto_field): Likewise.
2255 * tree.c (build_vector_type_for_mode): Likewise.
2256 * var-tracking.c (use_narrower_mode): Likewise.
2258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2259 Alan Hayward <alan.hayward@arm.com>
2260 David Sherwood <david.sherwood@arm.com>
2262 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2263 * builtins.c (expand_builtin_signbit): Use it.
2264 * cfgexpand.c (expand_debug_expr): Likewise.
2265 * dojump.c (do_jump): Likewise.
2266 (do_compare_and_jump): Likewise.
2267 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2268 * expmed.c (make_tree): Likewise.
2269 * expr.c (expand_expr_real_2): Likewise.
2270 (expand_expr_real_1): Likewise.
2271 (try_casesi): Likewise.
2272 * fold-const-call.c (fold_const_call_ss): Likewise.
2273 * fold-const.c (unextend): Likewise.
2274 (extract_muldiv_1): Likewise.
2275 (fold_single_bit_test): Likewise.
2276 (native_encode_int): Likewise.
2277 (native_encode_string): Likewise.
2278 (native_interpret_int): Likewise.
2279 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2280 * internal-fn.c (expand_addsub_overflow): Likewise.
2281 (expand_neg_overflow): Likewise.
2282 (expand_mul_overflow): Likewise.
2283 (expand_arith_overflow): Likewise.
2284 * match.pd: Likewise.
2285 * stor-layout.c (layout_type): Likewise.
2286 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2287 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2288 * tree-ssanames.c (get_range_info): Likewise.
2289 * tree-switch-conversion.c (array_value_type) Likewise.
2290 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2291 (vect_recog_divmod_pattern): Likewise.
2292 (vect_recog_mixed_size_cond_pattern): Likewise.
2293 * tree-vrp.c (extract_range_basic): Likewise.
2294 (simplify_float_conversion_using_ranges): Likewise.
2295 * tree.c (int_fits_type_p): Likewise.
2296 * ubsan.c (instrument_bool_enum_load): Likewise.
2297 * varasm.c (mergeable_string_section): Likewise.
2298 (narrowing_initializer_constant_valid_p): Likewise.
2299 (output_constant): Likewise.
2301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2302 Alan Hayward <alan.hayward@arm.com>
2303 David Sherwood <david.sherwood@arm.com>
2305 * machmode.h (NARROWEST_INT_MODE): New macro.
2306 * expr.c (alignment_for_piecewise_move): Use it instead of
2307 GET_CLASS_NARROWEST_MODE (MODE_INT).
2308 (push_block): Likewise.
2309 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2311 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2314 Alan Hayward <alan.hayward@arm.com>
2315 David Sherwood <david.sherwood@arm.com>
2317 * postreload.c (move2add_valid_value_p): Change the type of the
2318 mode parameter to scalar_int_mode.
2319 (move2add_use_add2_insn): Add a mode parameter and use it instead
2321 (move2add_use_add3_insn): Likewise.
2322 (reload_cse_move2add): Update accordingly.
2324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2325 Alan Hayward <alan.hayward@arm.com>
2326 David Sherwood <david.sherwood@arm.com>
2328 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2330 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2331 * optabs.c (expand_unop): Likewise.
2333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2334 Alan Hayward <alan.hayward@arm.com>
2335 David Sherwood <david.sherwood@arm.com>
2337 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2338 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2339 (popcount_loc_descriptor): Likewise.
2340 (bswap_loc_descriptor): Likewise.
2341 (rotate_loc_descriptor): Likewise.
2342 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2343 calling the functions above.
2345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2346 Alan Hayward <alan.hayward@arm.com>
2347 David Sherwood <david.sherwood@arm.com>
2349 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2351 (try_combine): Likewise.
2352 (simplify_if_then_else): Likewise.
2353 * cse.c (cse_insn): Likewise.
2354 * dwarf2out.c (mem_loc_descriptor): Likewise.
2355 * emit-rtl.c (gen_lowpart_common): Likewise.
2356 * simplify-rtx.c (simplify_truncation): Likewise.
2357 (simplify_binary_operation_1): Likewise.
2358 (simplify_const_relational_operation): Likewise.
2359 (simplify_ternary_operation): Likewise.
2360 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2363 Alan Hayward <alan.hayward@arm.com>
2364 David Sherwood <david.sherwood@arm.com>
2366 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2367 * reload.c (push_reload): Likewise.
2368 (find_reloads): Likewise.
2370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2371 Alan Hayward <alan.hayward@arm.com>
2372 David Sherwood <david.sherwood@arm.com>
2374 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2375 (make_compound_operation_int): Likewise.
2376 (change_zero_ext): Likewise.
2377 * expr.c (convert_move): Likewise.
2378 (convert_modes): Likewise.
2379 * fwprop.c (forward_propagate_subreg): Likewise.
2380 * loop-iv.c (get_biv_step_1): Likewise.
2381 * optabs.c (widen_operand): Likewise.
2382 * postreload.c (move2add_valid_value_p): Likewise.
2383 * recog.c (simplify_while_replacing): Likewise.
2384 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2385 (simplify_binary_operation_1): Likewise. Remove redundant
2386 mode equality check.
2388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2389 Alan Hayward <alan.hayward@arm.com>
2390 David Sherwood <david.sherwood@arm.com>
2392 * combine.c (combine_simplify_rtx): Add checks for
2393 is_a <scalar_int_mode>.
2394 (simplify_if_then_else): Likewise.
2395 (make_field_assignment): Likewise.
2396 (simplify_comparison): Likewise.
2397 * ifcvt.c (noce_try_bitop): Likewise.
2398 * loop-invariant.c (canonicalize_address_mult): Likewise.
2399 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2402 Alan Hayward <alan.hayward@arm.com>
2403 David Sherwood <david.sherwood@arm.com>
2405 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2406 is_a <scalar_int_mode> instead of != BLKmode.
2408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2409 Alan Hayward <alan.hayward@arm.com>
2410 David Sherwood <david.sherwood@arm.com>
2412 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2413 instead of != VOIDmode.
2414 * combine.c (if_then_else_cond): Likewise.
2415 (change_zero_ext): Likewise.
2416 * dwarf2out.c (mem_loc_descriptor): Likewise.
2417 (loc_descriptor): Likewise.
2418 * rtlanal.c (canonicalize_condition): Likewise.
2419 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2422 Alan Hayward <alan.hayward@arm.com>
2423 David Sherwood <david.sherwood@arm.com>
2425 * simplify-rtx.c (simplify_binary_operation_1): Use
2426 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2429 Alan Hayward <alan.hayward@arm.com>
2430 David Sherwood <david.sherwood@arm.com>
2432 * wide-int.h (int_traits<unsigned char>) New class.
2433 (int_traits<unsigned short>) Likewise.
2434 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2435 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2437 * combine.c (set_nonzero_bits_and_sign_copies): Use
2438 is_a <scalar_int_mode>.
2439 (find_split_point): Likewise.
2440 (combine_simplify_rtx): Likewise.
2441 (simplify_logical): Likewise.
2442 (expand_compound_operation): Likewise.
2443 (expand_field_assignment): Likewise.
2444 (make_compound_operation): Likewise.
2445 (extended_count): Likewise.
2446 (change_zero_ext): Likewise.
2447 (simplify_comparison): Likewise.
2448 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2449 (ucompare_loc_descriptor): Likewise.
2450 (minmax_loc_descriptor): Likewise.
2451 (mem_loc_descriptor): Likewise.
2452 (loc_descriptor): Likewise.
2453 * expmed.c (init_expmed_one_mode): Likewise.
2454 * lra-constraints.c (lra_constraint_offset): Likewise.
2455 * optabs.c (prepare_libcall_arg): Likewise.
2456 * postreload.c (move2add_note_store): Likewise.
2457 * reload.c (operands_match_p): Likewise.
2458 * rtl.h (load_extend_op): Likewise.
2459 * rtlhooks.c (gen_lowpart_general): Likewise.
2460 * simplify-rtx.c (simplify_truncation): Likewise.
2461 (simplify_unary_operation_1): Likewise.
2462 (simplify_binary_operation_1): Likewise.
2463 (simplify_const_binary_operation): Likewise.
2464 (simplify_const_relational_operation): Likewise.
2465 (simplify_subreg): Likewise.
2466 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2467 * var-tracking.c (adjust_mems): Likewise.
2468 (prepare_call_arguments): Likewise.
2470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2471 Alan Hayward <alan.hayward@arm.com>
2472 David Sherwood <david.sherwood@arm.com>
2474 * machmode.h (is_int_mode): New fuction.
2475 * combine.c (find_split_point): Use it.
2476 (combine_simplify_rtx): Likewise.
2477 (simplify_if_then_else): Likewise.
2478 (simplify_set): Likewise.
2479 (simplify_shift_const_1): Likewise.
2480 (simplify_comparison): Likewise.
2481 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2482 * cse.c (notreg_cost): Likewise.
2483 (cse_insn): Likewise.
2484 * cselib.c (cselib_lookup_1): Likewise.
2485 * dojump.c (do_jump_1): Likewise.
2486 (do_compare_rtx_and_jump): Likewise.
2487 * dse.c (get_call_args): Likewise.
2488 * dwarf2out.c (rtl_for_decl_init): Likewise.
2489 (native_encode_initializer): Likewise.
2490 * expmed.c (emit_store_flag_1): Likewise.
2491 (emit_store_flag): Likewise.
2492 * expr.c (convert_modes): Likewise.
2493 (store_field): Likewise.
2494 (expand_expr_real_1): Likewise.
2495 * fold-const.c (fold_read_from_constant_string): Likewise.
2496 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2497 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2498 * optabs.c (expand_binop): Likewise.
2499 (expand_unop): Likewise.
2500 (expand_abs_nojump): Likewise.
2501 (expand_one_cmpl_abs_nojump): Likewise.
2502 * simplify-rtx.c (mode_signbit_p): Likewise.
2503 (val_signbit_p): Likewise.
2504 (val_signbit_known_set_p): Likewise.
2505 (val_signbit_known_clear_p): Likewise.
2506 (simplify_relational_operation_1): Likewise.
2507 * tree.c (vector_type_mode): Likewise.
2509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2510 Alan Hayward <alan.hayward@arm.com>
2511 David Sherwood <david.sherwood@arm.com>
2513 * machmode.h (smallest_mode_for_size): Fix formatting.
2514 (smallest_int_mode_for_size): New function.
2515 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2516 instead of smallest_mode_for_size.
2517 * combine.c (make_extraction): Likewise.
2518 * config/arc/arc.c (arc_expand_movmem): Likewise.
2519 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2520 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2521 * config/s390/s390.c (s390_expand_insv): Likewise.
2522 * config/sparc/sparc.c (assign_int_registers): Likewise.
2523 * config/spu/spu.c (spu_function_value): Likewise.
2524 (spu_function_arg): Likewise.
2525 * coverage.c (get_gcov_type): Likewise.
2526 (get_gcov_unsigned_t): Likewise.
2527 * dse.c (find_shift_sequence): Likewise.
2528 * expmed.c (store_bit_field_1): Likewise.
2529 * expr.c (convert_move): Likewise.
2530 (store_field): Likewise.
2531 * internal-fn.c (expand_arith_overflow): Likewise.
2532 * optabs-query.c (get_best_extraction_insn): Likewise.
2533 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2534 * stor-layout.c (layout_type): Likewise.
2535 (initialize_sizetypes): Likewise.
2536 * targhooks.c (default_get_mask_mode): Likewise.
2537 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2540 Alan Hayward <alan.hayward@arm.com>
2541 David Sherwood <david.sherwood@arm.com>
2543 * machmode.h (opt_mode::else_blk): New function.
2544 (int_mode_for_mode): Declare.
2545 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2546 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2548 * cfgexpand.c (expand_debug_expr): Likewise.
2549 * combine.c (gen_lowpart_or_truncate): Likewise.
2550 (gen_lowpart_for_combine): Likewise.
2551 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2552 * config/avr/avr.c (avr_to_int_mode): Likewise.
2553 (avr_out_plus_1): Likewise.
2554 (avr_out_plus): Likewise.
2555 (avr_out_round): Likewise.
2556 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2557 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2558 (s390_expand_vcond): Likewise.
2559 * config/spu/spu.c (spu_split_immediate): Likewise.
2560 (spu_expand_mov): Likewise.
2561 * dse.c (get_stored_val): Likewise.
2562 * expmed.c (store_bit_field_1): Likewise.
2563 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2565 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2566 (extract_low_bits): Likewise.
2567 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2568 handling rather than repeating the check.
2569 (emit_group_store): Likewise.
2570 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2571 * optabs.c (expand_absneg_bit): Likewise.
2572 (expand_copysign_absneg): Likewise.
2573 (expand_copysign_bit): Likewise.
2574 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2575 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2576 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2577 * var-tracking.c (prepare_call_arguments): Likewise.
2578 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2579 int_mode_for_mode instead of mode_for_size.
2580 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2583 Alan Hayward <alan.hayward@arm.com>
2584 David Sherwood <david.sherwood@arm.com>
2586 * machmode.h (int_mode_for_size): New function.
2587 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2588 instead of mode_for_size.
2589 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2591 * combine.c (expand_field_assignment): Use int_mode_for_size
2592 instead of mode_for_size.
2593 (make_extraction): Likewise.
2594 (simplify_shift_const_1): Likewise.
2595 (simplify_comparison): Likewise.
2596 * dojump.c (do_jump): Likewise.
2597 * dwarf2out.c (mem_loc_descriptor): Likewise.
2598 * emit-rtl.c (init_derived_machine_modes): Likewise.
2599 * expmed.c (flip_storage_order): Likewise.
2600 (convert_extracted_bit_field): Likewise.
2601 * expr.c (copy_blkmode_from_reg): Likewise.
2602 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2603 * internal-fn.c (expand_mul_overflow): Likewise.
2604 * lower-subreg.c (simple_move): Likewise.
2605 * optabs-libfuncs.c (init_optabs): Likewise.
2606 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2607 * tree.c (vector_type_mode): Likewise.
2608 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2609 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2610 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2611 * tree-vect-stmts.c (vectorizable_load): Likewise.
2612 (vectorizable_store): Likewise.
2614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2615 Alan Hayward <alan.hayward@arm.com>
2616 David Sherwood <david.sherwood@arm.com>
2618 * coretypes.h (pod_mode): New type.
2619 (scalar_int_mode_pod): New typedef.
2620 * machmode.h (pod_mode): New class.
2621 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2622 * genmodes.c (emit_mode_int_n): Update accordingly.
2623 * lower-subreg.h (target_lower_subreg): Change type to
2624 scalar_int_mode_pod.
2625 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2626 scalar_int_mode_pod.
2628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2629 Alan Hayward <alan.hayward@arm.com>
2630 David Sherwood <david.sherwood@arm.com>
2632 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2633 machine_mode to scalar_int_mode.
2634 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2635 (rs6000_option_override_internal): Remove cast to int.
2636 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2637 machine_mode to scalar_int_mode.
2638 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2639 (rs6000_option_override_internal): Remove cast to int.
2640 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2641 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2643 * config/s390/s390.c (s390_expand_builtin): Likewise.
2644 * coretypes.h (scalar_int_mode): New type.
2645 (opt_scalar_int_mode): New typedef.
2646 * machmode.h (scalar_int_mode): New class.
2647 (scalar_int_mode::includes_p): New function.
2648 (byte_mode): Change type to scalar_int_mode.
2649 (word_mode): Likewise.
2650 (ptr_mode): Likewise.
2651 * emit-rtl.c (byte_mode): Likewise.
2652 (word_mode): Likewise.
2653 (ptr_mode): Likewise.
2654 (init_derived_machine_modes): Update accordingly.
2655 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2656 and MODE_PARTIAL_INT.
2657 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2658 opt_scalar_int_mode.
2660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2661 Alan Hayward <alan.hayward@arm.com>
2662 David Sherwood <david.sherwood@arm.com>
2664 * target.def (libgcc_floating_mode_supported_p): Take a
2666 * doc/tm.texi: Regenerate.
2667 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2669 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2670 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2674 Alan Hayward <alan.hayward@arm.com>
2675 David Sherwood <david.sherwood@arm.com>
2677 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2678 * doc/tm.texi: Regenerate.
2679 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2680 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2681 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2682 * targhooks.h (default_floatn_mode): Likewise.
2683 * targhooks.c (default_floatn_mode): Likewise.
2684 * tree.c (build_common_tree_nodes): Update accordingly.
2686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2687 Alan Hayward <alan.hayward@arm.com>
2688 David Sherwood <david.sherwood@arm.com>
2690 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2691 (mode_iterator::iterate_p): Likewise.
2692 (mode_iterator::get_wider): Likewise.
2693 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2696 Alan Hayward <alan.hayward@arm.com>
2697 David Sherwood <david.sherwood@arm.com>
2699 * coretypes.h (opt_scalar_float_mode): New typedef.
2700 * machmode.h (float_mode_for_size): New function.
2701 * emit-rtl.c (double_mode): Delete.
2702 (init_emit_once): Use float_mode_for_size.
2703 * stor-layout.c (layout_type): Likewise.
2704 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2707 Alan Hayward <alan.hayward@arm.com>
2708 David Sherwood <david.sherwood@arm.com>
2710 * output.h (assemble_real): Take a scalar_float_mode.
2711 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2712 * config/arm/arm.md (consttable_4): Likewise.
2713 (consttable_8): Likewise.
2714 (consttable_16): Likewise.
2715 * config/mips/mips.md (consttable_float): Likewise.
2716 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2717 * varasm.c (assemble_real): Take a scalar_float_mode.
2718 (output_constant_pool_2): Update accordingly.
2719 (output_constant): Likewise.
2721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2722 Alan Hayward <alan.hayward@arm.com>
2723 David Sherwood <david.sherwood@arm.com>
2725 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2726 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2727 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2728 (native_encode_real): Likewise.
2729 (native_interpret_real): Likewise.
2730 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2731 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2734 Alan Hayward <alan.hayward@arm.com>
2735 David Sherwood <david.sherwood@arm.com>
2737 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2738 <scalar_float_mode>. Simplify.
2739 (gen_extend_conv_libfunc): Likewise.
2741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2742 Alan Hayward <alan.hayward@arm.com>
2743 David Sherwood <david.sherwood@arm.com>
2745 * coretypes.h (scalar_float_mode): New type.
2746 * machmode.h (mode_traits::from_int): Use machine_mode if
2747 USE_ENUM_MODES is defined.
2748 (is_a): New function.
2750 (dyn_cast): Likewise.
2751 (scalar_float_mode): New class.
2752 (scalar_float_mode::includes_p): New function.
2753 (is_float_mode): Likewise.
2754 * gdbhooks.py (MachineModePrinter): New class.
2755 (build_pretty_printer): Use it for scalar_float_mode.
2756 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2757 (format_helper::format_helper): Turn into a template.
2758 * genmodes.c (get_mode_class): New function.
2759 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2760 or machine_mode if none.
2761 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2762 as_a <scalar_float_mode>.
2763 * dwarf2out.c (mem_loc_descriptor): Likewise.
2764 (insert_float): Likewise.
2765 (add_const_value_attribute): Likewise.
2766 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2767 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2768 (expand_unop): Update accordingly.
2769 (expand_abs_nojump): Likewise.
2770 (expand_copysign_absneg): Take a scalar_float_mode.
2771 (expand_copysign_bit): Likewise.
2772 (expand_copysign): Update accordingly.
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 (opt_mode): New class.
2779 * machmode.h (opt_mode): Likewise.
2780 (opt_mode::else_void): New function.
2781 (opt_mode::require): Likewise.
2782 (opt_mode::exists): Likewise.
2783 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2784 (GET_MODE_2XWIDER_MODE): Likewise.
2785 (mode_iterator::get_wider): Update accordingly.
2786 (mode_iterator::get_2xwider): Likewise.
2787 (mode_iterator::get_known_wider): Likewise, turning into a template.
2788 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2789 forcing a wider mode to exist.
2790 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2791 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2792 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2793 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2794 * lower-subreg.c (init_lower_subreg): Likewise.
2795 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2796 on the final iteration.
2797 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2798 a wider mode exists before asking for a move pattern.
2799 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2800 forcing a wider mode to exist.
2801 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2802 returning false if no such mode exists.
2803 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2804 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2805 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2806 Avoid checking for a MODE_INT if we already know the mode is not a
2808 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2809 forcing a wider mode to exist.
2810 (expmed_mult_highpart_optab): Likewise.
2811 (expmed_mult_highpart): Likewise.
2812 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2814 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2815 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2816 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2817 * internal-fn.c (expand_mul_overflow): Update use of
2818 GET_MODE_2XWIDER_MODE.
2819 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2820 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2821 GET_MODE_WIDER_MODE.
2822 (convert_plusminus_to_widen): Likewise.
2823 * tree-switch-conversion.c (array_value_type): Likewise.
2824 * var-tracking.c (emit_note_insn_var_location): Likewise.
2825 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2826 Return false inside rather than outside the loop if no wider mode
2828 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2829 and GET_MODE_2XWIDER_MODE
2830 (can_compare_p): Use else_void.
2831 * gdbhooks.py (OptMachineModePrinter): New class.
2832 (build_pretty_printer): Use it for opt_mode.
2834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2835 Alan Hayward <alan.hayward@arm.com>
2836 David Sherwood <david.sherwood@arm.com>
2838 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2839 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2842 Alan Hayward <alan.hayward@arm.com>
2843 David Sherwood <david.sherwood@arm.com>
2845 * machmode.h (mode_traits): New structure.
2846 (get_narrowest_mode): New function.
2847 (mode_iterator::start): Likewise.
2848 (mode_iterator::iterate_p): Likewise.
2849 (mode_iterator::get_wider): Likewise.
2850 (mode_iterator::get_known_wider): Likewise.
2851 (mode_iterator::get_2xwider): Likewise.
2852 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2853 (FOR_EACH_MODE): Likewise.
2854 (FOR_EACH_MODE_FROM): Likewise.
2855 (FOR_EACH_MODE_UNTIL): Likewise.
2856 (FOR_EACH_WIDER_MODE): Likewise.
2857 (FOR_EACH_2XWIDER_MODE): Likewise.
2858 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2859 * combine.c (simplify_comparison): Likewise
2860 * config/i386/i386.c (type_natural_mode): Likewise.
2861 * cse.c (cse_insn): Likewise.
2862 * dse.c (find_shift_sequence): Likewise.
2863 * emit-rtl.c (init_derived_machine_modes): Likewise.
2864 (init_emit_once): Likewise.
2865 * explow.c (hard_function_value): Likewise.
2866 * expmed.c (extract_fixed_bit_field_1): Likewise.
2867 (extract_bit_field_1): Likewise.
2868 (expand_divmod): Likewise.
2869 (emit_store_flag_1): Likewise.
2870 * expr.c (init_expr_target): Likewise.
2871 (convert_move): Likewise.
2872 (alignment_for_piecewise_move): Likewise.
2873 (widest_int_mode_for_size): Likewise.
2874 (emit_block_move_via_movmem): Likewise.
2875 (copy_blkmode_to_reg): Likewise.
2876 (set_storage_via_setmem): Likewise.
2877 (compress_float_constant): Likewise.
2878 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2879 * optabs-query.c (get_best_extraction_insn): Likewise.
2880 * optabs.c (expand_binop): Likewise.
2881 (expand_twoval_unop): Likewise.
2882 (expand_twoval_binop): Likewise.
2883 (widen_leading): Likewise.
2884 (widen_bswap): Likewise.
2885 (expand_parity): Likewise.
2886 (expand_unop): Likewise.
2887 (prepare_cmp_insn): Likewise.
2888 (prepare_float_lib_cmp): Likewise.
2889 (expand_float): Likewise.
2890 (expand_fix): Likewise.
2891 (expand_sfix_optab): Likewise.
2892 * postreload.c (move2add_use_add2_insn): Likewise.
2893 * reg-stack.c (reg_to_stack): Likewise.
2894 * reginfo.c (choose_hard_reg_mode): Likewise.
2895 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2896 * stor-layout.c (mode_for_size): Likewise.
2897 (smallest_mode_for_size): Likewise.
2898 (mode_for_vector): Likewise.
2899 (finish_bitfield_representative): Likewise.
2900 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2901 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2902 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2903 * var-tracking.c (prepare_call_arguments): Likewise.
2905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2906 Alan Hayward <alan.hayward@arm.com>
2907 David Sherwood <david.sherwood@arm.com>
2909 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2910 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2911 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2912 * machmode.h (mode_size): Move earlier in file.
2913 (mode_precision): Likewise.
2914 (mode_inner): Likewise.
2915 (mode_nunits): Likewise.
2916 (mode_unit_size): Likewise.
2917 (unit_unit_precision): Likewise.
2918 (mode_wider): Likewise.
2919 (mode_2xwider): Likewise.
2920 (machine_mode): New class.
2921 (mode_to_bytes): New function.
2922 (mode_to_bits): Likewise.
2923 (mode_to_precision): Likewise.
2924 (mode_to_inner): Likewise.
2925 (mode_to_unit_size): Likewise.
2926 (mode_to_unit_precision): Likewise.
2927 (mode_to_nunits): Likewise.
2928 (GET_MODE_SIZE): Use mode_to_bytes.
2929 (GET_MODE_BITSIZE): Use mode_to_bits.
2930 (GET_MODE_PRECISION): Use mode_to_precision.
2931 (GET_MODE_INNER): Use mode_to_inner.
2932 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2933 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2934 (GET_MODE_NUNITS): Use mode_to_nunits.
2935 * system.h (ALWAYS_INLINE): New macro.
2936 * config/powerpcspe/powerpcspe-c.c
2937 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2938 int for arg1_mode and arg2_mode.
2940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2941 Alan Hayward <alan.hayward@arm.com>
2942 David Sherwood <david.sherwood@arm.com>
2944 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2945 Prefix mode names with E_ in case statements.
2946 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2947 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2948 (aarch64_split_simd_move): Likewise.
2949 (aarch64_gen_storewb_pair): Likewise.
2950 (aarch64_gen_loadwb_pair): Likewise.
2951 (aarch64_gen_store_pair): Likewise.
2952 (aarch64_gen_load_pair): Likewise.
2953 (aarch64_get_condition_code_1): Likewise.
2954 (aarch64_constant_pool_reload_icode): Likewise.
2955 (get_rsqrte_type): Likewise.
2956 (get_rsqrts_type): Likewise.
2957 (get_recpe_type): Likewise.
2958 (get_recps_type): Likewise.
2959 (aarch64_gimplify_va_arg_expr): Likewise.
2960 (aarch64_simd_container_mode): Likewise.
2961 (aarch64_emit_load_exclusive): Likewise.
2962 (aarch64_emit_store_exclusive): Likewise.
2963 (aarch64_expand_compare_and_swap): Likewise.
2964 (aarch64_gen_atomic_cas): Likewise.
2965 (aarch64_emit_bic): Likewise.
2966 (aarch64_emit_atomic_swap): Likewise.
2967 (aarch64_emit_atomic_load_op): Likewise.
2968 (aarch64_evpc_trn): Likewise.
2969 (aarch64_evpc_uzp): Likewise.
2970 (aarch64_evpc_zip): Likewise.
2971 (aarch64_evpc_ext): Likewise.
2972 (aarch64_evpc_rev): Likewise.
2973 (aarch64_evpc_dup): Likewise.
2974 (aarch64_gen_ccmp_first): Likewise.
2975 (aarch64_gen_ccmp_next): Likewise.
2976 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2977 (alpha_emit_xfloating_libcall): Likewise.
2978 (emit_insxl): Likewise.
2979 (alpha_arg_type): Likewise.
2980 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2981 (arc_preferred_simd_mode): Likewise.
2982 (arc_secondary_reload): Likewise.
2983 (get_arc_condition_code): Likewise.
2984 (arc_print_operand): Likewise.
2985 (arc_legitimate_constant_p): Likewise.
2986 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2987 * config/arc/arc.md (casesi_load): Likewise.
2988 (casesi_compact_jump): Likewise.
2989 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2990 (cc_use_register): Likewise.
2991 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2992 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2993 (arm_init_iwmmxt_builtins): Likewise.
2994 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2995 (neon_expand_vector_init): Likewise.
2996 (arm_attr_length_move_neon): Likewise.
2997 (maybe_get_arm_condition_code): Likewise.
2998 (arm_emit_vector_const): Likewise.
2999 (arm_preferred_simd_mode): Likewise.
3000 (arm_output_iwmmxt_tinsr): Likewise.
3001 (thumb1_output_casesi): Likewise.
3002 (thumb2_output_casesi): Likewise.
3003 (arm_emit_load_exclusive): Likewise.
3004 (arm_emit_store_exclusive): Likewise.
3005 (arm_expand_compare_and_swap): Likewise.
3006 (arm_evpc_neon_vuzp): Likewise.
3007 (arm_evpc_neon_vzip): Likewise.
3008 (arm_evpc_neon_vrev): Likewise.
3009 (arm_evpc_neon_vtrn): Likewise.
3010 (arm_evpc_neon_vext): Likewise.
3011 (arm_validize_comparison): Likewise.
3012 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3013 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3014 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3015 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3016 (c6x_preferred_simd_mode): Likewise.
3017 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3018 (epiphany_rtx_costs): Likewise.
3019 * config/epiphany/predicates.md (proper_comparison_operator):
3021 * config/frv/frv.c (condexec_memory_operand): Likewise.
3022 (frv_emit_move): Likewise.
3023 (output_move_single): Likewise.
3024 (output_condmove_single): Likewise.
3025 (frv_hard_regno_mode_ok): Likewise.
3026 (frv_matching_accg_mode): Likewise.
3027 * config/h8300/h8300.c (split_adds_subs): Likewise.
3028 (h8300_rtx_costs): Likewise.
3029 (h8300_print_operand): Likewise.
3030 (compute_mov_length): Likewise.
3031 (output_logical_op): Likewise.
3032 (compute_logical_op_length): Likewise.
3033 (compute_logical_op_cc): Likewise.
3034 (h8300_shift_needs_scratch_p): Likewise.
3035 (output_a_shift): Likewise.
3036 (compute_a_shift_length): Likewise.
3037 (compute_a_shift_cc): Likewise.
3038 (expand_a_rotate): Likewise.
3039 (output_a_rotate): Likewise.
3040 * config/i386/i386.c (classify_argument): Likewise.
3041 (function_arg_advance_32): Likewise.
3042 (function_arg_32): Likewise.
3043 (function_arg_64): Likewise.
3044 (function_value_64): Likewise.
3045 (ix86_gimplify_va_arg): Likewise.
3046 (ix86_legitimate_constant_p): Likewise.
3047 (put_condition_code): Likewise.
3048 (split_double_mode): Likewise.
3049 (ix86_avx256_split_vector_move_misalign): Likewise.
3050 (ix86_expand_vector_logical_operator): Likewise.
3051 (ix86_split_idivmod): Likewise.
3052 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3053 (ix86_build_const_vector): Likewise.
3054 (ix86_build_signbit_mask): Likewise.
3055 (ix86_match_ccmode): Likewise.
3056 (ix86_cc_modes_compatible): Likewise.
3057 (ix86_expand_branch): Likewise.
3058 (ix86_expand_sse_cmp): Likewise.
3059 (ix86_expand_sse_movcc): Likewise.
3060 (ix86_expand_int_sse_cmp): Likewise.
3061 (ix86_expand_vec_perm_vpermi2): Likewise.
3062 (ix86_expand_vec_perm): Likewise.
3063 (ix86_expand_sse_unpack): Likewise.
3064 (ix86_expand_int_addcc): Likewise.
3065 (ix86_split_to_parts): Likewise.
3066 (ix86_vectorize_builtin_gather): Likewise.
3067 (ix86_vectorize_builtin_scatter): Likewise.
3068 (avx_vpermilp_parallel): Likewise.
3069 (inline_memory_move_cost): Likewise.
3070 (ix86_tieable_integer_mode_p): Likewise.
3071 (x86_maybe_negate_const_int): Likewise.
3072 (ix86_expand_vector_init_duplicate): Likewise.
3073 (ix86_expand_vector_init_one_nonzero): Likewise.
3074 (ix86_expand_vector_init_one_var): Likewise.
3075 (ix86_expand_vector_init_concat): Likewise.
3076 (ix86_expand_vector_init_interleave): Likewise.
3077 (ix86_expand_vector_init_general): Likewise.
3078 (ix86_expand_vector_set): Likewise.
3079 (ix86_expand_vector_extract): Likewise.
3080 (emit_reduc_half): Likewise.
3081 (ix86_emit_i387_round): Likewise.
3082 (ix86_mangle_type): Likewise.
3083 (ix86_expand_round_sse4): Likewise.
3084 (expand_vec_perm_blend): Likewise.
3085 (canonicalize_vector_int_perm): Likewise.
3086 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3087 (expand_vec_perm_1): Likewise.
3088 (expand_vec_perm_interleave3): Likewise.
3089 (expand_vec_perm_even_odd_pack): Likewise.
3090 (expand_vec_perm_even_odd_1): Likewise.
3091 (expand_vec_perm_broadcast_1): Likewise.
3092 (ix86_vectorize_vec_perm_const_ok): Likewise.
3093 (ix86_expand_vecop_qihi): Likewise.
3094 (ix86_expand_mul_widen_hilo): Likewise.
3095 (ix86_expand_sse2_abs): Likewise.
3096 (ix86_expand_pextr): Likewise.
3097 (ix86_expand_pinsr): Likewise.
3098 (ix86_preferred_simd_mode): Likewise.
3099 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3100 * config/i386/sse.md (*andnot<mode>3): Likewise.
3101 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3102 (*<code><mode>3): Likewise.
3103 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3104 (ia64_expand_atomic_op): Likewise.
3105 (ia64_arg_type): Likewise.
3106 (ia64_mode_to_int): Likewise.
3107 (ia64_scalar_mode_supported_p): Likewise.
3108 (ia64_vector_mode_supported_p): Likewise.
3109 (expand_vec_perm_broadcast): Likewise.
3110 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3111 (iq2000_function_arg_advance): Likewise.
3112 (iq2000_function_arg): Likewise.
3113 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3114 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3115 (m68k_libcall_value): Likewise.
3116 (m68k_function_value): Likewise.
3117 (sched_attr_op_type): Likewise.
3118 * config/mcore/mcore.c (mcore_output_move): Likewise.
3119 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3121 (microblaze_function_arg): Likewise.
3122 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3123 (mips_print_operand): Likewise.
3124 (mips_mode_ok_for_mov_fmt_p): Likewise.
3125 (mips_vector_mode_supported_p): Likewise.
3126 (mips_preferred_simd_mode): Likewise.
3127 (mips_expand_vpc_loongson_even_odd): Likewise.
3128 (mips_expand_vec_unpack): Likewise.
3129 (mips_expand_vi_broadcast): Likewise.
3130 (mips_expand_vector_init): Likewise.
3131 (mips_expand_vec_reduc): Likewise.
3132 (mips_expand_msa_cmp): Likewise.
3133 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3134 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3135 (cc_flags_for_mode): Likewise.
3136 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3137 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3138 (nds32_output_casesi_pc_relative): Likewise.
3139 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3140 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3141 (nvptx_gen_unpack): Likewise.
3142 (nvptx_gen_pack): Likewise.
3143 (nvptx_gen_shuffle): Likewise.
3144 (nvptx_gen_wcast): Likewise.
3145 (nvptx_preferred_simd_mode): Likewise.
3146 * config/pa/pa.c (pa_secondary_reload): Likewise.
3147 * config/pa/predicates.md (base14_operand): Likewise.
3148 * config/powerpcspe/powerpcspe-c.c
3149 (altivec_resolve_overloaded_builtin): Likewise.
3150 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3152 (rs6000_preferred_simd_mode): Likewise.
3153 (output_vec_const_move): Likewise.
3154 (rs6000_expand_vector_extract): Likewise.
3155 (rs6000_split_vec_extract_var): Likewise.
3156 (reg_offset_addressing_ok_p): Likewise.
3157 (rs6000_legitimate_offset_address_p): Likewise.
3158 (rs6000_legitimize_address): Likewise.
3159 (rs6000_emit_set_const): Likewise.
3160 (rs6000_const_vec): Likewise.
3161 (rs6000_emit_move): Likewise.
3162 (spe_build_register_parallel): Likewise.
3163 (rs6000_darwin64_record_arg_recurse): Likewise.
3164 (swap_selector_for_mode): Likewise.
3165 (spe_init_builtins): Likewise.
3166 (paired_init_builtins): Likewise.
3167 (altivec_init_builtins): Likewise.
3168 (do_load_for_compare): Likewise.
3169 (rs6000_generate_compare): Likewise.
3170 (rs6000_expand_float128_convert): Likewise.
3171 (emit_load_locked): Likewise.
3172 (emit_store_conditional): Likewise.
3173 (rs6000_output_function_epilogue): Likewise.
3174 (rs6000_handle_altivec_attribute): Likewise.
3175 (rs6000_function_value): Likewise.
3176 (emit_fusion_gpr_load): Likewise.
3177 (emit_fusion_p9_load): Likewise.
3178 (emit_fusion_p9_store): Likewise.
3179 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3180 (fusion_gpr_mem_load): Likewise.
3181 (fusion_addis_mem_combo_load): Likewise.
3182 (fusion_addis_mem_combo_store): Likewise.
3183 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3184 (fusion_gpr_mem_load): Likewise.
3185 (fusion_addis_mem_combo_load): Likewise.
3186 (fusion_addis_mem_combo_store): Likewise.
3187 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3189 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3190 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3191 (rs6000_preferred_simd_mode): Likewise.
3192 (output_vec_const_move): Likewise.
3193 (rs6000_expand_vector_extract): Likewise.
3194 (rs6000_split_vec_extract_var): Likewise.
3195 (reg_offset_addressing_ok_p): Likewise.
3196 (rs6000_legitimate_offset_address_p): Likewise.
3197 (rs6000_legitimize_address): Likewise.
3198 (rs6000_emit_set_const): Likewise.
3199 (rs6000_const_vec): Likewise.
3200 (rs6000_emit_move): Likewise.
3201 (rs6000_darwin64_record_arg_recurse): Likewise.
3202 (swap_selector_for_mode): Likewise.
3203 (paired_init_builtins): Likewise.
3204 (altivec_init_builtins): Likewise.
3205 (rs6000_expand_float128_convert): Likewise.
3206 (emit_load_locked): Likewise.
3207 (emit_store_conditional): Likewise.
3208 (rs6000_output_function_epilogue): Likewise.
3209 (rs6000_handle_altivec_attribute): Likewise.
3210 (rs6000_function_value): Likewise.
3211 (emit_fusion_gpr_load): Likewise.
3212 (emit_fusion_p9_load): Likewise.
3213 (emit_fusion_p9_store): Likewise.
3214 * config/rx/rx.c (rx_gen_move_template): Likewise.
3215 (flags_from_mode): Likewise.
3216 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3217 (s390_slb_comparison): Likewise.
3218 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3219 (s390_vector_mode_supported_p): Likewise.
3220 (s390_cc_modes_compatible): Likewise.
3221 (s390_match_ccmode_set): Likewise.
3222 (s390_canonicalize_comparison): Likewise.
3223 (s390_emit_compare_and_swap): Likewise.
3224 (s390_branch_condition_mask): Likewise.
3225 (s390_rtx_costs): Likewise.
3226 (s390_secondary_reload): Likewise.
3227 (__SECONDARY_RELOAD_CASE): Likewise.
3228 (s390_expand_cs): Likewise.
3229 (s390_preferred_simd_mode): Likewise.
3230 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3231 * config/sh/sh.c (sh_print_operand): Likewise.
3232 (dump_table): Likewise.
3233 (sh_secondary_reload): Likewise.
3234 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3235 * config/sh/sh.md (casesi_worker_1): Likewise.
3236 (casesi_worker_2): Likewise.
3237 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3238 (fcc_comparison_operator): Likewise.
3239 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3240 (emit_soft_tfmode_cvt): Likewise.
3241 (sparc_preferred_simd_mode): Likewise.
3242 (output_cbranch): Likewise.
3243 (sparc_print_operand): Likewise.
3244 (sparc_expand_vec_perm_bmask): Likewise.
3245 (vector_init_bshuffle): Likewise.
3246 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3247 (spu_vector_mode_supported_p): Likewise.
3248 (spu_expand_insv): Likewise.
3249 (spu_emit_branch_or_set): Likewise.
3250 (spu_handle_vector_attribute): Likewise.
3251 (spu_builtin_splats): Likewise.
3252 (spu_builtin_extract): Likewise.
3253 (spu_builtin_promote): Likewise.
3254 (spu_expand_sign_extend): Likewise.
3255 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3256 (tilegx_simd_int): Likewise.
3257 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3258 (tilepro_simd_int): Likewise.
3259 * config/v850/v850.c (const_double_split): Likewise.
3260 (v850_print_operand): Likewise.
3261 (ep_memory_offset): Likewise.
3262 * config/vax/vax.c (vax_rtx_costs): Likewise.
3263 (vax_output_int_move): Likewise.
3264 (vax_output_int_add): Likewise.
3265 (vax_output_int_subtract): Likewise.
3266 * config/visium/predicates.md (visium_branch_operator): Likewise.
3267 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3268 (visium_print_operand_address): Likewise.
3269 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3270 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3271 (xtensa_expand_conditional_branch): Likewise.
3272 (xtensa_copy_incoming_a7): Likewise.
3273 (xtensa_output_literal): Likewise.
3274 * dfp.c (decimal_real_maxval): Likewise.
3275 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3278 Alan Hayward <alan.hayward@arm.com>
3279 David Sherwood <david.sherwood@arm.com>
3281 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3282 (mode_nunits_inline): Likewise.
3283 (mode_inner_inline): Likewise.
3284 (mode_unit_size_inline): Likewise.
3285 (mode_unit_precision_inline): Likewise.
3286 (emit_insn_modes_h): Likewise. Also emit a #define of the
3288 (emit_mode_wider): Add an E_ prefix to mode names.
3289 (emit_mode_complex): Likewise.
3290 (emit_mode_inner): Likewise.
3291 (emit_mode_adjustments): Likewise.
3292 (emit_mode_int_n): Likewise.
3293 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3294 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3295 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3296 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3297 (CRC32_BUILTIN, ENTRY): Likewise.
3298 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3299 (aarch64_pop_regs): Likewise.
3300 (aarch64_process_components): Likewise.
3301 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3302 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3303 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3304 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3305 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3306 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3308 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3309 (pcmpistr): Likewise.
3310 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3311 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3312 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3314 * config/rl78/rl78.c (mduc_regs): Likewise.
3315 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3316 (htm_expand_builtin): Likewise.
3317 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3318 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3319 * config/xtensa/xtensa.c (print_operand): Likewise.
3320 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3321 (NUM_MODE_VECTOR_INT): Likewise.
3322 * genoutput.c (null_operand): Likewise.
3323 (output_operand_data): Likewise.
3324 * genrecog.c (print_parameter_value): Likewise.
3325 * lra.c (debug_operand_data): Likewise.
3327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3329 * dwarf2out.c (scompare_loc_descriptor_wide)
3330 (scompare_loc_descriptor_narrow): New functions, split out from...
3331 (scompare_loc_descriptor): ...here.
3332 * expmed.c (emit_store_flag_int): New function, split out from...
3333 (emit_store_flag): ...here.
3335 2017-08-30 Richard Biener <rguenther@suse.de>
3337 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3338 (dwarf2out_early_finish): Move setting of AT_pubnames from
3339 early debug output to early finish.
3341 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3343 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3344 and -mdata-region to the assembler.
3346 2017-08-30 Richard Biener <rguenther@suse.de>
3348 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3350 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3351 (dwarf2out_early_global_decl): Only generate a DIE for the
3352 abstract origin if it doesn't already exist or is a declaration DIE.
3353 (resolve_addr): Do not add the linkage name twice when
3354 generating a stub DIE for the DW_TAG_GNU_call_site target.
3356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3358 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3359 Use machine_mode rather than int for arg1_mode.
3361 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3364 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3365 that the second argument of the built-in functions to unpack
3366 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3367 switch statement instead a lot of if statements.
3368 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3369 Allow 64-bit values to be in Altivec registers as well as
3370 traditional floating point registers.
3371 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3373 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3375 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3376 MAX_REGS_PER_ADDRESS == 1.
3378 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3380 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3381 * config/i386/i386.c (ix86_option_override_internal): Simplify
3382 setting of opts->x_flag_entry.
3384 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3385 Jakub Jelinek <jakub@redhat.com>
3386 Richard Biener <rguenther@suse.de>
3388 PR tree-optimization/81503
3389 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3390 folded constant fits in the target type; reorder tests for clarity.
3392 2017-08-29 Martin Liska <mliska@suse.cz>
3394 * passes.def: Include pass_lower_switch.
3395 * stmt.c (dump_case_nodes): Remove and move to
3396 tree-switch-conversion.
3397 (case_values_threshold): Likewise.
3398 (expand_switch_as_decision_tree_p): Likewise.
3399 (emit_case_decision_tree): Likewise.
3400 (expand_case): Likewise.
3401 (balance_case_nodes): Likewise.
3402 (node_has_low_bound): Likewise.
3403 (node_has_high_bound): Likewise.
3404 (node_is_bounded): Likewise.
3405 (emit_case_nodes): Likewise.
3406 (struct simple_case_node): New struct.
3407 (add_case_node): Remove.
3408 (emit_case_dispatch_table): Use vector instead of case_list.
3409 (reset_out_edges_aux): Remove.
3410 (compute_cases_per_edge): Likewise.
3411 (expand_case): Build list of simple_case_node.
3412 (expand_sjlj_dispatch_table): Use it.
3413 * tree-switch-conversion.c (struct case_node): Moved from
3414 stmt.c and adjusted.
3415 (emit_case_nodes): Likewise.
3416 (node_has_low_bound): Likewise.
3417 (node_has_high_bound): Likewise.
3418 (node_is_bounded): Likewise.
3419 (case_values_threshold): Likewise.
3420 (reset_out_edges_aux): Likewise.
3421 (compute_cases_per_edge): Likewise.
3422 (add_case_node): Likewise.
3423 (dump_case_nodes): Likewise.
3424 (balance_case_nodes): Likewise.
3425 (expand_switch_as_decision_tree_p): Likewise.
3426 (emit_jump): Likewise.
3427 (emit_case_decision_tree): Likewise.
3428 (try_switch_expansion): Likewise.
3429 (do_jump_if_equal): Likewise.
3430 (emit_cmp_and_jump_insns): Likewise.
3431 (fix_phi_operands_for_edge): New function.
3432 (record_phi_operand_mapping): Likewise.
3433 (class pass_lower_switch): New pass.
3434 (pass_lower_switch::execute): New function.
3435 (make_pass_lower_switch): Likewise.
3436 (conditional_probability):
3437 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3438 * tree-pass.h: Add make_pass_lower_switch.
3440 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3443 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3446 2017-08-29 Richard Biener <rguenther@suse.de>
3448 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3449 we do not add a DW_AT_inline attribute twice.
3450 (gen_subprogram_die): Remove code setting DW_AT_inline on
3451 DECL_ABSTRACT_P nodes.
3453 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3455 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3456 calls to internal functions.
3457 (gimplify_modify_expr): Likewise.
3458 * tree-call-cdce.c (use_internal_fn): Likewise.
3459 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3460 (convert_to_divmod): Set the nothrow flag.
3461 * tree-if-conv.c (predicate_mem_writes): Likewise.
3462 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3463 (vectorizable_call): Likewise.
3464 (vectorizable_store): Likewise.
3465 (vectorizable_load): Likewise.
3466 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3467 (vect_recog_mask_conversion_pattern): Likewise.
3469 2017-08-29 Martin Liska <mliska@suse.cz>
3472 * gcc.c (driver_handle_option): Add new argument.
3473 * opts-common.c (handle_option): Pass
3474 target_option_override_hook.
3475 * opts-global.c (lang_handle_option): Add new option.
3476 (set_default_handlers): Add new argument.
3477 (decode_options): Likewise.
3478 * opts.c (target_handle_option): Likewise.
3479 (common_handle_option): Call target_option_override_hook.
3480 * opts.h (struct cl_option_handler_func): Add hook for
3481 target option override.
3482 (struct cl_option_handlers): Likewise.
3483 (set_default_handlers): Add new argument.
3484 (decode_options): Likewise.
3485 (common_handle_option): Likewise.
3486 (target_handle_option): Likewise.
3487 * toplev.c (toplev::main): Pass targetm.target_option.override
3490 2017-08-29 Richard Biener <rguenther@suse.de>
3491 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3493 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3494 life to the active subtree.
3496 2017-08-28 Jeff Law <law@redhat.com>
3498 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3499 derive_equivalences.
3500 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3502 (edge_info::derive_equivalences): New private member function
3504 * tree-ssa-dom.c (class edge_info): Changed from a struct
3505 to a class. Add ctor/dtor, methods and data members.
3506 (edge_info::edge_info): Renamed from allocate_edge_info.
3507 Initialize additional members.
3508 (edge_info::~edge_info): New.
3509 (free_dom_edge_info): Delete the edge info.
3510 (record_edge_info): Use new class & associated member functions.
3511 Tighten forms for testing for edge equivalences.
3512 (record_temporary_equivalences): Iterate over the simple
3513 equivalences rather than assuming there's only one per edge.
3514 (cprop_into_successor_phis): Iterate over the simple
3515 equivalences rather than assuming there's only one per edge.
3516 (optimize_stmt): Use operand_equal_p rather than pointer
3517 equality for mini-DSE code.
3519 2017-08-28 Nathan Sidwell <nathan@acm.org>
3521 * gcc.c (execute): Fold SIGPIPE handling into switch
3522 statement. Adjust internal error message.
3524 2017-08-28 Richard Biener <rguenther@suse.de>
3527 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3528 Do nothing for removed DIEs.
3530 2017-08-28 Richard Biener <rguenther@suse.de>
3532 PR tree-optimization/81977
3533 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3536 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3539 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3540 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3541 using targetm.gen_mem_thread_fence.
3543 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3546 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3547 predicate to register_operand. Reorder operands.
3548 (*btr<mode>): Ditto.
3549 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3550 (*btr<mode>_mask): Ditto.
3552 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3554 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3555 * config/rs6000/xmmintrin.h: New file.
3556 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3558 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3561 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3562 parameter and_insn and return it.
3563 (recombine_lvx_pattern): Insert a copy to ensure availability of
3564 the base register of the copied masking operation at the point of
3565 the instruction replacement.
3566 (recombine_stvx_pattern): Likewise.
3568 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3570 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3571 undocumented switches.
3572 (-mpower9-dform-vector): Likewise.
3573 (-mpower9-dform): Likewise.
3574 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3575 comments to delete references to -mpower9-dform* switches.
3576 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3577 Delete reference to -mpower9-dform* switches, test for
3578 -mpower9-vector instead.
3579 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3580 (OTHER_P9_VECTOR_MASKS): Likewise.
3581 (POWERPC_MASKS): Likewise.
3582 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3583 tests against -mpower9-dform* to -mpower9-vector. Delete code
3584 that checked for -mpower9-dform* consistancy with other options.
3585 Add test for -mpower9-misc to enable other power9 switches.
3586 (rs6000_init_hard_regno_mode_ok): Likewise.
3587 (rs6000_option_override_internal): Likewise.
3588 (rs6000_emit_prologue): Likewise.
3589 (rs6000_emit_epilogue): Likewise.
3590 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3591 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3592 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3594 (emit_fusion_p9_store): Likewise.
3595 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3596 resetting these macros if the assembler does not support ISA 3.0
3598 (TARGET_P9_DFORM_VECTOR): Likewise.
3599 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3600 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3602 2017-08-25 Alan Modra <amodra@gmail.com>
3604 PR rtl-optimization/81747
3605 * cse.c (cse_extended_basic_block): Don't attempt to record
3606 equivalences for degenerate conditional branches that branch
3607 to their fall-through.
3609 2017-08-24 Martin Sebor <msebor@redhat.com>
3612 * gimple-fold.c (size_must_be_zero_p): New function.
3613 (gimple_fold_builtin_memory_op): Call it.
3615 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3617 * config/rs6000/mm_malloc.h: New file.
3619 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3621 PR tree-optimization/81913
3622 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3623 analysis when either IVs in condition can wrap.
3625 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3627 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3628 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3630 2017-08-24 Richard Biener <rguenther@suse.de>
3633 * targhooks.c (default_target_can_inline_p): Properly
3634 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3635 is present and always compare.
3636 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3637 infer -mfpmath=sse from TARGET_SSE_P.
3638 (ix86_can_inline_p): Properly use target_option_default_node when
3639 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3641 2017-08-24 Richard Biener <rguenther@suse.de>
3644 * dwarf2out.c (output_die): Handle flag_generate_offload like
3646 (output_comp_unit): Likewise.
3647 (gen_array_type_die): Likewise.
3648 (dwarf2out_early_finish): Likewise.
3649 (note_variable_value_in_expr): Likewise.
3650 (dwarf2out_finish): Likewise. Adjust assert.
3651 * cgraphunit.c (symbol_table::compile): Move setting of
3652 flag_generate_offload earlier ...
3653 (symbol_table::finalize_compilation_unit): ... here, before
3654 early debug finalization.
3656 2017-08-24 Richard Biener <rguenther@suse.de>
3658 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3659 and ipa-fnsummary.h.
3660 (ix86_can_inline_p): When ix86_fpmath flags do not match
3661 check whether the callee uses FP math at all.
3663 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3666 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3667 instead of TYPE_PRECISION.
3669 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3670 Alan Hayward <alan.hayward@arm.com>
3671 David Sherwood <david.sherwood@arm.com>
3673 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3675 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3677 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3679 (ix86_frame::stack_realign_allocate): New field.
3680 (struct machine_frame_state): Modify comments.
3681 (machine_frame_state::sp_realigned_fp_end): New field.
3682 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3684 (sp_valid_at): Add assertion to assure no attempt to access invalid
3685 offset of a realigned stack.
3686 (fp_valid_at): Likewise.
3687 (choose_baseaddr): Modify comments.
3688 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3689 ix86_expand_prologue.
3690 (ix86_expand_prologue): Modify stack realignment and allocation.
3691 (ix86_expand_epilogue): Modify comments.
3692 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3693 avx2_runtime, avx512f, and avx512f_runtime.
3695 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3697 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3698 (mstackrealign): Do not init to -1.
3699 * config/i386/i386.c (ix86_option_override_internal):
3700 Check opts_set, not opts when setting default value of
3701 opts->x_ix86_force_align_arg_pointer.
3703 2017-08-23 Richard Biener <rguenther@suse.de>
3705 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3706 lang_hooks.decl_printable_name.
3707 * print-rtl-function.c (print_rtx_function): Likewise.
3708 * tree-pretty-print.c (dump_function_header): Likewise.
3710 2017-08-23 Richard Biener <rguenther@suse.de>
3713 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3714 -g0 at compile-time.
3716 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3719 * doc/sourcebuild.texi (Other hardware attributes):
3722 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3725 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3726 Fix single-use check.
3728 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3730 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3732 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3734 * doc/install.texi: Modify to add more details on running selected
3737 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3739 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3740 is combined with -mabi=ms.
3741 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3744 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3746 PR tree-optimization/81488
3747 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3748 and cached_basis fields.
3749 (MAX_SPREAD): New constant.
3750 (alloc_cand_and_find_basis): Initialize new fields.
3751 (clear_visited): New function.
3752 (create_phi_basis_1): Rename from create_phi_basis, set visited
3753 and cached_basis fields.
3754 (create_phi_basis): New wrapper function.
3755 (phi_add_costs_1): Rename from phi_add_costs, add spread
3756 parameter, set visited field, short-circuit when limits reached.
3757 (phi_add_costs): New wrapper function.
3758 (record_phi_increments_1): Rename from record_phi_increments, set
3760 (record_phi_increments): New wrapper function.
3761 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3762 (phi_incr_cost): New wrapper function.
3763 (all_phi_incrs_profitable_1): Rename from
3764 all_phi_incrs_profitable, set visited field.
3765 (all_phi_incrs_profitable): New wrapper function.
3767 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3768 Alan Hayward <alan.hayward@arm.com>
3769 David Sherwood <david.sherwood@arm.com>
3771 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3772 that takes the outer and inner modes.
3773 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3774 comparison as the canonical test for a paradoxical subreg.
3775 * combine.c (simplify_set): Use paradoxical_subreg_p.
3776 (make_extraction): Likewise.
3777 (force_to_mode): Likewise.
3778 (rtx_equal_for_field_assignment_p): Likewise.
3779 (gen_lowpart_for_combine): Likewise.
3780 (simplify_comparison): Likewise.
3781 * cse.c (equiv_constant): Likewise.
3782 * expmed.c (store_bit_field_1): Likewise.
3783 * final.c (alter_subreg): Likewise.
3784 * fwprop.c (propagate_rtx): Likewise.
3785 (forward_propagate_subreg): Likewise.
3786 * ira-conflicts.c (ira_build_conflicts): Likewise.
3787 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3788 * lra-constraints.c (curr_insn_transform): Likewise.
3789 (split_reg): Likewise.
3790 * lra-eliminations.c (move_plus_up): Likewise.
3791 (lra_eliminate_regs_1): Likewise.
3792 * recog.c (general_operand): Likewise.
3793 * ree.c (combine_reaching_defs): Likewise.
3794 * reload.c (push_reload): Likewise.
3795 (find_reloads): Likewise.
3796 * reload1.c (elimination_effects): Likewise.
3797 (compute_reload_subreg_offset): Likewise.
3798 (choose_reload_regs): Likewise.
3799 * rtlanal.c (subreg_lsb_1): Likewise.
3800 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3801 (simplify_subreg): Likewise.
3802 * var-tracking.c (track_loc_p): Likewise.
3803 * emit-rtl.c (byte_lowpart_offset): Likewise.
3804 (paradoxical_subreg_p): Delete out-of-line definition.
3806 2017-08-22 Jeff Law <law@redhat.com>
3808 PR tree-optimization/81741
3809 PR tree-optimization/71947
3810 * tree-ssa-dom.c: Include tree-inline.h.
3811 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3812 equivalences if one is more expensive to compute than the other.
3813 * tree-ssa-scopedtables.h (class const_or_copies): Make
3814 record_const_or_copy_raw method private.
3815 (class avail_exprs_stack): New method simplify_binary_operation.
3816 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3817 avail_exprs_stack::simplify_binary_operation as needed.
3818 (avail_exprs_stack::simplify_binary_operation): New function.
3820 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3822 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3823 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3824 (DOT_SYMBOLS): Likewise.
3825 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3826 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3827 (RS6000_ABI_NAME): Likewise.
3828 (TARGET_CMODEL): Likewise.
3829 (TOC_SECTION_ASM_OP): Likewise.
3830 (SET_CMODEL): New macro.
3831 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3833 2017-08-22 Richard Biener <rguenther@suse.de>
3835 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3836 to allow for free-lang-data replacements similar to verify_type_variant.
3838 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3839 Alan Hayward <alan.hayward@arm.com>
3840 David Sherwood <david.sherwood@arm.com>
3842 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3843 VOIDmode for the LABEL_REF.
3845 2017-08-22 Richard Biener <rguenther@suse.de>
3847 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3849 2017-08-22 Richard Biener <rguenther@suse.de>
3851 * common.opt (feliminate-dwarf2-dups): Ignore.
3852 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3853 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3854 same_die_p_wrap, compute_section_prefix,
3855 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3856 (comdat_symbol_id, comdat_symbol_number): Likewise.
3857 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3859 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3860 (output_die): Mark unreachable path unreachable.
3861 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3862 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3863 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3864 (dwarf2out_early_finish): Likewise.
3866 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
3868 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
3870 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
3873 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
3874 not VAR_P. Filter attribute warnings with OPT_Wattributes.
3875 (avr_attribute_table) <io, io_low, address>: Initialize
3876 .decl_required with true.
3878 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
3880 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
3881 undocumented debugging options.
3882 (-mvsx-scalar-double): Likewise.
3883 (-mallow-df-permute): Likewise.
3884 (-mvectorize-builtins): Likewise.
3885 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3886 (rs6000_builtin_vectorized_function): Likewise.
3887 (rs6000_builtin_md_vectorized_function): Likewise.
3888 (rs6000_opt_vars): Likewise.
3890 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
3893 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
3894 (*btrq_imm): Rename from *btrq.
3895 (*btcq_imm): Rename from *btcq.
3896 (btsc): New code attribute.
3897 (*<btsc><mode>): New insn pattern.
3898 (*btr<mode>): Ditto.
3899 (*<btsc><mode>_mask): New insn_and_split pattern.
3900 (*btr<mode>_mask): Ditto.
3902 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3903 Alan Hayward <alan.hayward@arm.com>
3904 David Sherwood <david.sherwood@arm.com>
3906 * function.c (pad_below): Simplify padding calculation.
3908 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3909 Alan Hayward <alan.hayward@arm.com>
3910 David Sherwood <david.sherwood@arm.com>
3912 * target.def (function_prologue): Remove frame size argument.
3913 (function_epilogue): Likewise.
3914 * doc/tm.texi: Regenerate.
3915 * final.c (final_start_function): Update call to function_prologue.
3916 (final_end_function): Update call to function_epilogue.
3917 (default_function_pro_epilogue): Remove frame size argument.
3918 * output.h (default_function_pro_epilogue): Likewise.
3919 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3920 (arm_output_function_prologue): Likewise.
3921 * config/frv/frv.c (frv_function_prologue): Likewise.
3922 (frv_function_epilogue): Likewise.
3923 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3924 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3925 (ia64_output_function_epilogue): Likewise.
3926 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3927 (m32r_output_function_epilogue): Likewise.
3928 * config/microblaze/microblaze.c (microblaze_function_prologue)
3929 (microblaze_function_epilogue): Likewise.
3930 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3931 (mips_output_function_epilogue): Likewise.
3932 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3933 (mmix_target_asm_function_epilogue): Likewise.
3934 * config/msp430/msp430.c (msp430_start_function): Likewise.
3935 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3936 (nds32_asm_function_epilogue): Likewise.
3937 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3938 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3939 (pa_output_function_epilogue): Likewise.
3940 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3941 (rs6000_output_function_epilogue): Likewise.
3942 * config/rl78/rl78.c (rl78_start_function): Likewise.
3943 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3944 (rs6000_output_function_epilogue): Likewise.
3945 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3946 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3947 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3948 (sparc_asm_function_epilogue): Likewise.
3950 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3952 * tree.h (type_has_mode_precision_p): New function.
3953 * convert.c (convert_to_integer_1): Use it.
3954 * expr.c (expand_expr_real_2): Likewise.
3955 (expand_expr_real_1): Likewise.
3956 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3957 * match.pd: Likewise.
3958 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3959 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3960 * tree-tailcall.c (process_assignment): Likewise.
3961 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3962 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3963 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3964 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3965 (vectorizable_assignment): Likewise.
3966 (vectorizable_shift): Likewise.
3967 (vectorizable_operation): Likewise.
3968 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3970 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3972 * match.pd: Add pow (C, x) simplification.
3974 2017-08-21 Richard Biener <rguenther@suse.de>
3976 PR tree-optimization/81900
3977 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3978 for blocks with abnormal predecessors.
3979 (compute_antic): Do not set visited flag prematurely.
3981 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3984 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3986 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3988 * stor-layout.h (vector_type_mode): Move to...
3989 * tree.h (vector_type_mode): ...here.
3990 * stor-layout.c (vector_type_mode): Move to...
3991 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3993 2017-08-21 Richard Biener <rguenther@suse.de>
3995 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3996 register_external_die hooks.
3997 (debug_false_tree_charstarstar_uhwistar): Declare.
3998 (debug_nothing_tree_charstar_uhwi): Likewise.
3999 * debug.c (do_nothing_debug_hooks): Adjust.
4000 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4001 (debug_nothing_tree_charstar_uhwi): Likewise.
4002 * dbxout.c (dbx_debug_hooks): Adjust.
4003 (xcoff_debug_hooks): Likewise.
4004 * sdbout.c (sdb_debug_hooks): Likewise.
4005 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4006 * dwarf2out.c (macinfo_label_base): New global.
4007 (dwarf2out_register_external_die): New function for the
4008 register_external_die hook.
4009 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4010 (dwarf2_debug_hooks): Use them.
4011 (dwarf2_lineno_debug_hooks): Adjust.
4012 (struct die_struct): Add with_offset flag.
4013 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4014 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4015 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4016 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4017 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4018 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4019 defining section names for the early LTO debug variants.
4020 (reset_indirect_string): New helper.
4021 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4022 (print_dw_val): Add support for offsetted symbol references.
4023 (get_ultimate_context): Split out from is_cxx.
4024 (is_cxx): Use get_ultimate_context.
4025 (is_fortran): Add decl overload.
4026 (compute_comp_unit_symbol): Split out worker from
4027 compute_section_prefix.
4028 (compute_section_prefix): Call compute_comp_unit_symbol and
4029 set comdat_type_p here.
4030 (output_die): Skip DIE symbol output for the LTO added one.
4031 Handle DIE symbol references with offset.
4032 (output_comp_unit): Guard section name mangling properly.
4033 For LTO debug sections emit a symbol at the section beginning
4034 which we use to refer to its DIEs.
4035 (add_abstract_origin_attribute): For DIEs registered via
4036 dwarf2out_register_external_die directly refer to the early
4037 DIE rather than indirectly through the shadow one we created.
4038 Remove obsolete call to dwarf2out_abstract_function for
4039 non-function/block origins.
4040 (gen_array_type_die): When generating early LTO debug do
4041 not emit DW_AT_string_length.
4042 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4043 late when in LTO. As suggested place a gcc_unreachable for
4044 the DECL_ABSTRACT_P case.
4045 (gen_subprogram_die): Avoid another specification DIE
4046 for early built declarations/definitions for the late LTO case.
4047 (gen_variable_die): Add type references for late duplicated VLA dies
4049 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4050 we have the abstract instance already.
4051 (process_scope_var): Adjust decl DIE contexts in LTO which
4052 first puts them in limbo.
4053 (gen_decl_die): Do not generate type DIEs late apart from
4054 types for VLAs or for decls we do not yet have a DIE. Do not
4055 call dwarf2out_abstract_function late.
4056 (dwarf2out_early_global_decl): Make sure to create DIEs
4057 for abstract instances of a decl first.
4058 (dwarf2out_late_global_decl): Adjust comment.
4059 (output_macinfo_op): With multiple macro sections use
4060 macinfo_label_base to distinguish labels.
4061 (output_macinfo): Likewise. Update macinfo_label_base.
4062 Pass in the line info label.
4063 (note_variable_value_in_expr): When generating LTO resolve
4064 all variable values here by generating DIEs as needed.
4065 (init_sections_and_labels): Add early LTO debug flag parameter
4066 and generate different sections and names if set. Add generation
4067 counter for the labels so we can have multiple of them.
4068 (reset_dies): Helper to allow DIEs to be output multiple times.
4069 (dwarf2out_finish): When outputting DIEs to the fat part of an
4070 LTO object first reset DIEs.
4071 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4072 (modified_type_die): Check for decl_ultimate_origin being self
4074 (gen_type_die_with_usage): Likewise.
4075 (gen_typedef_die): Allow decl_ultimate_origin being self.
4076 (set_decl_abstract_flags): Remove.
4077 (set_block_abstract_flags): Likewise.
4078 (dwarf2out_abstract_function): Treat the early generated DIEs
4079 as the abstract copy and only add DW_AT_inline and
4080 DW_AT_artificial here and call set_decl_origin_self.
4081 If the DIE has an abstract origin don't do anything.
4082 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4083 if we have none yet (Go fails to build one, PR78628).
4084 (variably_modified_type_p): Prevent endless recursion for Ada
4085 cyclic pointer types.
4086 * lto-streamer-in.c: Include debug.h.
4087 (dref_queue): New global.
4088 (lto_read_tree_1): Stream in DIE references.
4089 (lto_input_tree): Register DIE references.
4090 (input_function): Stream DECL_DEBUG_ARGS.
4091 * lto-streamer-out.c: Include debug.h.
4092 (lto_write_tree_1): Output DIE references.
4093 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4094 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4095 (output_function): Stream DECL_DEBUG_ARGS.
4096 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4097 Stream DECL_ABSTRACT_ORIGIN.
4098 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4099 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4100 DECL_CONTEXT for file-scope decls.
4101 * lto-streamer.h (struct dref_entry): Declare.
4102 (dref_queue): Likewise.
4103 * cfgexpand.c (pass_expand::execute): Do not call the
4104 outlining_inline_function hook here.
4105 * lto-wrapper.c (debug_obj): New global.
4106 (tool_cleanup): Unlink it if required.
4107 (debug_objcopy): New function.
4108 (run_gcc): Handle early debug sections in the IL files by
4109 extracting them to separate files, partially linkin them and
4110 feeding the result back as result to the linker.
4111 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4112 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4113 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4114 sections into a separate segment.
4115 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4117 (darwin_asm_dwarf_section): Likewise.
4118 (darwin_asm_output_dwarf_offset): Likewise.
4119 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4121 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4122 Alan Hayward <alan.hayward@arm.com>
4123 David Sherwood <david.sherwood@arm.com>
4125 * read-md.h (md_reader::record_potential_iterator_use): Replace
4126 pointer argument with an rtx and an index.
4127 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4128 (apply_mode_iterator): Likewise.
4129 (apply_code_iterator): Likewise.
4130 (apply_int_iterator): Likewise.
4131 (apply_subst_iterator): Likewise.
4132 (record_iterator_use): Likewise.
4133 (record_attribute_use): Likewise.
4134 (md_reader::record_potential_iterator_use): Likewise. Update calls
4135 to record_iterator_use and apply_iterator.
4136 (iterator_use): Replace ptr with x and index.
4137 (attribute_use): Likewise.
4138 (apply_attribute_uses): Update calls to apply_iterator.
4139 (apply_iterators): Likewise. Update initialization of iterator_use.
4140 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4141 and record_potential_iterator_use.
4142 (rtx_reader::read_rtx_operand): Likewise.
4144 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4145 Alan Hayward <alan.hayward@arm.com>
4146 David Sherwood <david.sherwood@arm.com>
4148 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4151 2017-08-21 Richard Biener <rguenther@suse.de>
4154 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4155 at struct end conservatively when comparing common bases.
4157 2017-08-21 Richard Biener <rguenther@suse.de>
4159 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4160 (mem_ref_in_stmt): Remove.
4161 (determine_max_movement): Use ref index to get at the reference.
4162 (invariantness_dom_walker::before_dom_children): Deal with
4163 lim data already initialized.
4164 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4166 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4168 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4169 (TARGET_ISA_ROUND): Ditto.
4170 (TARGET_ROUND): Ditto.
4171 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4172 * config/i386/i386.md: Ditto.
4173 * config/i386/sse.md: Ditto.
4174 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4175 with OPTION_MASK_ISA_SSE4_1.
4177 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4180 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4181 __builtin_ia32_lzcnt_u16.
4183 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4186 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4187 (rs6000_set_current_function): Rewrite function to use it.
4189 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4192 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4193 and TYPE_WARN_IF_NOT_ALIGN.
4194 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4195 (handle_warn_if_not_align): New.
4196 (place_union_field): Call handle_warn_if_not_align.
4197 (place_field): Call handle_warn_if_not_align.
4198 Copy TYPE_WARN_IF_NOT_ALIGN.
4199 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4200 (layout_type): Likewise.
4201 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4203 (tree_decl_common): Add warn_if_not_align.
4204 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4205 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4206 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4207 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4208 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4209 * doc/extend.texi: Document warn_if_not_aligned attribute.
4210 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4212 2017-08-17 Martin Liska <mliska@suse.cz>
4215 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4216 (get_data_dependence): Use it as pointer type.
4217 (distribute_loop): Likewise.
4219 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4221 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4222 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4223 (p8_vmrgow_<mode>_direct): New define_insn.
4224 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4225 handle endianness for vmrgew and vmrgow permute patterns.
4227 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4229 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4230 * config/rs6000/rs6000-cpus.def: Remove comment.
4231 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4232 (POWERPC_MASKS): Likewise.
4233 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4234 use of TARGET_VSX_TIMODE.
4235 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4236 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4237 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4238 (rs6000_option_override_internal): Remove dead code.
4239 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4240 (rs6000_legitimize_reload_address): Likewise.
4241 (rs6000_legitimate_address_p): Likewise.
4242 (rs6000_opt_masks): Delete "vsx-timode".
4243 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4244 from function comment.
4245 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4246 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4247 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4249 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4250 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4251 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4254 (define_peephole2): Likewise.
4256 2017-08-17 Martin Sebor <msebor@redhat.com>
4259 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4260 past the end of an array.
4261 (test_pp_format): Add test cases.
4263 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4265 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4266 missing ECF_NOTHROW flags.
4268 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4271 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4272 operands residing in integer registers.
4273 (*vsx_le_perm_load_<mode>): Likewise.
4274 (*vsx_le_perm_store_<mode>): Likewise.
4275 (define_peephole2): Add peepholes to optimize the above.
4277 2017-08-17 Marek Polacek <polacek@redhat.com>
4280 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4281 to mimic what shorten_compare did. Change the return type to bool.
4282 (fold_cond_expr_with_comparison): Update call to
4283 operand_equal_for_comparison_p.
4284 (fold_ternary_loc): Likewise.
4286 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4288 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4290 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4292 2017-08-17 Richard Biener <rguenther@suse.de>
4294 * tree-ssa-structalias.c (solve_graph): When propagating
4295 to successors update the graphs succ edges and avoid duplicate work.
4297 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4300 * config/i386/i386.c (ix86_option_override_internal): Save target
4301 specific options after ix86_stack_protector_guard_reg was changed.
4303 2017-08-17 Richard Biener <rguenther@suse.de>
4305 PR tree-optimization/81827
4306 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4307 (new_var_info): Initialize it conservatively.
4308 (get_call_vi): Mark register vars.
4309 (new_scalar_tmp_constraint_exp): Likewise.
4310 (handle_rhs_call): Likewise.
4311 (handle_const_call): Likewise.
4312 (create_function_info_for): Likewise.
4313 (solve_constraints): Sort varinfos to separate register from
4314 non-register vars to pack points-to solution bitmaps during
4317 2017-08-17 Marek Polacek <polacek@redhat.com>
4319 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4321 2017-08-17 Richard Biener <rguenther@suse.de>
4323 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4324 to true when overflow is undefined and we saturated the result.
4326 2017-08-17 Alan Modra <amodra@gmail.com>
4329 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4330 Don't use store multiple if only one reg needs saving.
4331 (interesting_frame_related_regno): New function.
4332 (rs6000_frame_related): Don't emit frame info for regs that
4334 (rs6000_emit_epilogue): Likewise.
4336 2017-08-16 Nathan Sidwell <nathan@acm.org>
4338 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4339 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4340 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4341 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4342 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4343 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4344 (verify_type): Adjust for TYPE_BINFO move.
4345 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4346 process TYPE_BINFO directly.
4347 (hash_tree): Likewise.
4348 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4350 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4353 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4355 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4357 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4360 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4361 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4362 Add ix86_binary_operator_ok to insn constraint.
4363 (*iordi_1_bts): Ditto.
4364 (*xordi_1_btc): Ditto.
4365 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4366 Update corresponding peephole2 pattern.
4370 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4372 PR tree-optimization/81832
4373 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4374 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4376 2017-08-16 Marek Polacek <polacek@redhat.com>
4379 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4380 regarding size_zero_node.
4382 2017-08-16 Martin Liska <mliska@suse.cz>
4385 * config.gcc: Respect previously set extra_objs in case
4388 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4390 PR tree-optimization/81835
4391 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4392 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4393 not depend on the phi.
4395 2017-08-16 Alan Modra <amodra@gmail.com>
4397 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4400 2017-08-16 Alan Modra <amodra@gmail.com>
4402 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4403 (save_reg_p): ..into this. Update all callers.
4404 (first_reg_to_save): Simplify.
4406 2017-08-16 Alan Modra <amodra@gmail.com>
4408 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4411 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4415 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4416 constant count if that count is less than 32.
4418 2017-08-15 Nathan Sidwell <nathan@acm.org>
4420 * gcc.c (execute): Emit friendlier message if inferior is killed
4421 by an external cause.
4423 2017-08-15 Richard Biener <rguenther@suse.de>
4425 PR tree-optimization/81790
4426 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4427 CONSTRUCTORs from simplifying and VN.
4429 2017-08-14 Martin Sebor <msebor@redhat.com>
4431 * builtin-attrs.def: Add comments.
4433 2017-08-14 Martin Sebor <msebor@redhat.com>
4436 * doc/extend.texi (attribute nonstring): Document new attribute.
4438 2017-08-14 Martin Sebor <msebor@redhat.com>
4441 * tree-diagnostic.c (default_tree_printer): Handle %G.
4442 * gimple-pretty-print.h (percent_G_format): Declare new function.
4443 * gimple-pretty-print.c (percent_G_format): Define.
4444 * tree-pretty-print.c (percent_K_format): Add argument.
4446 2017-08-14 Martin Sebor <msebor@redhat.com>
4448 PR translation/79998
4449 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4450 Remove a stray space.
4452 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4455 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4456 (*iordi_1_bts): Ditto.
4457 (*xordi_1_btc): Ditto.
4459 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4462 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4463 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4465 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4466 quoted strings, and make more translator-friendly.
4467 (darwin_rs6000_override_options): Likewise.
4468 (rs6000_option_override_internal): Likewise.
4469 (rs6000_return_in_memory): Fix overlong line.
4470 (init_cmulative_args): Use quoted strings, and make more
4471 translator-friendly.
4472 (rs6000_pass_by_reference): Fix overlong line.
4473 (def_builtin): Use quoted strings.
4474 (altivec_expand_predicate_builtin): Use quoted strings, and make
4475 more translator-friendly.
4476 (htm_expand_builtin): Use quoted strings.
4477 (cpu_expand_builtin): Use quoted strings, and make more
4478 translator-friendly.
4479 (altivec_expand_builtin): Likewise.
4480 (paired_expand_predicate_builtin): Likewise.
4481 (rs6000_invalid_builtin): Likewise.
4482 (builtin_function_type): Use quoted strings.
4483 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4484 more translator-friendly.
4485 (rs6000_trampoline_init): Likewise.
4486 (rs6000_handle_altivec_attribute): Likewise.
4487 (rs6000_inner_target_options): Use quoted strings.
4488 (rs6000_disable_incompatible_switches): Likewise.
4489 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4490 strings, and make more translator-friendly.
4491 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4493 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4495 PR tree-optimization/81799
4496 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4497 cond_expr to simple gimple operand.
4499 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4502 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4504 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4508 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4509 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4510 TARGET_GASISR_PROLOGUES.
4511 * config/avr/avr.c (avr_option_override): Same.
4512 (avr_pass_pre_proep::execute): Same.
4514 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4517 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4518 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4520 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4522 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4523 define_insn_and_split. Split to a memory load from 0 in
4524 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4525 using PTR mode iterator.
4526 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4527 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4528 (*add_tp_<mode>): Redefine as define_insn_and_split.
4529 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4530 address space. Merge with *add_tp_x32 using PTR mode iterator.
4531 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4532 Split to an add with a memory load from 0 in
4533 DEFAULT_TLS_SEG_REG address space.
4535 2017-08-12 Andrew Pinski <apinski@cavium.com>
4537 * config/aarch64/aarch64-option-extensions.def (rdma):
4538 Fix feature string to what Linux prints out in /proc/cpuinfo.
4540 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4543 * dwarf2out.c (modified_type_die): For C typedef types that have
4544 an ultimate origin, process the ultimate origin instead of the
4546 (gen_typedef_die): Assert that input DECLs have no ultimate
4548 (gen_type_die_with_usage): For typedef variants that have an
4549 ultimate origin, just call gen_decl_die on the original DECL.
4550 (process_scope_var): Avoid creating DIEs for local typedefs and
4551 concrete static variables.
4553 2017-08-12 Alan Modra <amodra@gmail.com>
4557 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4558 match gnu-user.h startfile.
4559 (ENDFILE_LINUX_SPEC): Similarly.
4561 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4564 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4566 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4568 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4569 * config/aarch64/aarch64.md (mov<mode>): Change.
4570 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4571 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4572 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4574 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4576 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4577 for storage order barriers.
4579 2017-08-11 Martin Liska <mliska@suse.cz>
4582 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4583 variables of void type.
4585 2017-08-11 Martin Liska <mliska@suse.cz>
4587 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4588 TARGET_SUPPORTS_ALIASES.
4589 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4590 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4591 (optimize_weakref): Likewise.
4592 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4593 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4594 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4596 2017-08-11 Martin Liska <mliska@suse.cz>
4599 * config/i386/i386.c (make_resolver_func): Do complete
4600 refactoring of the function.
4602 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4605 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4606 * config/i386/i386.c (ix86_stack_protect_guard): Use
4607 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4608 * doc/invoke.texi (x86 Options): Document
4609 -mstack-protector-guard-symbol= option.
4611 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4613 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4614 * config/i386/i386.c (ix86_split_stack_guard): New function.
4615 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4616 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4617 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4618 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4619 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4620 (split_stack_space_check): Call ix86_split_stack_guard.
4622 2017-08-10 Martin Sebor <msebor@redhat.com>
4624 * print-tree.c (print_node): Print location using the established
4626 Replace spaces with colons.
4627 (debug_raw, debug): Ditto.
4629 2017-08-10 Martin Sebor <msebor@redhat.com>
4632 * pretty-print.c (pp_format): Correct the handling of %s precision.
4634 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4637 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4639 (ix86_finalize_stack_frame_flags): This. Also clear
4640 frame_pointer_needed if -fno-omit-frame-pointer is used without
4642 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4643 with ix86_finalize_stack_frame_flags.
4644 (ix86_expand_epilogue): Likewise.
4645 (ix86_expand_split_stack_prologue): Likewise.
4646 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4648 2017-08-10 Martin Liska <mliska@suse.cz>
4651 * c-attribs.c (handle_target_attribute):
4652 Report warning for an empty string argument of target attribute.
4654 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4657 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4659 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4660 or DECL_NONLOCAL labels.
4661 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4662 or DECL_NONLOCAL labels here.
4664 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4666 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4667 to indicate when early gimple folding has been disabled.
4668 (rs6000_gimple_fold_builtin): Add debug content.
4669 (rs6000_invalid_builtin): Fix whitespace.
4670 (rs6000_expand_builtin): Fix whitespace.
4671 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4673 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4676 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4677 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4679 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4681 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4682 (qdf24xx): Likewise.
4683 * config/aarch64/aarch64.md: Include falkor.md.
4684 * config/aarch64/falkor.md: New.
4686 2017-08-09 Marek Polacek <polacek@redhat.com>
4689 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4690 * diagnostic.c (emit_diagnostic): Add a comment.
4691 (emit_diagnostic_valist): New function.
4693 2017-08-09 Marek Polacek <polacek@redhat.com>
4696 * input.c (make_location): New overload.
4697 * input.h (make_location): Declare.
4699 2017-08-08 Alan Modra <amodra@gmail.com>
4700 H.J. Lu <hongjiu.lu@intel.com>
4703 * gcc.c (NO_PIE_SPEC): Delete.
4704 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4706 (LINK_PIE_SPEC): ..to here.
4707 (LINK_COMMAND_SPEC): Support -no-pie.
4708 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4709 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4711 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4712 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4713 (ENDFILE_CRTEND_SPEC): Similarly.
4715 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4718 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4719 (mstack-protector-guard-offset=): Ditto.
4720 * config/i386/i386.c (ix86_option_override): Handle
4721 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4723 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4724 ix86_stack_protect_guard_offset variables.
4725 (TARGET_STACK_PROTECT_GUARD): Always define.
4726 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4727 and -mstack-protector-guard-offset= options.
4729 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4731 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4732 boundary case for the last candidate.
4734 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4736 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4737 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4739 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4742 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4743 * config/aarch64/aarch64-builtins.c
4744 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4745 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4746 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4748 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4749 Andrew Pinski <pinskia@gmail.com>
4752 * internal-fn.def (XORSIGN): New.
4753 * optabs.def (xorsign_optab): New.
4754 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4755 (convert_expand_mult_copysign): New.
4756 (pass_optimize_widening_mul::execute): Call
4757 convert_expand_mult_copysign.
4759 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4761 PR tree-optimization/81354
4762 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4763 Insert on edges rather than explicitly creating landing pads.
4764 (analyze_candidates_and_replace): Commit edge inserts.
4766 2017-08-08 Richard Biener <rguenther@suse.de>
4769 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4770 (expand_simple_operations): Also look through ADDR_EXPRs with
4771 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4773 2017-08-08 Richard Biener <rguenther@suse.de>
4775 PR tree-optimization/81723
4776 * tree-vect-slp.c (struct bst_traits): New hash traits.
4777 (bst_fail): New global.
4778 (vect_build_slp_tree_2): New worker, split out from ...
4779 (vect_build_slp_tree): ... this now wrapping it with using
4780 bst_fail set to cache SLP tree build fails. Properly handle
4782 (vect_analyze_slp_instance): Allocate and free bst_fail.
4784 2017-08-08 Martin Liska <mliska@suse.cz>
4787 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4788 LABEL_DECLs that can be from a different function.
4790 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4792 PR tree-optimization/81744
4793 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4794 loop's number of iterations.
4796 2017-08-08 Martin Liska <mliska@suse.cz>
4798 * asan.c: Include header files.
4799 * attribs.c (build_decl_attribute_variant): New function moved
4801 (build_type_attribute_qual_variant): Likewise.
4802 (cmp_attrib_identifiers): Likewise.
4803 (simple_cst_list_equal): Likewise.
4804 (omp_declare_simd_clauses_equal): Likewise.
4805 (attribute_value_equal): Likewise.
4806 (comp_type_attributes): Likewise.
4807 (build_type_attribute_variant): Likewise.
4808 (lookup_ident_attribute): Likewise.
4809 (remove_attribute): Likewise.
4810 (merge_attributes): Likewise.
4811 (merge_type_attributes): Likewise.
4812 (merge_decl_attributes): Likewise.
4813 (merge_dllimport_decl_attributes): Likewise.
4814 (handle_dll_attribute): Likewise.
4815 (attribute_list_equal): Likewise.
4816 (attribute_list_contained): Likewise.
4817 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4818 (lookup_attribute_by_prefix): Likewise.
4819 * bb-reorder.c: Include header files.
4820 * builtins.c: Likewise.
4821 * calls.c: Likewise.
4822 * cfgexpand.c: Likewise.
4823 * cgraph.c: Likewise.
4824 * cgraphunit.c: Likewise.
4825 * convert.c: Likewise.
4826 * dwarf2out.c: Likewise.
4827 * final.c: Likewise.
4828 * fold-const.c: Likewise.
4829 * function.c: Likewise.
4830 * gimple-expr.c: Likewise.
4831 * gimple-fold.c: Likewise.
4832 * gimple-pretty-print.c: Likewise.
4833 * gimple.c: Likewise.
4834 * gimplify.c: Likewise.
4835 * hsa-common.c: Likewise.
4836 * hsa-gen.c: Likewise.
4837 * internal-fn.c: Likewise.
4838 * ipa-chkp.c: Likewise.
4839 * ipa-cp.c: Likewise.
4840 * ipa-devirt.c: Likewise.
4841 * ipa-fnsummary.c: Likewise.
4842 * ipa-inline.c: Likewise.
4843 * ipa-visibility.c: Likewise.
4845 * lto-cgraph.c: Likewise.
4846 * omp-expand.c: Likewise.
4847 * omp-general.c: Likewise.
4848 * omp-low.c: Likewise.
4849 * omp-offload.c: Likewise.
4850 * omp-simd-clone.c: Likewise.
4851 * opts-global.c: Likewise.
4852 * passes.c: Likewise.
4853 * predict.c: Likewise.
4854 * sancov.c: Likewise.
4855 * sanopt.c: Likewise.
4856 * symtab.c: Likewise.
4857 * toplev.c: Likewise.
4858 * trans-mem.c: Likewise.
4859 * tree-chkp.c: Likewise.
4860 * tree-eh.c: Likewise.
4861 * tree-into-ssa.c: Likewise.
4862 * tree-object-size.c: Likewise.
4863 * tree-parloops.c: Likewise.
4864 * tree-profile.c: Likewise.
4865 * tree-ssa-ccp.c: Likewise.
4866 * tree-ssa-live.c: Likewise.
4867 * tree-ssa-loop.c: Likewise.
4868 * tree-ssa-sccvn.c: Likewise.
4869 * tree-ssa-structalias.c: Likewise.
4870 * tree-ssa.c: Likewise.
4871 * tree-streamer-in.c: Likewise.
4872 * tree-vectorizer.c: Likewise.
4873 * tree-vrp.c: Likewise.
4875 * ubsan.c: Likewise.
4876 * varasm.c: Likewise.
4877 * varpool.c: Likewise.
4878 * tree.c: Remove functions moved to attribs.[ch].
4880 * config/aarch64/aarch64.c: Add attrs.h header file.
4881 * config/alpha/alpha.c: Likewise.
4882 * config/arc/arc.c: Likewise.
4883 * config/arm/arm.c: Likewise.
4884 * config/avr/avr.c: Likewise.
4885 * config/bfin/bfin.c: Likewise.
4886 * config/c6x/c6x.c: Likewise.
4887 * config/cr16/cr16.c: Likewise.
4888 * config/cris/cris.c: Likewise.
4889 * config/darwin.c: Likewise.
4890 * config/epiphany/epiphany.c: Likewise.
4891 * config/fr30/fr30.c: Likewise.
4892 * config/frv/frv.c: Likewise.
4893 * config/ft32/ft32.c: Likewise.
4894 * config/h8300/h8300.c: Likewise.
4895 * config/i386/winnt.c: Likewise.
4896 * config/ia64/ia64.c: Likewise.
4897 * config/iq2000/iq2000.c: Likewise.
4898 * config/lm32/lm32.c: Likewise.
4899 * config/m32c/m32c.c: Likewise.
4900 * config/m32r/m32r.c: Likewise.
4901 * config/m68k/m68k.c: Likewise.
4902 * config/mcore/mcore.c: Likewise.
4903 * config/microblaze/microblaze.c: Likewise.
4904 * config/mips/mips.c: Likewise.
4905 * config/mmix/mmix.c: Likewise.
4906 * config/mn10300/mn10300.c: Likewise.
4907 * config/moxie/moxie.c: Likewise.
4908 * config/msp430/msp430.c: Likewise.
4909 * config/nds32/nds32-isr.c: Likewise.
4910 * config/nds32/nds32.c: Likewise.
4911 * config/nios2/nios2.c: Likewise.
4912 * config/nvptx/nvptx.c: Likewise.
4913 * config/pa/pa.c: Likewise.
4914 * config/pdp11/pdp11.c: Likewise.
4915 * config/powerpcspe/powerpcspe.c: Likewise.
4916 * config/riscv/riscv.c: Likewise.
4917 * config/rl78/rl78.c: Likewise.
4918 * config/rx/rx.c: Likewise.
4919 * config/s390/s390.c: Likewise.
4920 * config/sh/sh.c: Likewise.
4921 * config/sol2.c: Likewise.
4922 * config/sparc/sparc.c: Likewise.
4923 * config/spu/spu.c: Likewise.
4924 * config/stormy16/stormy16.c: Likewise.
4925 * config/tilegx/tilegx.c: Likewise.
4926 * config/tilepro/tilepro.c: Likewise.
4927 * config/v850/v850.c: Likewise.
4928 * config/vax/vax.c: Likewise.
4929 * config/visium/visium.c: Likewise.
4930 * config/xtensa/xtensa.c: Likewise.
4932 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4935 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4936 constraints since the -mupper-regs-* switches have been
4938 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4939 into a vector from a double word element that was extracted from
4940 another vector, and eliminate extra XXPERMDI instructions.
4941 (vsx_concat_<mode>_2): Likewise.
4942 (vsx_concat_<mode>_3): Likewise.
4943 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4944 concat to allow optimizing inserts from previous extracts.
4946 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4948 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4949 memory reference to a SSP offset in TLS address space.
4950 (ix86_print_operand) <case '@'>: Remove.
4951 (ix86_print_operand_punct_valid_p): Remove '@' code.
4952 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4954 (stack_tls_protect_set_<mode>): Remove.
4955 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4956 (stack_tls_protect_test_<mode>): Remove.
4957 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4959 2017-08-07 Olivier Hainque <hainque@adacore.com>
4962 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4964 2017-08-07 Douglas Rupp <rupp@adacore.com>
4966 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4967 variable was referenced as multidir in command.
4969 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4972 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4975 2017-08-07 Martin Liska <mliska@suse.cz>
4977 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4978 * config/rl78/rl78.c: Add include of attribs.h.
4979 * config/sh/sh.c: Likewise.
4980 * config/v850/v850.c: Likewise.
4982 2017-08-07 Tom de Vries <tom@codesourcery.com>
4985 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4987 2017-08-07 Martin Liska <mliska@suse.cz>
4989 * config/mips/mips.c: Include attribs.h.
4991 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4994 * doc/invoke.texi: Document change in behvaior for -Ofast for
4997 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4999 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5001 (aarch64_pop_regs): Likewise.
5002 (aarch64_gen_load_pair): Likewise.
5003 (aarch64_save_callee_saves): Likewise.
5004 (aarch64_restore_callee_saves): Likewise.
5006 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5008 * config/i386/i386.c: Revert the last change.
5010 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5013 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5015 (ix86_finalize_stack_frame_flags): This. Also clear
5016 frame_pointer_needed if -fno-omit-frame-pointer is used without
5018 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5019 with ix86_finalize_stack_frame_flags.
5020 (ix86_expand_epilogue): Likewise.
5021 (ix86_expand_split_stack_prologue): Likewise.
5023 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5026 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5027 to P_AES for Westmere.
5029 2017-08-07 Jonathan Yong <10walls@gmail.com>
5031 * config/i386/mingw.opt (fset-stack-executable): Removed.
5032 * config/i386/cygming.opt (fset-stack-executable): Moved
5034 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5036 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5038 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5040 2017-08-07 Marek Polacek <polacek@redhat.com>
5043 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5045 2017-08-07 Martin Liska <mliska@suse.cz>
5047 * attribs.h (canonicalize_attr_name): New function.
5048 (cmp_attribs): Move from c-format.c and adjusted.
5049 (is_attribute_p): Moved from tree.h.
5050 * tree-inline.c: Add new includes.
5051 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5052 (private_is_attribute_p): Remove.
5053 (private_lookup_attribute): Likewise.
5054 (private_lookup_attribute_by_prefix): Simplify.
5055 (remove_attribute): Use is_attribute_p.
5056 * tree.h: Remove removed declarations.
5058 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5061 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5062 instead of computing it in the function. Formatting fix.
5063 (expand_case): Don't rely on default_edge being the first edge,
5064 clear it if removing it, pass default_edge to
5065 emit_case_dispatch_table.
5066 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5069 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5071 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5072 insn in the function, emit NOP after the insn.
5074 2017-08-06 Tom de Vries <tom@codesourcery.com>
5076 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5079 2017-08-06 Tom de Vries <tom@codesourcery.com>
5081 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5084 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5086 PR tree-optimization/57371
5087 * match.pd: New pattern.
5089 2017-08-04 Marek Polacek <polacek@redhat.com>
5092 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5093 perform the computation in offset_int.
5095 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5097 PR tree-optimization/81136
5098 * tree-vectorizer.h: Include tree-hash-traits.h.
5099 (vec_base_alignments): New typedef.
5100 (vec_info): Add a base_alignments field.
5101 (vect_record_base_alignments): Declare.
5102 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5104 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5105 (create_data_ref): Add an is_conditional_in_stmt argument.
5106 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5107 the is_conditional_in_stmt field.
5108 (data_ref_loc): Add an is_conditional_in_stmt field.
5109 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5110 (find_data_references_in_stmt): Update call to create_data_ref.
5111 (graphite_find_data_references_in_stmt): Likewise.
5112 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5113 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5114 (vect_record_base_alignment): New function.
5115 (vect_record_base_alignments): Likewise.
5116 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5117 for nested statements even if we fail to compute a misalignment.
5118 Use pooled base alignments for unconditional references.
5119 (vect_find_same_alignment_drs): Compare base addresses instead
5121 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5122 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5124 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5126 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5127 Add an explicit name for the enum. Use auto_vec for slp_instances
5129 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5131 (_bb_vec_info): Add a constructor and destructor.
5132 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5133 (destroy_loop_vec_info): Delete.
5134 (vect_destroy_datarefs): Likewise.
5135 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5136 (vec_info::vec_info): New function.
5137 (vec_info::~vec_info): Likewise.
5138 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5139 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5140 destroy_loop_vec_info.
5141 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5142 (_loop_vec_info::_loop_vec_info): ...this.
5143 (destroy_loop_vec_info): Replace with...
5144 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5145 the stmt_vec_infos. Leave handling of vec_info information to its
5146 destructor. Remove explicit vector releases.
5147 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5148 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5149 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5150 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5151 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5152 (destroy_bb_vec_info): Replace with...
5153 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5154 information to its destructor.
5155 (vect_slp_analyze_bb_1): Use new and delete instead of
5156 new_bb_vec_info and destroy_bb_vec_info.
5157 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5160 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5162 * tree-data-ref.h (subscript): Add access_fn field.
5163 (data_dependence_relation): Add could_be_independent_p.
5164 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5165 (same_access_functions): Move to tree-data-ref.c.
5166 * tree-data-ref.c (ref_contains_union_access_p): New function.
5167 (access_fn_component_p): Likewise.
5168 (access_fn_components_comparable_p): Likewise.
5169 (dr_analyze_indices): Add a reference to access_fn_component_p.
5170 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5172 (constant_access_functions): Likewise.
5173 (add_other_self_distances): Likewise.
5174 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5175 (initialize_data_dependence_relation): Use XCNEW and remove
5176 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5177 of access functions that have the same type. Allow the
5178 subsequence to end with different bases in some circumstances.
5179 Record the chosen access functions in SUB_ACCESS_FN.
5180 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5181 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5182 (subscript_dependence_tester_1): Likewise dra and drb.
5183 (build_classic_dist_vector): Update calls accordingly.
5184 (subscript_dependence_tester): Likewise.
5185 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5186 DDR_COULD_BE_INDEPENDENT_P.
5187 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5188 comp_alias_ddrs instead of may_alias_ddrs.
5189 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5191 (vect_analyze_data_ref_dependence): Use it if
5192 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5193 distance vectors if that fails.
5194 (dependence_distance_ge_vf): New function.
5195 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5196 LOOP_VINFO_MAY_ALIAS_DDRS.
5198 2017-08-04 Richard Biener <rguenther@suse.de>
5201 * fold-const.c (fold_binary_loc): Properly restrict
5202 minus_var0 && minus_var1 case when associating undefined overflow
5205 2017-08-04 Tom de Vries <tom@codesourcery.com>
5207 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5209 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5211 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5212 Don't start diagnostic messages with a capital letter.
5213 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5215 (rs6000_invalid_builtin): Likewise.
5216 (rs6000_trampoline_init): Likewise.
5218 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5221 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5222 after setting changeable df flags.
5224 2017-08-03 Richard Biener <rguenther@suse.de>
5226 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5227 up if the use is in USE - X.
5229 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5231 * toplev.c (dumpfile.h): New include.
5232 (internal_error_reentered): New static function. Use it...
5233 (internal_error_function): ...here to handle reentered internal_error.
5235 2017-08-03 Richard Biener <rguenther@suse.de>
5238 * fold-const.c (split_tree): Add minus_var and minus_con
5239 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5240 here but always use minus_*.
5241 (associate_trees): Assert we never associate with MINUS_EXPR
5242 and NULL first operand. Do not recurse for PLUS_EXPR operands
5243 when associating as MINUS_EXPR either.
5244 (fold_binary_loc): Track minus_var and minus_con.
5246 2017-08-03 Tom de Vries <tom@codesourcery.com>
5249 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5250 ACCEL_COMPILER, apply finish_options on
5251 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5253 2017-08-03 Tom de Vries <tom@codesourcery.com>
5256 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5257 function_entry_patch_area_size > 0.
5259 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5262 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5263 instead of 10??LU, perform unit multiplication in wide_int,
5264 don't change alloc_object_size_limit if the limit is larger
5267 PR tree-optimization/81655
5268 PR tree-optimization/81588
5269 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5270 the case when ranges[i].low and high are 1 for unsigned type with
5274 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5275 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5277 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5279 * tree-vrp.h: Add include guard.
5281 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5284 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5285 (ud2): New insn pattern.
5286 * config/i386/i386.c (ix86_expand_epilogue):
5287 For naked functions, generate ud2 instead of trap insn.
5289 2017-08-02 Marek Polacek <polacek@redhat.com>
5292 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5294 2017-08-02 Tom de Vries <tom@codesourcery.com>
5295 Cesar Philippidis <cesar@codesourcery.com>
5297 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5298 Add missing edge probabilities.
5300 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5302 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5305 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5308 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5309 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5310 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5311 prologue_seq sequences - if any.
5313 2017-08-02 Richard Biener <rguenther@suse.de>
5315 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5316 via vectors if supported, integer extracts via punning if supported
5317 or otherwise vector extracts.
5319 2017-08-02 Richard Biener <rguenther@suse.de>
5321 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5323 (bitmap_insert_into_set): ... this.
5325 2017-08-02 Richard Biener <rguenther@suse.de>
5327 PR tree-optimization/81633
5329 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5331 PR tree-optimization/71752
5332 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5334 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5336 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5337 (machine_function::call_ms2sysv_pad_out): Remove field.
5338 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5339 (ix86_compute_frame_layout): Likewise.
5341 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5344 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5345 attribute with interrupt attribute.
5347 2017-08-01 Andrew Pinski <apinski@cavium.com>
5349 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5350 BIT_INSERT_EXPR's operand 1
5351 to see if the types precision matches.
5353 2017-08-01 Martin Liska <mliska@suse.cz>
5356 * builtins.c (expand_builtin_memcpy_args): Remove.
5357 (expand_builtin_memcpy): Call newly added function
5358 expand_builtin_memory_copy_args.
5359 (expand_builtin_memcpy_with_bounds): Likewise.
5360 (expand_builtin_mempcpy): Remove last argument.
5361 (expand_builtin_mempcpy_with_bounds): Likewise.
5362 (expand_builtin_memory_copy_args): New function created from
5363 expand_builtin_mempcpy_args with small modifications.
5364 (expand_builtin_mempcpy_args): Remove.
5365 (expand_builtin_stpcpy): Remove unused argument.
5366 (expand_builtin): Likewise.
5367 (expand_builtin_with_bounds): Likewise.
5369 2017-08-01 Martin Liska <mliska@suse.cz>
5372 Make mempcpy more optimal (PR middle-end/70140).
5374 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5377 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5378 __builtin_vec_cmpne verify both arguments are compatible vectors
5379 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5380 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5381 move computation of aligned to after checking the argument types.
5385 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5386 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5389 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5390 Doug Rupp <rupp@adacore.com>
5391 Olivier Hainque <hainque@adacore.com>
5393 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5394 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5396 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5397 for TARGET_OS_CPP_BUILTIN.
5398 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5399 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5401 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5402 passing required abi options to the assembler for EABI configurations.
5403 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5404 of .text.hot and .text.unlikely sections for kernel modules when
5405 using ARM style exceptions.
5406 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5407 options. Add EXTRA_CC1_SPEC.
5408 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5410 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5412 (ARM_TARGET2_DWARF_FORMAT): Define.
5413 * config/arm/t-vxworks: Adjust multilib control to removal of the
5414 Diab command line options.
5416 2017-08-01 Martin Liska <mliska@suse.cz>
5418 PR gcov-profile/81561
5419 * gcov.c (unblock): Make unblocking safe as we need to preserve
5420 index correspondence of blocks and block_lists.
5422 2017-08-01 Richard Biener <rguenther@suse.de>
5424 PR tree-optimization/81181
5425 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5426 (compute_antic): ... end of iteration here.
5428 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5430 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5431 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5432 (ftree-slp-vectorize): Likewise.
5433 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5434 can no longer be set independent of flag_tree_loop_vectorize.
5435 * omp-general.c (emp_max_vf): Likewise.
5436 * opts.c (enable_fdo_optimizations): Remove references to
5437 flag_tree_vectorize, these are now implicit.
5438 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5439 and leave it for the options machinery.
5441 2017-08-01 Martin Liska <mliska@suse.cz>
5444 * builtins.c (expand_builtin_memcpy_args): Remove.
5445 (expand_builtin_memcpy): Call newly added function
5446 expand_builtin_memory_copy_args.
5447 (expand_builtin_memcpy_with_bounds): Likewise.
5448 (expand_builtin_mempcpy): Remove last argument.
5449 (expand_builtin_mempcpy_with_bounds): Likewise.
5450 (expand_builtin_memory_copy_args): New function created from
5451 expand_builtin_mempcpy_args with small modifications.
5452 (expand_builtin_mempcpy_args): Remove.
5453 (expand_builtin_stpcpy): Remove unused argument.
5454 (expand_builtin): Likewise.
5455 (expand_builtin_with_bounds): Likewise.
5457 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5460 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5461 print "ds:" only for immediates in generic address space.
5463 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5466 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5467 (ix86_function_ok_for_sibcall): Return false for naked functions.
5469 2017-08-01 Richard Biener <rguenther@suse.de>
5471 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5472 (compute_antic): Seed worklist with exit block predecessors.
5473 * cfganal.c (dfs_find_deadend): For a cycle return the source
5474 of the edge closing it.
5476 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5478 * config/aarch64/aarch64.c
5479 (aarch64_can_const_movi_rtx_p): Move 0 check.
5481 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5483 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5484 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5486 * match.pd: Ditto in address comparison pattern.
5488 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5490 PR tree-optimization/81627
5491 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5492 closed ssa form for store-store chain.
5494 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5496 PR tree-optimization/81620
5497 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5498 for store-store chain.
5500 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5502 PR tree-optimization/81588
5503 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5504 ranges[i].in_p, invert comparison code ccode. For >/>=,
5505 swap rhs1 and rhs2 and comparison code unconditionally,
5506 for </<= don't do that. Don't swap rhs1/rhs2 again if
5507 ranges[i].in_p, instead invert comparison code ccode if
5508 opcode or oe->rank is BIT_IOR_EXPR.
5511 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5512 a direct optab to conversion optab.
5513 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5514 with GET_MODE_INNER as last argument instead of optab_handler.
5515 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5516 vector extraction if possible and optab is available.
5517 * expr.c (store_constructor): Use convert_optab_handler instead
5518 of optab_handler. Use vector initialization from smaller
5519 vectors if possible and optab is available.
5520 * tree-vect-stmts.c (vectorizable_load): Likewise.
5521 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5523 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5524 of vec_init from half-sized vectors with the same element mode.
5525 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5526 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5527 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5528 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5529 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5530 after mode in gen_vec_extract* calls.
5531 (vec_extract<mode>): Renamed to ...
5532 (vec_extract<mode><ssescalarmodelower>): ... this.
5533 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5534 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5535 element mode after mode in gen_vec_init* calls.
5536 (VEC_INIT_HALF_MODE): New mode iterator.
5537 (vec_init<mode>): Renamed to ...
5538 (vec_init<mode><ssescalarmodelower>): ... this.
5539 (vec_init<mode><ssehalfvecmodelower>): New expander.
5540 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5541 (vec_extractv2sfsf): ... this.
5542 (vec_initv2sf): Renamed to ...
5543 (vec_initv2sfsf): ... this.
5544 (vec_extractv2si): Renamed to ...
5545 (vec_extractv2sisi): ... this.
5546 (vec_initv2si): Renamed to ...
5547 (vec_initv2sisi): ... this.
5548 (vec_extractv4hi): Renamed to ...
5549 (vec_extractv4hihi): ... this.
5550 (vec_initv4hi): Renamed to ...
5551 (vec_initv4hihi): ... this.
5552 (vec_extractv8qi): Renamed to ...
5553 (vec_extractv8qiqi): ... this.
5554 (vec_initv8qi): Renamed to ...
5555 (vec_initv8qiqi): ... this.
5556 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5557 (vec_init<mode>): Renamed to ...
5558 (vec_init<mode><VEC_base_l>): ... this.
5559 (vec_extract<mode>): Renamed to ...
5560 (vec_extract<mode><VEC_base_l>): ... this.
5561 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5562 (vec_initv2sfsf): ... this.
5563 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5564 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5565 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5566 element mode after mode in gen_vec_init* calls.
5567 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5568 (vec_init<mode><Vel>): ... this.
5569 (vec_extract<mode>): Renamed to ...
5570 (vec_extract<mode><Vel>): ... this.
5571 * config/aarch64/iterators.md (Vel): New mode attribute.
5572 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5573 Add element mode after mode in gen_vec_extract* calls.
5574 * config/s390/vector.md (non_vec_l): New mode attribute.
5575 (vec_extract<mode>): Renamed to ...
5576 (vec_extract<mode><non_vec_l>): ... this.
5577 (vec_init<mode>): Renamed to ...
5578 (vec_init<mode><non_vec_l>): ... this.
5579 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5580 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5582 * config/arm/iterators.md (V_elem_l): New mode attribute.
5583 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5584 (vec_extract<mode><V_elem_l>): ... this.
5585 (vec_extractv2di): Renamed to ...
5586 (vec_extractv2didi): ... this.
5587 (vec_init<mode>): Renamed to ...
5588 (vec_init<mode><V_elem_l>): ... this.
5589 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5590 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5591 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5592 Add element mode after gen_vec_extract* calls.
5593 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5594 (vec_init<mode><unitmode>): ... this.
5595 (vec_extract<mode>): Renamed to ...
5596 (vec_extract<mode><unitmode>): ... this.
5597 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5598 (vec_init<mode><unitmode>): ... this.
5599 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5600 (vec_initv2sfsf): ... this.
5601 (vec_extractv2sf): Renamed to ...
5602 (vec_extractv2sfsf): ... this.
5603 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5604 Add element mode after gen_vec_extract* calls.
5605 * config/mips/mips.md (unitmode): New mode iterator.
5606 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5607 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5608 * config/spu/spu.md (inner_l): New mode attribute.
5609 (vec_init<mode>): Renamed to ...
5610 (vec_init<mode><inner_l>): ... this.
5611 (vec_extract<mode>): Renamed to ...
5612 (vec_extract<mode><inner_l>): ... this.
5613 * config/sparc/sparc.md (veltmode): New mode iterator.
5614 (vec_init<VMALL:mode>): Renamed to ...
5615 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5616 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5617 (vec_initv2sisi): ... this.
5618 (vec_initv2sf): Renamed to ...
5619 (vec_initv2sfsf): ... this.
5620 (vec_extractv2sf): Renamed to ...
5621 (vec_extractv2sfsf): ... this.
5622 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5623 (vec_init<mode>): Renamed to ...
5624 (vec_init<mode><VEC_base_l>): ... this.
5625 (vec_extract<mode>): Renamed to ...
5626 (vec_extract<mode><VEC_base_l>): ... this.
5627 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5628 (vec_initv2sfsf): ... this.
5629 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5630 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5631 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5632 gen_vec_init* calls.
5634 2017-08-01 Richard Biener <rguenther@suse.de>
5636 PR tree-optimization/81297
5637 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5638 TREE_OVERFLOW from INTEGER_CSTs.
5640 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5642 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5644 2017-07-31 Carl Love <cel@us.ibm.com>
5646 * config/rs6000/rs6000-c: Add support for built-in functions
5647 vector signed char vec_xl_be (signed long long, signed char *);
5648 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5649 vector signed int vec_xl_be (signed long long, signed int *);
5650 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5651 vector signed long long vec_xl_be (signed long long, signed long long *);
5652 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5653 vector signed short vec_xl_be (signed long long, signed short *);
5654 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5655 vector double vec_xl_be (signed long long, double *);
5656 vector float vec_xl_be (signed long long, float *);
5657 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5658 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5659 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5661 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5662 (altivec_expand_builtin): Add switch statement to call
5663 altivec_expand_xl_be for each builtin.
5664 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5665 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5666 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5667 __builtin_vsx_le_be_v16qi.
5668 * doc/extend.texi: Update the built-in documentation file for the
5669 new built-in functions.
5671 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5674 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5676 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5678 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5680 * config.gcc: Add z14.
5681 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5682 CPU model numbers for z13s and z14.
5683 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5685 * config/s390/s390-opts.h (enum processor_type): Rename
5686 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5687 * config/s390/s390.c (processor_table): Add field for CPU name to
5688 be passed to Binutils.
5689 (s390_asm_output_machine_for_arch): Use the new field in
5690 processor_table for Binutils.
5691 (s390_expand_builtin): Replace arch12 with z14.
5692 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5693 (s390_get_sched_attrmask): Likewise.
5694 (s390_get_unit_mask): Likewise.
5695 * config/s390/s390.opt: Add z14 to processor_type enum.
5697 2017-07-31 Martin Jambor <mjambor@suse.cz>
5700 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5701 regardless of optimization level.
5703 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5704 Martin Liska <mliska@suse.cz>
5706 * predict.def: Remove old comment and adjust probability.
5707 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5710 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5713 * config/i386/i386.c (ix86_function_naked): New function.
5714 (ix86_can_use_return_insn_p): Return false for naked functions.
5715 (ix86_expand_prologue): Skip prologue for naked functions.
5716 (ix86_expand_epilogue): Skip epilogue for naked functions
5717 and emit trap instruction.
5718 (ix86_warn_func_return): New function.
5719 (ix86_attribute_table): Add "naked" attribute specification.
5720 (TARGET_WARN_FUNC_RETURN): Define.
5721 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5723 2017-07-31 Martin Liska <mliska@suse.cz>
5725 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5726 (dump_gimple_bb_header): Always dump BB info.
5727 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5729 2017-07-31 Martin Liska <mliska@suse.cz>
5732 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5733 also with current_function_decl non-null equality.
5735 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5738 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5739 change type to the element type, instead add eltype variable and
5740 use it where we are interested in the element type.
5742 PR tree-optimization/81603
5743 * ipa-polymorphic-call.c
5744 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5745 offset arithmetic in offset_int, bail out if the resulting bit offset
5746 doesn't fit into shwi.
5748 2017-07-31 Martin Liska <mliska@suse.cz>
5750 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5751 (gimplify_save_expr): Fix comment.
5753 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5756 * config/i386/i386.c (ix86_function_arg): Update arguments for
5758 (ix86_compute_frame_layout): Set the initial stack offset to
5759 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5760 INCOMING_FRAME_SP_OFFSET.
5761 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5762 stack before exception handler returns.
5763 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5764 the 'ERROR_CODE' for exception handler.
5766 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5768 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5769 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5770 (ASM_OUTPUT_REG_POP): Ditto.
5771 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5772 instead of asm_fprintf to output pure string.
5774 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5776 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5777 to imported_module_or_decl hook.
5778 (debug_nothing_tree_tree_tree_bool): Remove.
5779 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5780 * debug.c (do_nothing_debug_hooks): Use
5781 debug_nothing_tree_tree_tree_bool_bool instead of
5782 debug_nothing_tree_tree_tree_bool.
5783 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5784 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5785 * sdbout.c (sdb_debug_hooks): Likewise.
5786 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5787 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5789 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5790 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5791 attribute, don't add anything.
5793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5795 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5796 (fold_build2_stat_loc): Likewise.
5797 (fold_build3_stat_loc): Likewise.
5798 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5799 (fold_build1_loc): Remove macro.
5800 (fold_build2_loc): Likewise.
5801 (fold_build3_loc): Likewise.
5803 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5805 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5806 (gimple_build_debug_bind_source_stat): Likewise.
5807 * gimple.h (gimple_build_debug_bind): Remove macro.
5808 (gimple_build_debug_bind_source): Likewise.
5810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5812 * bitmap.c (bitmap_alloc): Adjust.
5813 (bitmap_gc_alloc): Likewise.
5814 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5816 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5818 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5819 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5820 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5821 (bitmap_gc_alloc_stat): Likewise.
5822 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5826 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5827 * rtl.h (shallow_copy_rtx): Remove macro.
5829 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5831 * emit-rtl.c (gen_raw_REG): Adjust.
5832 * gengenrtl.c (gendef): Likewise.
5833 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5834 * rtl.h (rtx_alloc): Remove macro.
5836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5838 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5839 (build_tree_list_stat): Likewise.
5840 * tree.h (build_tree_list): Remove macro.
5841 (build_tree_list_vec): Likewise.
5843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5845 * tree.c (make_vector_stat): Remove _stat from name.
5846 (build_vector_stat): Likewise.
5847 * tree.h (make_vector_stat): Remove macro.
5848 (build_vector_stat): Likewise.
5850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5852 * tree.h (build_var_debug_value): Remove prototype.
5854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5856 * tree.c (tree_cons_stat): Remove _stat from name.
5857 * tree.h (tree_cons): Remove macro.
5859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5861 * tree.c (build_vl_exp_stat): Remove _stat from name.
5862 * tree.h (build_vl_exp): Remove macro.
5864 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5866 * tree.c (build_decl_stat): Remove _stat from name.
5867 * tree.h (build_decl): Remove macro.
5869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5871 * gimple.c (gimple_build_with_ops_stat): Adjust.
5872 (gimple_alloc_stat): Remove _stat from name.
5873 * gimple.h (gimple_alloc): Remove macro.
5875 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5877 * tree.c (make_tree_vec_stat): Remove _stat from name.
5878 (grow_tree_vec_stat): Likewise.
5879 * tree.h (make_tree_vec_stat): Adjust prototype.
5880 (grow_tree_vec_stat): Likewise.
5881 (make_tree_vec): Remove macro.
5882 (grow_tree_vec): Likewise.
5884 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5886 * fold-const.c (fold_build1_stat_loc): Adjust.
5887 (fold_build2_stat_loc): Likewise.
5888 (fold_build3_stat_loc): Likewise.
5889 * tree.c (build0_stat): Remove _stat from name.
5890 (build1_stat): Likewise.
5891 (build2_stat): Likewise.
5892 (build3_stat): Likewise.
5893 (build4_stat): Likewise.
5894 (build5_stat): Likewise.
5895 * tree.h (build1_loc): Remove macro, and rename _stat function
5897 (build2_loc): Likewise.
5898 (build3_loc): Likewise.
5899 (build4_loc): Likewise.
5900 (build5_loc): Likewise.
5902 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5904 * tree.c (make_int_cst_stat): Remove _stat from name.
5905 * tree.h (make_int_cst_stat): Adjust prototype.
5906 (make_int_cst): Remove macro.
5908 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5910 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5911 * tree.h (make_tree_binfo_stat): Adjust prototype.
5912 (make_tree_binfo): Remove.
5914 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5916 * tree.c (copy_node_stat): Rename to copy_node.
5917 (build_distinct_type_copy): Adjust.
5918 * tree.h (copy_node_stat): Adjust prototype.
5919 (copy_node): Remove macro.
5921 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5923 * tree.c (make_node_stat): rename to make_node.
5924 (build_tree_list_stat): Adjust.
5925 (build0_stat): Likewise.
5926 (build2_stat): Likewise.
5927 (build3_stat): Likewise.
5928 (build4_stat): Likewise.
5929 (build5_stat): Likewise.
5930 (build_decl_stat): Likewise.
5931 * tree.h (make_node_stat): Adjust prototype.
5932 (make_node): remove macro.
5934 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5936 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5937 (PPC_FEATURE2_SCV): Likewise.
5938 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5940 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5942 * config/aarch64/aarch64.c
5943 (aarch64_internal_mov_immediate): Add new special pattern.
5944 * config/aarch64/aarch64.md (*movdi_aarch64):
5945 Add reg/32bit const mov case.
5947 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5948 Richard Sandiford <richard.sandiford@linaro.org>
5950 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5951 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5952 Add integer and movi cases.
5953 (movi-split-hf-df-sf split, fp16): New.
5954 (enabled): Added TARGET_FP_F16INST.
5955 * config/aarch64/iterators.md (GPF_HF): New.
5956 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5958 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5960 * config/aarch64/aarch64.c
5961 (aarch64_simd_container_mode): Add prototype.
5962 (aarch64_expand_mov_immediate): Add HI support.
5963 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5964 (aarch64_can_const_movi_rtx_p): New.
5965 (aarch64_preferred_reload_class):
5966 Remove restrictions of using FP registers for certain SIMD operations.
5967 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5968 (aarch64_valid_floating_const): Add integer move validation.
5969 (aarch64_simd_imm_scalar_p): Remove.
5970 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5971 (aarch64_legitimate_constant_p): Expand list of supported cases.
5972 * config/aarch64/aarch64-protos.h
5973 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5974 (aarch64_reinterpret_float_as_int): New.
5975 (aarch64_simd_imm_scalar_p): Remove.
5976 * config/aarch64/constraints.md (Uvi): New.
5977 (Dd): Split into Ds and new Dd.
5978 * config/aarch64/aarch64.md (*movsi_aarch64):
5980 (*movdi_aarch64): Add SIMD mov case.
5982 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5984 * tree-predcom.c: (struct chain): Handle store-store chain in which
5985 stores for elimination only store loop invariant values.
5986 (execute_pred_commoning_chain): Ditto.
5987 (prepare_initializers_chain_store_elim): Ditto.
5988 (prepare_finalizers): Ditto.
5989 (is_inv_store_elimination_chain): New function.
5990 (initialize_root_vars_store_elim_1): New function.
5992 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5994 * tree-predcom.c: Revise general description of the pass.
5995 (enum chain_type): New enum type for store elimination.
5996 (struct chain): New field supporting store elimination.
5997 (struct component): Ditto.
5998 (dump_chain): Dump store-stores chain.
5999 (release_chain): Release resources.
6000 (split_data_refs_to_components): Compute and create component
6001 contains only stores for elimination.
6002 (get_chain_last_ref_at): New function.
6003 (make_invariant_chain): Initialization.
6004 (make_rooted_chain): Specify chain type in parameter and record it.
6005 (add_looparound_copies): Skip for store-stores chain.
6006 (determine_roots_comp): Compute type of chain and pass it to
6008 (initialize_root_vars_store_elim_2): New function.
6009 (finalize_eliminated_stores): New function.
6010 (remove_stmt): Handle store for elimination.
6011 (execute_pred_commoning_chain): Execute predictive commoning on
6013 (determine_unroll_factor): Skip unroll for store-stores chain.
6014 (prepare_initializers_chain_store_elim): New function.
6015 (prepare_initializers_chain): Hanlde store-store chain.
6016 (prepare_finalizers_chain, prepare_finalizers): New function.
6017 (tree_predictive_commoning_loop): Return integer value indicating
6018 if loop is unrolled or lcssa form is corrupted.
6019 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6021 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6023 * tree-predcom.c (initialize_root): Delete.
6024 (execute_pred_commoning_chain): Initialize root vars and replace
6025 reference of non-combined chain directly, rather than call above
6028 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6030 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6031 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6033 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6035 * tree-predcom.c (struct chain): New field init_seq.
6036 (release_chain): Release init_seq.
6037 (prepare_initializers_chain): Record intialization stmts in above
6039 (insert_init_seqs): New function.
6040 (tree_predictive_commoning_loop): Call insert_init_seqs.
6042 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6044 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6046 2017-07-28 Richard Biener <rguenther@suse.de>
6048 * match.pd: Remove superfluous :c.
6049 * genmatch.c (simplify::id): Add member.
6050 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6052 (current_id): New global.
6053 (dt_node::parent): Move from ...
6054 (dt_operand::parent): ... here. Add for_id member.
6055 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6056 (decision_tree::find_node): Relax order requirement when
6057 merging DT_TRUE nodes to ones inbetween the current simplify
6058 and the one we try to merge with. Add diagnostic whenever
6059 we need to enforce pattern order by not merging.
6060 (decision_tree::insert): Set current_id.
6061 (decision_tree::print_node): Dump parent node and for_id.
6062 (parser::last_id): Add member.
6063 (parser::push_simplify): Assign unique id.
6064 (parser::parser): Initialize last_id.
6066 2017-07-28 Martin Liska <mliska@suse.cz>
6069 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6070 gimple_build_debug_bind.
6072 2017-07-28 Richard Biener <rguenther@suse.de>
6074 PR tree-optimization/81502
6075 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6077 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6079 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6080 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6081 for BIT_FIELD_REF args.
6082 * fold-const.c (make_bit_field_ref): Likewise.
6083 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6085 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6088 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6089 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6090 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6091 Or it into SANITIZER_UNDEFINED.
6092 * ubsan.c: Include gimple-fold.h and varasm.h.
6093 (ubsan_expand_ptr_ifn): New function.
6094 (instrument_pointer_overflow): New function.
6095 (maybe_instrument_pointer_overflow): New function.
6096 (instrument_object_size): Formatting fix.
6097 (pass_ubsan::execute): Call instrument_pointer_overflow
6098 and maybe_instrument_pointer_overflow.
6099 * internal-fn.c (expand_UBSAN_PTR): New function.
6100 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6101 * sanitizer.def (__ubsan_handle_pointer_overflow,
6102 __ubsan_handle_pointer_overflow_abort): New builtins.
6103 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6104 * internal-fn.def (UBSAN_PTR): New internal function.
6105 * opts.c (sanitizer_opts): Add pointer-overflow.
6106 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6107 * fold-const.c (build_range_check): Compute pointer range check in
6108 integral type if pointer arithmetics would be needed. Formatting
6111 2017-07-28 Martin Liska <mliska@suse.cz>
6114 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6115 parameters that are of a variable-length.
6117 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6119 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6121 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6122 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6123 (CRT_CALL_STATIC_FUNCTION): Likewise.
6124 (ASM_DEFAULT_SPEC): New define.
6125 (ASM_SPEC32): Likewise.
6126 (ASM_SPEC64): Likewise.
6127 (ASM_SPEC_COMMON): Likewise.
6128 (ASM_SPEC): Likewise.
6129 (INVALID_64BIT): Likewise.
6130 (LINK_OS_DEFAULT_SPEC): Likewise.
6131 (LINK_OS_SPEC32): Likewise.
6132 (LINK_OS_SPEC64): Likewise.
6133 (POWERPC_LINUX): Likewise.
6134 (PTRDIFF_TYPE): Likewise.
6135 (RESTORE_FP_PREFIX): Likewise.
6136 (RESTORE_FP_SUFFIX): Likewise.
6137 (SAVE_FP_PREFIX): Likewise.
6138 (SAVE_FP_SUFFIX): Likewise.
6139 (SIZE_TYPE): Likewise.
6140 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6141 (TARGET_64BIT): Likewise.
6142 (TARGET_64BIT): Likewise.
6143 (TARGET_AIX): Likewise.
6144 (WCHAR_TYPE_SIZE): Likewise.
6145 (WCHAR_TYPE): Undefine.
6146 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6147 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6148 (CPP_OS_RTEMS_SPEC): Delete.
6149 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6150 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6152 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6154 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6156 PR tree-optimization/81578
6157 * tree-parloops.c (build_new_reduction): Bail out if
6158 reduction_code isn't one of the standard OpenMP reductions.
6159 Move the details printing after that decision.
6161 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6163 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6164 related to reload_in_progress.
6165 (splat_input_operand): Likewise.
6166 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6168 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6170 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6171 (TARGET_INSTANTIATE_DECLS): Likewise.
6172 (legitimate_indexed_address_p): Delete reload_in_progress code.
6173 (rs6000_debug_legitimate_address_p): Likewise.
6174 (rs6000_eliminate_indexed_memrefs): Likewise.
6175 (rs6000_emit_le_vsx_store): Likewise.
6176 (rs6000_emit_move_si_sf_subreg): Likewise.
6177 (rs6000_emit_move): Likewise.
6178 (register_to_reg_type): Likewise.
6179 (rs6000_pre_atomic_barrier): Likewise.
6180 (rs6000_machopic_legitimize_pic_address): Likewise.
6181 (rs6000_allocate_stack_temp): Likewise.
6182 (rs6000_address_for_fpconvert): Likewise.
6183 (rs6000_address_for_altivec): Likewise.
6184 (rs6000_secondary_memory_needed_rtx): Delete function.
6185 (rs6000_check_sdmode): Likewise.
6186 (rs6000_alloc_sdmode_stack_slot): Likewise.
6187 (rs6000_instantiate_decls): Likewise.
6188 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6189 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6190 Delete reload_in_progress.
6191 (*vec_reload_and_plus_<mptrsize>): Likewise.
6192 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6193 (vsx_div_v2di): Likewise.
6194 (vsx_udiv_v2di): Likewise.
6196 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6198 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6199 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6200 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6201 (rs6000_debug_reg_global): Delete print of LRA status.
6202 (rs6000_option_override_internal): Delete dead LRA related code.
6203 (rs6000_lra_p): Delete function.
6204 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6206 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6208 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6209 * config/riscv/rtems.h: New file.
6211 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6212 Sudakshina Das <sudi.das@arm.com>
6214 * config/aarch64/aarch64.md
6215 (define_split for and<mode>3nr_compare): Move
6216 non aarch64_logical_operand to a register.
6217 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6218 register immediate operand to a register.
6219 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6221 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6224 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6226 2017-07-27 Richard Biener <rguenther@suse.de>
6228 PR tree-optimization/81573
6229 PR tree-optimization/81494
6230 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6231 multi defuse cycle case.
6233 2017-07-27 Richard Biener <rguenther@suse.de>
6235 PR tree-optimization/81571
6236 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6239 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6241 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6242 earlier and only if MASK_FPU is set. Adjust formatting.
6244 2017-07-27 Martin Liska <mliska@suse.cz>
6246 * opt-functions.awk: Add validation of value of Init.
6247 * optc-gen.awk: Pass new argument.
6249 2017-07-27 Martin Liska <mliska@suse.cz>
6251 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6252 Fix wrong condition.
6254 2017-07-27 Martin Liska <mliska@suse.cz>
6256 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6257 BBs and edges seen by autoFDO.
6259 2017-07-27 Richard Biener <rguenther@suse.de>
6261 PR tree-optimization/81502
6262 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6263 with incompatible but same sized type.
6264 (execute_update_addresses_taken): Likewise.
6266 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6268 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6269 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6271 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6274 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6275 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6276 Change s_operand to memory_operand.
6278 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6280 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6281 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6282 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6283 Emit instructions rather than returning an expression. Handle TFmode
6284 and KFmode by casting to TImode.
6285 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6286 (rs6000_emit_le_vsx_store): Likewise.
6287 * config/rs6000/vsx.md (VSX_TI): New iterator.
6288 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6289 (*vsx_le_undo_permute_<mode>): Likewise.
6290 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6291 emit the split sequence.
6292 (*vsx_le_perm_store_<mode>): Likewise.
6294 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6296 PR tree-optimization/81555
6297 PR tree-optimization/81556
6298 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6299 if true, force CHANGED for the recursive invocation.
6300 (reassociate_bb): Remember original length of ops array, pass
6301 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6303 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6304 attributes for noipa attribute. For naked attribute use
6305 lookup_attribute first before lookup_attribute_spec.
6306 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6308 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6309 for functions with noipa attribute.
6310 (cgraph_externally_visible_p): Return true for functions with noipa
6312 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6313 for functions with noipa attribute.
6314 * doc/extend.texi: Document noipa function attribute.
6315 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6316 also for functions with noipa attribute.
6317 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6319 2017-07-26 Andrew Pinski <apinski@cavium.com>
6321 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6322 vec_unalign_load_cost and vec_unalign_store_cost.
6324 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6326 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6327 -mvsx-small-integer option.
6328 (ISA_3_0_MASKS_IEEE): Likewise.
6329 (OTHER_VSX_VECTOR_MASKS): Likewise.
6330 (POWERPC_MASKS): Likewise.
6331 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6332 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6333 code, only testing for DImode being allowed in non-VSX floating
6335 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6336 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6338 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6339 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6340 TARGET_P8_VECTOR test.
6341 (rs6000_secondary_reload_simple_move): Likewise.
6342 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6343 since TARGET_P9_VECTOR was already tested.
6344 (rs6000_opt_masks): Remove -mvsx-small-integer.
6345 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6346 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6348 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6349 test for TARGET_VEXTRACTUB was used, and that uses
6351 (p9 extract splitter): Likewise.
6352 (vsx_extract_<mode>_di_p9): Likewise.
6353 (vsx_extract_<mode>_store_p9): Likewise.
6354 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6355 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6356 the elimination of TARGET_VSX_SMALL_INTEGER.
6357 (vsx_extract_<mode>_p8): Likewise.
6358 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6359 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6360 (vsx_set_<mode>_p9): Likewise.
6361 (vsx_set_v4sf_p9): Likewise.
6362 (vsx_set_v4sf_p9_zero): Likewise.
6363 (vsx_insert_extract_v4sf_p9): Likewise.
6364 (vsx_insert_extract_v4sf_p9_2): Likewise.
6365 * config/rs6000/rs6000.md (sign extend splitter): Change
6366 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6367 (floatsi<mode>2_lfiwax_mem): Likewise.
6368 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6369 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6370 since a test for TARGET_P9_VECTOR was used.
6371 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6372 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6373 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6374 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6375 TARGET_P8_VECTOR test.
6376 (fix_trunc<mode>si2_stfiwx): Likewise.
6377 (fix_trunc<mode>si2_internal): Likewise.
6378 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6379 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6381 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6382 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6383 TARGET_P8_VECTOR test.
6384 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6385 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6386 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6388 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6389 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6390 since a test for TARGET_P9_VECTOR was used.
6391 (splitter for loading small constants): Likewise.
6393 2017-07-26 Andrew Pinski <apinski@cavium.com>
6395 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6398 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6401 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6402 (fp_valid_at): Likewise.
6404 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6406 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6407 (qdf24xx_addrcost_table): Likewise.
6408 (cortexa57_tunings): Update to use generic_branch_cost.
6409 (cortexa72_tunings): Likewise.
6410 (cortexa73_tunings): Likewise.
6411 (qdf24xx_tunings): Likewise.
6413 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6415 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6416 (thunderx2t99_branch_cost): Likewise.
6417 (cortexa35_tunings): Update to use generic_branch_cost.
6418 (cortexa53_tunings): Likewise.
6419 (cortexa57_tunings): Likewise.
6420 (cortexa72_tunings): Likewise.
6421 (cortexa73_tunings): Likewise.
6422 (thunderx2t99_tunings): Likewise.
6424 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6426 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6427 (sparc_option_override): Honour MASK_FSMULD.
6428 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6429 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6430 * config/sparc/sparc.opt (mfsmuld): New option.
6431 * doc/invoke.texi (mfsmuld): Document option.
6433 2017-07-26 Marek Polacek <polacek@redhat.com>
6436 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6438 2017-07-26 Richard Biener <rguenther@suse.de>
6440 * gimple-match-head.c (do_valueize): Return OP if valueize
6442 (get_def): New helper to get at the def stmt of a SSA name
6444 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6445 do_valueize to get at the def stmt.
6446 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6448 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6451 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6453 2017-07-26 Martin Liska <mliska@suse.cz>
6456 * function.c (expand_function_start): Make expansion of
6457 nonlocal_goto_save_area after parm_birth_insn.
6459 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6461 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6462 from all CPU target flags enable members.
6464 2017-07-26 Richard Biener <rguenther@suse.de>
6466 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6467 (decision_tree::gen): Make 'type' const.
6468 (write_predicate): Likewise.
6470 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6472 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6473 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6474 (rs6000_option_override_internal): Likewise.
6475 (rs6000_expand_vector_set): Likewise.
6476 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6477 (TARGET_UPPER_REGS_SF): Likewise.
6478 (TARGET_UPPER_REGS_DI): Likewise.
6479 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6480 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6481 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6482 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6483 (Splitters for DI constants in Altivec registers): Likewise.
6484 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6485 (vsx_set_v4sf_p9): Likewise.
6486 (vsx_set_v4sf_p9_zero): Likewise.
6487 (vsx_insert_extract_v4sf_p9): Likewise.
6488 (vsx_insert_extract_v4sf_p9_2): Likewise.
6490 2017-07-25 Carl Love <cel@us.ibm.com>
6492 * doc/extend.texi: Update the built-in documentation file for the
6493 existing built-in functions
6494 vector signed char vec_cnttz (vector signed char);
6495 vector unsigned char vec_cnttz (vector unsigned char);
6496 vector signed short vec_cnttz (vector signed short);
6497 vector unsigned short vec_cnttz (vector unsigned short);
6498 vector signed int vec_cnttz (vector signed int);
6499 vector unsigned int vec_cnttz (vector unsigned int);
6500 vector signed long long vec_cnttz (vector signed long long);
6501 vector unsigned long long vec_cnttz (vector unsigned long long);
6503 2017-07-25 Andrew Pinski <apinski@cavium.com>
6505 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6506 accesses where the use is for the first operand of a BIT_INSERT.
6508 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6511 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6512 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6514 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6516 * config/i386/gstabs.h: Delete.
6517 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6519 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6521 * config/i386/i386.c (ix86_decompose_address): Do not check for
6522 register RTX when looking at index_reg or base_reg.
6523 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6525 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6527 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6528 to update EH info here.
6530 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6532 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6534 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6536 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6538 2017-07-25 Torsten Duwe <duwe@suse.de>
6540 * common.opt: Introduce -fpatchable-function-entry
6541 command line option, and its variables function_entry_patch_area_size
6542 and function_entry_patch_area_start.
6543 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6544 including a two-value parser.
6545 * target.def (print_patchable_function_entry): New target hook.
6546 * targhooks.h (default_print_patchable_function_entry): New function.
6547 * targhooks.c (default_print_patchable_function_entry): Likewise.
6548 * toplev.c (process_options): Switch off IPA-RA if
6549 patchable function entries are being generated.
6550 * varasm.c (assemble_start_function): Look at the
6551 patchable-function-entry command line switch and current
6552 function attributes and maybe generate NOP instructions by
6553 calling the print_patchable_function_entry hook.
6554 * doc/extend.texi: Document patchable_function_entry attribute.
6555 * doc/invoke.texi: Document -fpatchable_function_entry
6556 command line option.
6557 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6559 * doc/tm.texi: Re-generate.
6561 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6564 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6565 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6567 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6569 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6571 2017-07-25 Richard Biener <rguenther@suse.de>
6573 PR tree-optimization/81455
6574 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6575 not walk in cycles when looking for guards.
6577 2017-07-25 Richard Biener <rguenther@suse.de>
6579 PR tree-optimization/81529
6580 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6581 when optimizing backedge uses.
6583 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6585 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6587 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6588 to dl_section_ref. On AIX, append an expression to subtract
6589 the size of the section length to dl_section_ref.
6591 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6593 * configure.ac: If any of the config.* scripts fail, exit 1.
6594 * configure: Regenerate.
6596 2017-07-25 Richard Biener <rguenther@suse.de>
6599 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6600 of immediate uses, be more verbose on errors.
6602 2017-07-25 Richard Biener <rguenther@suse.de>
6604 PR tree-optimization/81510
6605 * tree-vect-loop.c (vect_is_simple_reduction): When the
6606 reduction stmt is not inside the loop bail out.
6608 2017-07-25 Richard Biener <rguenther@suse.de>
6610 PR tree-optimization/81303
6611 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6612 profitability check against LOOP_VINFO_NITERSM1.
6614 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6616 * domwalk.c (cmp_bb_postorder): Simplify.
6617 (sort_bbs_postorder): New function. Use it...
6618 (dom_walker::walk): ...here to optimize common cases.
6620 2017-07-25 Martin Liska <mliska@suse.cz>
6623 * ipa-visibility.c (function_and_variable_visibility): Make the
6624 redirection just on target that supports aliasing.
6625 Fix GNU coding style.
6627 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6630 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6632 * config/arm/rtems.h: Likewise.
6633 * config/bfin/rtems.h: Likewise.
6634 * config/i386/rtemself.h: Likewise.
6635 * config/lm32/rtems.h: Likewise.
6636 * config/m32c/rtems.h: Likewise.
6637 * config/m68k/rtemself.h: Likewise.
6638 * config/microblaze/rtems.h: Likewise.
6639 * config/mips/rtems.h: Likewise.
6640 * config/moxie/rtems.h: Likewise.
6641 * config/nios2/rtems.h: Likewise.
6642 * config/powerpcspe/rtems.h: Likewise.
6643 * config/rs6000/rtems.h: Likewise.
6644 * config/rtems.h: Likewise.
6645 * config/sh/rtems.h: Likewise.
6646 * config/sh/rtemself.h: Likewise.
6647 * config/sparc/rtemself.h: Likewise.
6649 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6652 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6653 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6654 * tree-ssa-structalias.c (alias_get_name): Same.
6656 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6659 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6660 instructions if no du chain is found.
6662 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6664 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6666 2017-07-25 Richard Biener <rguenther@suse.de>
6669 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6672 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6674 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6676 (ISA_2_7_MASKS_SERVER): Likewise.
6677 (ISA_3_0_MASKS_IEEE): Likewise.
6678 (OTHER_P8_VECTOR_MASKS): Likewise.
6679 (OTHER_VSX_VECTOR_MASKS): Likewise.
6680 (POWERPC_MASKS): Likewise.
6681 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6682 duplicate list of options.
6683 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6684 explicit -mupper-regs options.
6685 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6686 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6687 alias for -mupper-regs-df.
6688 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6689 (rs6000_init_hard_regno_mode_ok): Likewise.
6690 (rs6000_option_override_internal): Likewise.
6691 (rs6000_opt_masks): Likewise.
6692 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6693 options in terms of whether -mvsx or -mpower8-vector was used.
6694 (TARGET_UPPER_REGS_DI): Likewise.
6695 (TARGET_UPPER_REGS_SF): Likewise.
6696 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6697 -mupper-regs-* options.
6699 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6701 * passes.c (emergency_dump_function): Print some empty lines and a
6702 header before the RTL dump.
6704 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6706 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6708 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6711 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6712 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6714 2017-07-24 Carl Love <cel@us.ibm.com>
6716 * config/rs6000/rs6000-c.c: Add support for built-in functions
6717 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6718 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6719 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6720 vec_extract_fp_from_shortl): Add defines for the two builtins.
6721 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6722 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6724 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6725 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6726 * doc/extend.texi: Update the built-in documentation file for the
6727 new built-in function.
6729 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6732 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6734 * doc/generic.texi: Likewise.
6735 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6736 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6738 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6740 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6741 (aarch64_mls_elt_merge<mode>): Likewise.
6743 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6745 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6746 having __cxa_atexit.
6748 2017-07-23 Michael Collison <michael.collison@arm.com>
6750 * config/arm/arm.c (arm_option_override): Deprecate
6751 use of -mstructure-size-boundary.
6752 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6753 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6755 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6758 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6759 Reduce cost estimate for direct moves.
6761 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6764 * config/i386/i386.c (ix86_option_override_internal): Disable
6765 BMI, BMI2 and TBM instructions for -m16.
6767 2017-07-21 Carl Love <cel@us.ibm.com>
6769 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6770 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6771 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6772 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6773 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6774 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6775 VMULOSW): New enum "unspec" values.
6776 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6777 altivec_vmulosw): New patterns.
6778 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6779 VMULOSW): Add definitions.
6781 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6783 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6784 (qdf24xx): Likewise.
6785 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6786 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6787 (AARCH64_FL_V8_1): Renumber.
6788 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6789 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6790 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6791 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6792 rdma to feature modifiers list.
6794 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6797 * ipa-visibility (function_and_variable_visibility): Convert
6798 recursive PLT call to direct call if appropriate.
6800 2017-07-21 Andrew Pinski <apinski@cavium.com>
6802 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6803 operand 1 to see if the types precision matches.
6804 * fold-const.c (operand_equal_p): Likewise.
6806 2017-07-21 Richard Biener <rguenther@suse.de>
6808 PR tree-optimization/81303
6809 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6810 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6811 (vect_peeling_hash_get_lowest_cost): Adjust.
6812 (vect_enhance_data_refs_alignment): Likewise. Use
6813 vect_get_peeling_costs_all_drs to compute the penalty for no
6814 peeling to match up costs.
6816 2017-07-21 Richard Biener <rguenther@suse.de>
6818 PR tree-optimization/81500
6819 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6820 we didn't identify a reduction path.
6822 2017-07-21 Tom de Vries <tom@codesourcery.com>
6823 Cesar Philippidis <cesar@codesourcery.com>
6825 PR gcov-profile/81442
6826 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6829 2017-07-21 Tom de Vries <tom@codesourcery.com>
6832 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6834 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6835 nvptx_override_options_after_change.
6837 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6839 * dwarf2out.c (output_file_names): Avoid double testing for
6842 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6844 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6846 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6848 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6850 (try_crossjump_to_edge): Do not punt on partitioned functions.
6852 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6854 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6855 Put all BBs reachable only via paths crossing cold region to cold
6857 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6859 2016-07-21 Richard Biener <rguenther@suse.de>
6861 PR tree-optimization/81303
6862 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6863 into account prologue and epilogue iterations when raising
6864 min_profitable_iters to sth at least covering one vector iteration.
6866 2017-07-21 Tamar Christina <tamar.christina@arm.com>
6868 * config/arm/arm.c (arm_test_cpu_arch_dat):
6871 2017-07-20 Nathan Sidwell <nathan@acm.org>
6873 Remove TYPE_METHODS.
6874 * tree.h (TYPE_METHODS): Delete.
6875 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
6876 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
6877 (dbxout_type_methods): Scan TYPE_FIELDS.
6878 (dbxout_type): Don't check TYPE_METHODS here.
6879 * function.c (use_register_for_decl): Always ignore register for
6880 class types when not optimizing.
6881 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
6882 * tree.c (free_lang_data_in_type): Stitch out member functions and
6883 templates from TYPE_FIELDS.
6884 (build_distinct_type_copy, verify_type_variant,
6885 verify_type): Member fns are on TYPE_FIELDS.
6886 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
6887 * tree-pretty-print.c (dump_generic_node): Likewise.
6889 2017-07-20 Jakub Jelinek <jakub@redhat.com>
6892 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
6893 V2TImode and V4TImode.
6894 (ix86_expand_vector_extract): Likewise.
6895 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
6896 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
6897 (ssescalarmode): Handle V4TImode and V2TImode.
6898 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6899 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6900 (VEXTRACTI128_MODE): New mode iterator.
6901 (splitter for *vec_extractv?ti first element): New.
6902 (VEC_INIT_MODE): New mode iterator.
6903 (vec_init<mode>): Consolidate 3 expanders into one using
6904 VEC_INIT_MODE mode iterator.
6906 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6908 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6909 non_spilled_static_chain_regno_p.
6911 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6913 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6915 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6917 * bb-reorder.c (connect_traces): Allow copying of blocks within
6920 2017-07-20 Richard Biener <rguenther@suse.de>
6922 * gimple.h (gimple_phi_result): Add gphi * overload.
6923 (gimple_phi_result_ptr): Likewise.
6924 (gimple_phi_arg): Likewise. Adjust index assert to only
6925 allow actual argument accesses rather than all slots available
6927 (gimple_phi_arg_def): Add gphi * overload.
6928 * tree-phinodes.c (make_phi_node): Initialize only actual
6930 (resize_phi_node): Clear memory not covered by old node,
6931 do not initialize excess argument slots.
6932 (reserve_phi_args_for_new_edge): Initialize new argument slot
6935 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6937 PR tree-optimization/81388
6939 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6941 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6942 by removing computation of may_be_zero.
6944 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6945 Tom de Vries <tom@codesourcery.com>
6948 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6949 when gimple level profile disagrees with what RTL expander did.
6951 2017-07-20 Richard Biener <rguenther@suse.de>
6953 PR tree-optimization/61171
6954 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6955 (vect_analyze_stmt): Add slp instance parameter.
6956 (vectorizable_reduction): Likewise.
6957 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6958 (vect_is_simple_reduction): Deal with chains not detected
6959 as SLP reduction chain, specifically not properly associated
6960 chains containing a mix of plus/minus.
6961 (get_reduction_op): Remove.
6962 (get_initial_defs_for_reduction): Simplify, pass in whether
6963 this is a reduction chain, pass in the SLP node for the PHIs.
6964 (vect_create_epilog_for_reduction): Get the SLP instance as
6966 (vectorizable_reduction): Get the SLP instance as arg.
6967 During analysis remember the SLP node with the PHIs in the
6968 instance. Simplify getting at the vectorized reduction PHIs.
6969 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6970 through SLP instance.
6971 (vect_slp_analyze_operations): Likewise.
6972 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6973 (vect_transform_stmt): Likewise.
6975 2017-07-20 Tom de Vries <tom@codesourcery.com>
6977 PR tree-optimization/81489
6978 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6979 read of phi arg location to before loop that modifies phi.
6981 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6983 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6986 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6989 * except.c (execute): Fix ordering issue.
6991 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6993 PR rtl-optimization/81423
6994 * combine.c (make_compound_operation_int): Don't try to optimize
6995 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6997 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6999 PR rtl-optimization/81423
7000 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7001 with a constant that is -1 in the truncated to mode.
7003 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7005 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7006 (determine_unlikely_bbs): ... here.
7007 * predict.h (propagate_unlikely_bbs_forward): Declare.
7008 * cfgexpand.c (pass_expand::execute): Use it.
7009 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7011 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7012 propagate_unlikely_bbs_forward.
7014 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7017 * except.c (maybe_add_nop_after_section_switch): New function.
7020 2017-07-19 Tom de Vries <tom@codesourcery.com>
7022 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7024 2017-07-19 Tom de Vries <tom@codesourcery.com>
7026 * gimple.h (gimple_phi_arg): Make assert more strict.
7028 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7030 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7031 * config/rs6000/mmintrin.h: New file.
7032 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7034 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7036 PR tree-optimization/81346
7037 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7039 2017-07-19 Tom de Vries <tom@codesourcery.com>
7041 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7043 2017-07-19 Tom de Vries <tom@codesourcery.com>
7045 * config/nvptx/nvptx-modes.def: Add V2DImode.
7046 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7047 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7048 (nvptx_output_mov_insn): Handle lack of mov.b128.
7049 (nvptx_print_operand): Handle 'H' and 'L' codes.
7050 (nvptx_vector_mode_supported): Allow V2DImode.
7051 (nvptx_preferred_simd_mode): New function.
7052 (nvptx_data_alignment): New function.
7053 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7054 nvptx_preferred_simd_mode.
7055 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7057 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7059 2017-07-19 Tom de Vries <tom@codesourcery.com>
7061 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7062 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7063 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7064 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7065 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7066 (mov<VECIM>_insn): New define_insn.
7067 (define_expand "mov<VECIM>): New define_expand.
7069 2017-07-19 Tom de Vries <tom@codesourcery.com>
7071 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7073 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7075 PR tree-optimization/81346
7076 * fold-const.h (fold_div_compare, range_check_type): Declare.
7077 * fold-const.c (range_check_type): New function.
7078 (build_range_check): Use range_check_type.
7079 (fold_div_compare): No longer static, rewritten into
7080 a match.pd helper function.
7081 (fold_comparison): Don't call fold_div_compare here.
7082 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7085 2017-07-19 Nathan Sidwell <nathan@acm.org>
7087 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7088 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7089 * tree.c (find_decls_types_r, verify_type): Use
7090 TYPE_{MIN,MAX}_VALUE_RAW.
7091 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7092 (hash_tree): Likewise.
7093 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7095 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7098 2017-07-18 Tom de Vries <tom@codesourcery.com>
7101 * omp-expand.c (expand_omp_for_static_chunk): Handle
7102 equal-argument loop exit phi.
7104 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7107 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7108 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7109 operand 2 predicate.
7110 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7111 operand 2 predicate.
7112 (ror,rol -> rorx splitters): Use const_int_operand as
7113 operand 2 predicate.
7115 2017-06-18 Richard Biener <rguenther@suse.de>
7117 PR tree-optimization/81410
7118 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7119 the gap in the ! slp_perm SLP case after each group.
7121 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7124 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7127 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7130 * predict.c (set_even_probabilities): Cleanup; do not affect
7131 probabilities that are already known.
7132 (combine_predictions_for_bb): Call even when count is set.
7134 2017-07-18 Nathan Sidwell <nathan@acm.org>
7136 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7139 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7142 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7143 optimization for loop niter analysis.
7145 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7148 * config/avr/avr.c (avr_optimize_casesi): Don't use
7149 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7151 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7153 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7154 body_cost_vec from _vect_peel_extended_info.
7155 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7156 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7159 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7161 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7163 2017-07-18 Richard Biener <rguenther@suse.de>
7165 PR tree-optimization/80620
7166 PR tree-optimization/81403
7167 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7168 info when re-using a VN table entry.
7170 2017-07-18 Richard Biener <rguenther@suse.de>
7172 PR tree-optimization/81418
7173 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7174 vectype_in. Verify that with lane-reducing reduction operations
7175 we have a single def-use cycle.
7177 2017-07-17 Carl Love <cel@us.ibm.com>
7179 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7181 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7182 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7183 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7184 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7185 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7186 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7187 VMULOSW): New enum "unspec" values.
7188 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7189 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7190 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7191 altivec_vmulosw): New patterns.
7192 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7193 VMULOSW): Add definitions.
7194 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7196 * config/alpha/alpha.c: Include predict.h.
7198 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7200 * tree-vrp.c (compare_assert_loc): Fix comparison function
7201 to return predictable results.
7203 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7205 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7208 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7209 * doc/invoke.texi (mexpand-adddi): Update text.
7211 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7213 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7214 that also clobbers the CC register. The old expand code is moved
7216 (*arc_clzsi2): ... here.
7217 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7218 the CC register. The old expand code is moved to ...
7219 (arc_ctzsi2): ... here.
7221 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7223 * config/arc/arc.opt (mindexed-loads): Use initial value
7224 TARGET_INDEXED_LOADS_DEFAULT.
7225 (mauto-modify-reg): Use initial value
7226 TARGET_AUTO_MODIFY_REG_DEFAULT.
7227 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7228 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7229 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7230 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7232 2017-07-17 Martin Liska <mliska@suse.cz>
7235 * opts.c (finish_options): Do not allow -fgnu-tm
7236 w/ -fsanitize={kernel-,}address. Say sorry.
7238 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7241 * tree-loop-distribution.c (classify_partition): Only assert on
7242 numer of iterations.
7243 (merge_dep_scc_partitions): Delete prameter. Update function call.
7244 (distribute_loop): Remove code handling loop with unknown niters.
7245 (pass_loop_distribution::execute): Skip loop with unknown niters.
7247 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7250 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7251 function sort_partitions_by_post_order.
7253 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7255 PR tree-optimization/81374
7256 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7257 the max index of basic blocks, rather than number of basic blocks.
7259 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7261 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7263 (arc_legitimate_pic_operand_p): Likewise.
7264 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7266 (arc_needs_pcl_p): Likewise.
7267 (arc_legitimate_pc_offset_p): Likewise.
7268 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7269 function is also used in constrains.md.
7270 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7271 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7272 PLUS. Only return true/false in known cases, otherwise assert.
7273 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7274 is already called in arc_legitimate_constant_p.
7275 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7277 (LEGITIMATE_PIC_OPERAND_P): Use
7278 arc_raw_symbolic_reference_mentioned_p function.
7279 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7284 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7285 Andrew Burgess <andrew.burgess@embecosm.com>
7287 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7288 (arc_return_address_register): New function.
7289 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7290 (arc_handle_fndecl_attribute): Add naked attribute.
7291 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7292 (TARGET_WARN_FUNC_RETURN): Likewise.
7293 (arc_allocate_stack_slots_for_args): New function.
7294 (arc_warn_func_return): Likewise.
7295 (machine_function): Change type fn_type.
7296 (arc_compute_function_type): Consider new naked function type,
7297 change function return type.
7298 (arc_must_save_register): Adapt to handle new
7299 arc_compute_function_type's return type.
7300 (arc_expand_prologue): Likewise.
7301 (arc_expand_epilogue): Likewise.
7302 (arc_return_address_regs): Delete.
7303 (arc_return_address_register): New function.
7304 (arc_epilogue_uses): Use above function.
7305 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7306 (arc_function_type): Change encoding, add naked type.
7307 (ARC_INTERRUPT_P): Change to handle the new encoding.
7308 (ARC_FAST_INTERRUPT_P): Likewise.
7309 (ARC_NORMAL_P): Define.
7310 (ARC_NAKED_P): Likewise.
7311 (arc_compute_function_type): Delete prototype.
7312 * config/arc/arc.md (in_ret_delay_slot): Use
7313 arc_return_address_register function.
7314 (simple_return): Likewise.
7315 (p_return_i): Likewise.
7317 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7319 PR tree-optimization/81428
7320 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7321 can't be built for those types.
7323 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7325 Remove stuff dead since r239246.
7327 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7328 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7329 (avr_inform_devices): Remove dead stuff.
7331 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7333 * config/arm/arm_neon.h: Fix softp typo.
7335 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7337 PR tree-optimization/81365
7338 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7339 aggregate moves onto bb predecessor edges, make sure there are no
7340 loads that could alias the lhs in between the start of bb and the
7343 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7346 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7347 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7348 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7350 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7352 PR tree-optimization/81396
7353 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7354 (init_symbolic_number): Initialize it to 1.
7355 (perform_symbolic_merge): Add n_ops from both operands into the new
7357 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7358 without base_addr as useless if they need more than one operation.
7359 (bswap_replace): Handle !bswap case for NULL base_addr.
7361 2017-07-17 Tom de Vries <tom@codesourcery.com>
7364 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7367 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7369 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7370 conditional builtin define __FIX_LEON3FT_B2BST.
7372 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7374 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7375 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7378 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7380 PR rtl-optimization/81424
7381 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7382 to remove potential trapping from operands if -fnon-call-exceptions.
7384 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7386 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7387 profile_proability for scalling.
7388 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7390 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7392 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7394 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7396 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7397 fixpoint arithmetics.
7399 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7401 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7402 fixpoint arithmetics.
7404 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7406 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7407 fixpoint arithmetics.
7409 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7411 * profile-count.h (profile_probability::from_reg_br_prob_note,
7412 profile_probability::to_reg_br_prob_note): New functions.
7413 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7414 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7415 * predict.c (probability_reliable_p): Update.
7416 (edge_probability_reliable_p): Update.
7417 (br_prob_note_reliable_p): Update.
7418 (invert_br_probabilities): Update.
7419 (add_reg_br_prob_note): New function.
7420 (combine_predictions_for_insn): Update.
7421 * asan.c (asan_clear_shadow): Update.
7422 * cfgbuild.c (compute_outgoing_frequencies): Update.
7423 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7424 (update_br_prob_note): Update.
7425 (rtl_verify_edges): Update.
7426 (purge_dead_edges): Update.
7427 (fixup_reorder_chain): Update.
7428 * emit-rtl.c (try_split): Update.
7429 * ifcvt.c (cond_exec_process_insns): Update.
7430 (cond_exec_process_if_block): Update.
7431 (dead_or_predicable): Update.
7432 * internal-fn.c (expand_addsub_overflow): Update.
7433 (expand_neg_overflow): Update.
7434 (expand_mul_overflow): Update.
7435 * loop-doloop.c (doloop_modify): Update.
7436 * loop-unroll.c (compare_and_jump_seq): Update.
7437 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7438 * predict.h: Update.
7439 * reorg.c (mostly_true_jump): Update.
7441 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7442 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7443 * config/arc/arc.c: (emit_unlikely_jump): Update.
7444 * config/arm/arm.c: (emit_unlikely_jump): Update.
7445 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7446 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7447 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7448 (ix86_print_operand): Update.
7449 (ix86_split_fp_branch): Update.
7450 (predict_jump): Update.
7451 * config/ia64/ia64.c (ia64_print_operand): Update.
7452 * config/mmix/mmix.c (mmix_print_operand): Update.
7453 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7454 (rs6000_expand_split_stack_prologue): Update.
7455 * config/rs6000/rs6000.c: Update.
7456 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7457 (s390_expand_vec_movstr): Update.
7458 (s390_expand_cs_tdsi): Update.
7459 (s390_expand_split_stack_prologue): Update.
7460 * config/sh/sh.c (sh_print_operand): Update.
7461 (expand_cbranchsi4): Update.
7462 (expand_cbranchdi4): Update.
7463 * config/sparc/sparc.c (output_v9branch): Update.
7464 * config/spu/spu.c (get_branch_target): Update.
7465 (ea_load_store_inline): Update.
7466 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7467 * config/tilepro/tilepro.c: Update.
7469 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7471 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7472 (gimplify_save_expr): Likewise.
7474 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7476 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7478 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7480 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7482 * timevar.def (TV_IPA_FNSUMMARY): Define.
7484 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7486 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7487 to back store errata sensitive sequence from being generated.
7488 (sqrtdf2_fix): Likewise.
7490 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7492 * tree-ssa-threadupdate.c (compute_path_counts,
7493 update_joiner_offpath_counts): Use profile_probability.
7495 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7498 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7500 * config/arm/arm-c.c (arm_cpu_builtins): Define
7501 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7503 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7505 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7506 array entries to represent __ieee128 versions of the
7507 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7508 scalar_extract_sig, and scalar_insert_exp built-in functions.
7509 (altivec_resolve_overloaded_builtin): Add special case handling
7510 for the __builtin_scalar_insert_exp function, as represented by
7511 the P9V_BUILTIN_VEC_VSIEDP constant.
7512 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7513 exponent support for __ieee128 argument.
7514 (VSESQP): Add scalar extract signature support for __ieee128
7516 (VSTDCNQP): Add scalar test negative support for __ieee128
7518 (VSIEQP): Add scalar insert exponent support for __int128 argument
7519 with __ieee128 result.
7520 (VSIEQPF): Add scalar insert exponent support for __ieee128
7521 argument with __ieee128 result.
7522 (VSTDCQP): Add scalar test data class support for __ieee128
7524 (VSTDCNQP): Add overload support for scalar test negative with
7526 (VSTDCQP): Add overload support for scalar test data class
7528 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7530 (UNSPEC_VSX_SIEXPQP): New constant.
7531 (xsxexpqp): New insn for VSX scalar extract exponent quad
7533 (xsxsigqp): New insn for VSX scalar extract significand quad
7535 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7536 precision with floating point argument.
7537 (xststdcqp): New expand for VSX scalar test data class quad
7539 (xststdcnegqp): New expand for VSX scalar test negative quad
7541 (xststdcqp): New insn to match expansions for VSX scalar test data
7542 class quad precision and VSX scalar test negative quad precision.
7543 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7544 special case operand checking to enforce that second operand of
7545 VSX scalar test data class with quad precision argument is a 7-bit
7547 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7548 prototypes and descriptions of __ieee128 versions of
7549 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7550 scalar_test_data_class, and scalar_test_neg built-in functions.
7552 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7554 PR tree-optimization/81162
7555 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7556 replace a negate with an add.
7558 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7560 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7562 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7564 * config/arm/arm-c.c (arm_cpu_builtins): Define
7565 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7567 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7569 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7570 (armv8-r): Set ARM Cortex-R52 as default CPU.
7571 * config/arm/arm-tables.opt: Regenerate.
7572 * config/arm/arm-tune.md: Regenerate.
7573 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7575 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7576 extension for -mcpu=cortex-r52.
7578 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7580 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7581 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7582 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7583 (fp-armv8): Define it as FP_ARMv8 only.
7584 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7585 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7587 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7588 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7589 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7590 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7591 than TARGET_FPU_ARMV8.
7592 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7593 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7594 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7596 * config/arm/neon.md (neon_vrint): Likewise.
7597 (neon_vcvt): Likewise.
7598 (neon_<fmaxmin_op><mode>): Likewise.
7599 (<fmaxmin><mode>3): Likewise.
7600 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7601 * config/arm/predicates.md (arm_cond_move_operator): Check against
7602 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7604 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7606 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7609 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7611 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7612 loop in comment with memset.
7614 2017-07-14 Martin Liska <mliska@suse.cz>
7616 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7617 * dwarf2out.c (is_java): Remove the function.
7618 (output_pubname): Remove usage of the function.
7619 (lower_bound_default): Remove usage of DW_LANG_Java.
7620 (gen_compile_unit_die): Likewise.
7621 * gcc.c: Remove compiler defaults for .java and .zip files.
7622 * gimple-expr.c (remove_suffix): Change as there's no longer
7623 extension than 4-letter one.
7624 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7625 (gimplify_save_expr): Likewise.
7626 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7627 as it's possible even for other languages than Java.
7628 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7629 * lto-opts.c (lto_write_options): Remove reference to Java.
7630 * opts.c (strip_off_ending): Update file extension handling.
7631 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7632 * tree-eh.c (lower_resx): Likewise.
7633 * tree.c (free_lang_data_in_type): Remove dead code.
7634 (find_decls_types_r): Likewise.
7635 (build_common_builtin_nodes): Remove Java from a comment.
7636 (verify_type): Remove dead code.
7637 * varasm.c (assemble_external): Remove Java from a comment.
7639 2017-07-14 Martin Liska <mliska@suse.cz>
7641 * opts.c (finish_options): Add quotes.
7642 (common_handle_option): Likewise.
7644 2017-07-14 Martin Liska <mliska@suse.cz>
7646 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7647 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7649 * dwarf2out.c (lower_bound_default): Do not handle
7651 (gen_compile_unit_die): Likewise.
7652 * gcc.c: Remove default extension binding for GNU Pascal.
7653 * stmt.c: Remove Pascal language from a comment.
7654 * xcoffout.c: Likewise.
7656 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7659 * diagnostic-show-locus.c (fixit_cmp): New function.
7660 (layout::layout): Sort m_fixit_hints.
7661 (column_range::column_range): Assert that the values are valid.
7662 (struct char_span): New struct.
7663 (correction::overwrite): New method.
7664 (struct source_line): New struct.
7665 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7666 calls in terms of classes source_line and char_span, and
7667 correction::overwrite.
7668 (selftest::test_overlapped_fixit_printing_2): New function.
7669 (selftest::diagnostic_show_locus_c_tests): Call it.
7671 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7674 early if there is no lhs.
7676 2017-07-13 Martin Liska <mliska@suse.cz>
7678 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7679 (gen_reference_type_die): Likewise.
7680 * stor-layout.c: Remove Pascal-related comment.
7682 2017-07-13 Martin Liska <mliska@suse.cz>
7684 * opts.c (finish_options): Add quotes to error messages.
7685 (parse_sanitizer_options): Likewise.
7687 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7689 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7691 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7693 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7695 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7697 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7699 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7701 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7704 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7705 provides the hardware capability bits, define the macro
7706 __BUILTIN_CPU_SUPPORTS__.
7707 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7708 if GLIBC does not provide the hardware capability bits. Add a
7709 gcc_unreachable call if the built-in cpu function is neither
7710 __builtin_cpu_is nor __builtin_cpu_supports.
7711 (rs6000_get_function_versions_dispatcher): Change the warning
7712 that an old GLIBC is used which does not export the capability
7713 bits to be an error.
7714 * doc/extend.texi (target_clones attribute): Document the
7715 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7716 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7717 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7718 the macros defined by GCC if the newer GLIBC is available.
7720 2017-07-12 Jeff Law <law@redhat.com>
7722 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7723 remaining includes slightly.
7724 * config/riscv/riscv-builtins.c: Include profile-count.h.
7726 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7729 * config/avr/avr.c (avr_set_current_function): In diagnostic
7730 messages: Quote keywords and (parts of) identifiers.
7731 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7734 2017-07-12 Carl Love <cel@us.ibm.com>
7736 * config/rs6000/rs6000-c.c: Add support for built-in functions
7737 vector bool char vec_revb (vector bool char);
7738 vector bool short vec_revb (vector short char);
7739 vector bool int vec_revb (vector bool int);
7740 vector bool long long vec_revb (vector bool long long);
7741 * doc/extend.texi: Update the built-in documentation file for the
7742 new built-in functions.
7744 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7746 * config/s390/s390.md: Remove movcc splitter.
7748 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7750 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7751 load/store on condition.
7753 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7756 * config/avr/avr.c (avr_encode_section_info)
7757 [progmem && !TREE_READONLY]: Error if progmem object needs
7760 2017-07-11 Michael Collison <michael.collison@arm.com>
7762 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7765 2017-07-11 Carl Love <cel@us.ibm.com>
7767 * config/rs6000/rs6000-c.c: Add support for builtins
7768 vector unsigned int vec_parity_lsbb (vector signed int);
7769 vector unsigned int vec_parity_lsbb (vector unsigned int);
7770 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7771 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7772 vector unsigned long long vec_parity_lsbb (vector signed long long);
7773 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7774 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7775 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7776 * doc/extend.texi: Update the built-in documentation file for the
7777 new built-in functions.
7779 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7781 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7782 (layout::m_primary_loc): New field.
7783 (layout::layout): Initialize new field. Move location filtering
7784 logic from here to...
7785 (layout::maybe_add_location_range): ...this new method. Add
7786 support for filtering to just the lines already specified by other
7788 (layout::will_show_line_p): New method.
7789 (gcc_rich_location::add_location_if_nearby): New method.
7790 (selftest::test_add_location_if_nearby): New test function.
7791 (selftest::diagnostic_show_locus_c_tests): Call it.
7792 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7795 2017-07-11 Tom de Vries <tom@codesourcery.com>
7797 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7798 (bb_first_real_insn): New function.
7799 (nvptx_single): Add extra initialization of broadcasted condition
7802 2017-07-11 Nathan Sidwell <nathan@acm.org>
7804 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7806 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7808 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7809 Binutils doc as TEXI will mess them up.
7810 * doc/invoke.texi (AVR Options): Same here.
7812 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7814 * config/sparc/sparc.opt (mfix-ut700): New option.
7815 (mfix-gr712rc): Likewise.
7816 (sparc_fix_b2bst): New variable.
7817 * doc/invoke.texi (SPARC options): Document them.
7818 (ARM options): Fix warnings.
7819 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7820 instructions to prevent sequences that can trigger the store-store
7821 errata for certain LEON3FT processors.
7822 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7823 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7824 * config/sparc/sparc.md (fix_b2bst): New attribute.
7825 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7827 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7830 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7832 (*rsqrtsf2_sse): Ditto.
7834 (div<mode>3): Macroize insn from divdf3 and divsf3
7835 using MODEF mode iterator.
7837 2017-07-10 Martin Sebor <msebor@redhat.com>
7839 PR tree-optimization/80397
7840 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7841 instead of testing for equality to INTEGER_TYPE.
7843 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7845 * config.gcc: Remove uclibc from arc target spec.
7847 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7849 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7851 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7854 * lto-wrapper.c (remove_option): New function.
7855 (merge_and_complain): Merge PIC/PIE options more realistically.
7857 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7859 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7863 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7864 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7865 * config.in: Regenerate.
7866 * configure: Regenerate.
7867 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
7868 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
7869 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
7870 (TARGET_GASISR_PROLOGUES): ...target mask.
7871 * common/config/avr/avr-common.c
7872 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
7873 Set -mgas-isr-prologues.
7874 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
7875 INSERT_PASS_BEFORE for it.
7876 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
7877 * config/avr/avr.c (avr_option_override)
7878 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
7879 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
7880 (avr_attribute_table) <no_gccisr>: Add new function attribute.
7881 (avr_set_current_function) <is_no_gccisr>: Init machine field.
7882 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
7884 (make_avr_pass_pre_proep): New function.
7885 (emit_push_sfr) <treg>: Add argument to function and use it
7887 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
7888 and set machine->gasisr.yes.
7889 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
7890 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
7891 __gcc_isr.n_pushed to .L__stack_usage.
7892 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
7893 (avr_asm_final_postscan_insn): ...this new static function.
7894 * config/avr/avr.h (machine_function)
7895 <is_no_gccisr, use_L__stack_usage>: New fields.
7896 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
7897 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
7898 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7899 (gasisr, *gasisr): New expander and insn.
7900 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7901 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7902 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7904 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7906 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7909 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7911 Move jump-tables out of .text again.
7914 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7915 (ASM_OUTPUT_ADDR_VEC): New function.
7916 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7917 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7918 INSN_ADDRESSes as asm comment.
7919 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7920 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7921 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7922 * config/avr/avr.md (*tablejump): Adjust comment.
7923 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7924 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7926 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7927 (avr_output_addr_vec): New proto.
7928 (avr_log_t) <insn_addresses>: New field.
7930 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7933 * config/i386/i386.c (ix86_function_arg_advance): Set
7934 outgoing_args_on_stack to true if there are outgoing arguments
7936 (ix86_function_arg): Likewise.
7937 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7938 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7939 * config/i386/i386.h (machine_function): Add
7940 outgoing_args_on_stack.
7942 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7944 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7946 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7948 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7950 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7951 (REAL_H): Remove $(MACHMODE_H).
7952 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7954 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7955 $(MACHMODE_H) and double-int.h.
7956 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7958 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7961 2017-07-07 Andrew Pinski <apinski@cavium.com>
7963 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7964 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7966 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7968 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7969 Add warning if GCC was not configured to link against a GLIBC that
7970 exports the hardware capability bits.
7971 (make_resolver_func): Make resolver function private and not a
7972 COMDAT function. Create the name with clone_function_name instead
7973 of make_unique_name.
7976 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7977 correct operand in doing the split.
7979 2017-07-07 Carl Love <cel@us.ibm.com>
7981 * config/rs6000/rs6000-c: Add support for built-in function
7982 vector unsigned short vec_pack_to_short_fp32 (vector float,
7984 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7985 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7986 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7987 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7988 (convert_4f32_8i16): Add define_expand.
7989 * doc/extend.texi: Update the built-in documentation file for the
7990 new built-in function.
7992 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7994 * config/sparc/m8.md: New file.
7995 * config/sparc/sparc.md: Include m8.md.
7997 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7999 * config/sparc/sparc.opt: New option -mvis4b.
8000 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8001 (sparc_option_override): Handle VIS4B.
8002 (enum sparc_builtins): Define
8003 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8004 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8005 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8006 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8007 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8008 (check_constant_argument): New function.
8009 (sparc_vis_init_builtins): Define builtins
8010 __builtin_vis_dictunpack{8,16,32},
8011 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8012 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8013 __builtin_vis_fpcmpde{8,16,32}shl and
8014 __builtin_vis_fpcmpur{8,16,32}shl.
8015 (sparc_expand_builtin): Check that the constant operands to
8016 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8017 constant and in range.
8018 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8020 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8021 (SPARC_IMM5_P): Likewise.
8022 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8023 (enabled): Handle vis4b.
8024 (UNSPEC_DICTUNPACK): New unspec.
8025 (UNSPEC_FPCMPSHL): Likewise.
8026 (UNSPEC_FPUCMPSHL): Likewise.
8027 (UNSPEC_FPCMPDESHL): Likewise.
8028 (UNSPEC_FPCMPURSHL): Likewise.
8029 (cpu_feature): New CPU feature `vis4b'.
8030 (dictunpack{8,16,32}): New insns.
8031 (FPCSMODE): New mode iterator.
8032 (fpcscond): New code iterator.
8033 (fpcsucond): Likewise.
8034 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8035 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8036 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8037 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8038 * config/sparc/constraints.md: Define constraints `q' for unsigned
8039 2-bit integer constants and `t' for unsigned 5-bit integer
8041 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8043 (imm5_operand_dictunpack16): Likewise.
8044 (imm5_operand_dictunpack32): Likewise.
8045 (imm2_operand): Likewise.
8046 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8047 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8048 ditunpack* and fpcmp*shl builtins.
8050 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8052 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8053 * config.in: Add HAVE_AS_SPARC6 define.
8054 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8056 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8058 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8059 (CPP_CPU_SPEC): Handle m8.
8060 (ASM_CPU_SPEC): Likewise.
8061 * config/sparc/sparc-opts.h (enum processor_type): Add
8063 * config/sparc/sparc.c (m8_costs): New struct.
8064 (sparc_option_override): Handle TARGET_CPU_m8.
8065 (sparc32_initialize_trampoline): Likewise.
8066 (sparc64_initialize_trampoline): Likewise.
8067 (sparc_issue_rate): Likewise.
8068 (sparc_register_move_cost): Likewise.
8069 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8070 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8071 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8072 (CPP_CPU_SPEC): Handle M8.
8073 (ASM_CPU_SPEC): Likewise.
8074 (AS_M8_FLAG): Define.
8075 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8076 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8077 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8079 * configure: Regenerate.
8080 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8083 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8085 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8087 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8088 ("*movdi_insn_sp32"): Do not set v3pipe.
8089 ("*movsi_insn"): Likewise.
8090 ("*movdi_insn_sp64"): Likewise.
8091 ("*movsf_insn"): Likewise.
8092 ("*movdf_insn_sp32"): Likewise.
8093 ("*movdf_insn_sp64"): Likewise.
8094 ("*zero_extendsidi2_insn_sp64"): Likewise.
8095 ("*sign_extendsidi2_insn"): Likewise.
8096 ("*mov<VM32:mode>_insn"): Likewise.
8097 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8098 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8099 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8100 ("<vlop:code><VL:mode>3"): Likewise.
8101 ("*not_<vlop:code><VL:mode>3"): Likewise.
8102 ("*nand<VL:mode>_vis"): Likewise.
8103 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8104 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8105 ("one_cmpl<VL:mode>2"): Likewise.
8106 ("faligndata<VM64:mode>_vis"): Likewise.
8107 ("alignaddrsi_vis"): Likewise.
8108 ("alignaddrdi_vis"): Likweise.
8109 ("alignaddrlsi_vis"): Likewise.
8110 ("alignaddrldi_vis"): Likewise.
8111 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8112 ("bmaskdi_vis"): Likewise.
8113 ("bmasksi_vis"): Likewise.
8114 ("bshuffle<VM64:mode>_vis"): Likewise.
8115 ("cmask8<P:mode>_vis"): Likewise.
8116 ("cmask16<P:mode>_vis"): Likewise.
8117 ("cmask32<P:mode>_vis"): Likewise.
8118 ("pdistn<P:mode>_vis"): Likewise.
8119 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8121 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8123 * config/sparc/sparc.md ("subtype"): New insn attribute.
8124 ("*wrgsr_sp64"): Set insn subtype.
8125 ("*rdgsr_sp64"): Likewise.
8126 ("alignaddrsi_vis"): Likewise.
8127 ("alignaddrdi_vis"): Likewise.
8128 ("alignaddrlsi_vis"): Likewise.
8129 ("alignaddrldi_vis"): Likewise.
8130 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8131 ("fexpand_vis"): Likewise.
8132 ("fpmerge_vis"): Likewise.
8133 ("faligndata<VM64:mode>_vis"): Likewise.
8134 ("bshuffle<VM64:mode>_vis"): Likewise.
8135 ("cmask8<P:mode>_vis"): Likewise.
8136 ("cmask16<P:mode>_vis"): Likewise.
8137 ("cmask32<P:mode>_vis"): Likewise.
8138 ("fchksm16_vis"): Likewise.
8139 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8140 ("fmean16_vis"): Likewise.
8141 ("fp<plusminus_insn>64_vis"): Likewise.
8142 ("<plusminus_insn>v8qi3"): Likewise.
8143 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8144 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8145 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8146 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8147 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8148 ("*movqi_insn"): Likewise.
8149 ("*movhi_insn"): Likewise.
8150 ("*movsi_insn"): Likewise.
8151 ("movsi_pic_gotdata_op"): Likewise.
8152 ("*movdi_insn_sp32"): Likewise.
8153 ("*movdi_insn_sp64"): Likewise.
8154 ("movdi_pic_gotdata_op"): Likewise.
8155 ("*movsf_insn"): Likewise.
8156 ("*movdf_insn_sp32"): Likewise.
8157 ("*movdf_insn_sp64"): Likewise.
8158 ("*zero_extendhisi2_insn"): Likewise.
8159 ("*zero_extendqihi2_insn"): Likewise.
8160 ("*zero_extendqisi2_insn"): Likewise.
8161 ("*zero_extendqidi2_insn"): Likewise.
8162 ("*zero_extendhidi2_insn"): Likewise.
8163 ("*zero_extendsidi2_insn_sp64"): Likewise.
8164 ("ldfsr"): Likewise.
8165 ("prefetch_64"): Likewise.
8166 ("prefetch_32"): Likewise.
8167 ("tie_ld32"): Likewise.
8168 ("tie_ld64"): Likewise.
8169 ("*tldo_ldub_sp32"): Likewise.
8170 ("*tldo_ldub1_sp32"): Likewise.
8171 ("*tldo_ldub2_sp32"): Likewise.
8172 ("*tldo_ldub_sp64"): Likewise.
8173 ("*tldo_ldub1_sp64"): Likewise.
8174 ("*tldo_ldub2_sp64"): Likewise.
8175 ("*tldo_ldub3_sp64"): Likewise.
8176 ("*tldo_lduh_sp32"): Likewise.
8177 ("*tldo_lduh1_sp32"): Likewise.
8178 ("*tldo_lduh_sp64"): Likewise.
8179 ("*tldo_lduh1_sp64"): Likewise.
8180 ("*tldo_lduh2_sp64"): Likewise.
8181 ("*tldo_lduw_sp32"): Likewise.
8182 ("*tldo_lduw_sp64"): Likewise.
8183 ("*tldo_lduw1_sp64"): Likewise.
8184 ("*tldo_ldx_sp64"): Likewise.
8185 ("*mov<VM32:mode>_insn"): Likewise.
8186 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8187 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8189 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8191 * config/sparc/sparc.md ("type"): New insn type viscmp.
8192 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8194 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8195 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8196 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8197 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8199 ("n7_vis_logical_11cycle"): Likewise.
8200 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8201 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8202 * config/sparc/niagara.md ("niag_vis"): Likewise.
8203 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8204 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8206 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8208 * config/sparc/sparc.md: New instruction type `bmask'.
8209 (bmaskdi_vis): Use the `bmask' type.
8210 (bmasksi_vis): Likewise.
8211 * config/sparc/ultra3.md (us3_array): Likewise.
8212 * config/sparc/niagara7.md (n7_array): Likewise.
8213 * config/sparc/niagara4.md (n4_array): Likewise.
8214 * config/sparc/niagara2.md (niag2_vis): Likewise.
8215 (niag3_vis): Likewise.
8216 * config/sparc/niagara.md (niag_vis): Likewise.
8218 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8220 * ipa-comdats.c: Remove optimize check from gate.
8221 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8222 for functions not optimized.
8223 (ipa_fn_summary_read): Skip optimize check.
8224 (ipa_fn_summary_write): Likewise.
8225 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8227 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8229 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8230 (check_callers): Give up on caller which is not optimized.
8231 (inline_small_functions): Likewise.
8232 (ipa_inline): Do not give up when not optimizing.
8233 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8234 away unoptimizes cdtors.
8235 (whole_program_function_and_variable_visibility): Do
8236 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8237 * ipa.c (process_references): Do not check optimize.
8238 (symbol_table::remove_unreachable_nodes): Update optimize check.
8239 (set_writeonly_bit): Update optimize check.
8240 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8241 (pass_ipa_single_use::gate): Remove.
8243 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8245 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8246 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8247 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8248 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8249 permute_load, permute_store, adjust_extract, adjust_splat,
8250 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8251 replace_swap_with_copy, dump_swap_insn_table,
8252 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8253 recombine_lvx_pattern, recombine_stvx_pattern,
8254 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8255 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8256 to file rs6000-p8swap.c.
8257 * config/rs6000/rs6000-p8swap.c: New file.
8258 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8259 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8260 and rs6000*-*-* targets.
8262 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8264 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8266 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8268 * lto-wrapper.c (merge_and_complain): Do not merge
8269 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8270 fsigned_zeros, ftrapping_math, fwrapv.
8271 (append_compiler_options): Do not track these options.
8272 (append_linker_options): Likewie
8274 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8276 * cgraphunit.c (cgraph_node::finalize_function): When
8277 !flag_toplevel_reorde set no_reorder flag.
8278 (varpool_node::finalize_decl): Likewise.
8279 (symbol_table::compile): Drop no toplevel reorder path.
8281 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8283 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8284 edges; zero probability is not better than uninitialized.
8286 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8288 * asan.h (asan_sanitize_allocas_p): Declare.
8289 * asan.c (asan_sanitize_allocas_p): New function.
8290 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8291 (handle_builtin_alloca): Likewise.
8292 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8293 if !asan_sanitize_allocas_p.
8294 * params.def (asan-instrument-allocas): Add new option.
8295 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8296 * opts.c (common_handle_option): Disable allocas sanitization for
8299 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8301 * asan.c: Include gimple-fold.h.
8302 (get_last_alloca_addr): New function.
8303 (handle_builtin_stackrestore): Likewise.
8304 (handle_builtin_alloca): Likewise.
8305 (asan_emit_allocas_unpoison): Likewise.
8306 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8307 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8308 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8309 (instrument_builtin_call): Pass gimple iterator to
8310 get_mem_refs_of_builtin_call.
8311 (last_alloca_addr): New global.
8312 * asan.h (asan_emit_allocas_unpoison): Declare.
8313 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8314 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8315 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8316 if function calls alloca.
8317 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8318 * gimple-fold.h (replace_call_with_value): Declare.
8319 * internal-fn.c: Include asan.h.
8320 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8321 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8323 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8325 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8326 (C_SELFTEST_FLAGS): New.
8327 (CPP_SELFTEST_FLAGS): New.
8328 (SELFTEST_DEPS): New, from deps of s-selftest.
8329 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8330 (CPP_SELFTEST_DEPS): New.
8331 (selftest): Add dependency on s-selftest-c++.
8332 (s-selftest): Rename to...
8333 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8334 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8335 than SELFTEST_FLAGS.
8336 (selftest-gdb): Rename to...
8337 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8339 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8340 (selftest-valgrind): Rename to...
8341 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8343 (selftest-valgrind): Reintroduce as an alias for
8344 selftest-c-valgrind.
8345 (s-selftest-c++): New.
8346 (selftest-c++-gdb): New.
8347 (selftest-c++-valgrind): New.
8349 2017-07-06 Olivier Hainque <hainque@adacore.com>
8351 * gcc.c (process_command): When deciding if undefined variables
8352 should be ignored when processing specs, accept "gcc -v" as well.
8354 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8356 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8357 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8359 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8361 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8362 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8363 * config/arm/arm-tables.opt: Regenerate.
8364 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8366 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8368 2017-07-06 Carl Love <cel@us.ibm.com>
8370 * ChangeLog: Clean up from mid air collision
8372 2017-07-06 Carl Love <cel@us.ibm.com>
8374 * config/rs6000/rs6000-c.c: Add support for built-in functions
8375 vector signed int vec_subc (vector signed int, vector signed int);
8376 vector signed __int128 vec_subc (vector signed __int128,
8377 vector signed __int128);
8378 vector unsigned __int128 vec_subc (vector unsigned __int128,
8379 vector unsigned __int128);
8380 vector signed int vec_sube (vector signed int, vector signed int,
8382 vector unsigned int vec_sube (vector unsigned int,
8383 vector unsigned int,
8384 vector unsigned int);
8385 vector signed __int128 vec_sube (vector signed __int128,
8386 vector signed __int128,
8387 vector signed__int128);
8388 vector unsigned __int128 vec_sube (vector unsigned __int128,
8389 vector unsigned __int128,
8390 vector unsigned __int128);
8391 vector signed int vec_subec (vector signed int, vector signed int,
8393 vector unsigned int vec_subec (vector unsigned int,
8394 vector unsigned int,
8395 vector unsigned int);
8396 vector signed __int128 vec_subec (vector signed __int128,
8397 vector signed __int128,
8398 vector signed__int128);
8399 vector unsigned __int128 vec_subec (vector unsigned __int128,
8400 vector unsigned __int128,
8401 vector unsigned __int128);
8402 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8403 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8404 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8405 BU_ALTIVEC_OVERLOAD_X definitions.
8406 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8407 * doc/extend.texi: Update the built-in documentation file for the new
8410 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8413 * diagnostic-show-locus.c (layout::layout): Use start and finish
8414 spelling location for the start and finish of each range.
8415 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8416 Add unused aspect param.
8417 * input.c (expand_location_1): Add "aspect" param, and use it
8418 to access the correct part of the location.
8419 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8421 (expand_location_to_spelling_point): Likewise.
8422 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8423 param, and pass it to expand_location_1.
8425 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8427 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8428 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8429 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8430 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8431 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8432 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8433 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8434 _mm_maskz_getmant_ss): New intrinsics.
8435 (__builtin_ia32_getexpss128_mask): Changed to ...
8436 __builtin_ia32_getexpss128_round ... this.
8437 (__builtin_ia32_getexpsd128_mask): Changed to ...
8438 __builtin_ia32_getexpsd128_round ... this.
8439 * config/i386/i386-builtin-types.def
8440 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8441 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8442 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8443 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8444 __builtin_ia32_getmantss_mask_round): New builtins.
8445 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8446 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8447 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8448 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8449 * config/i386/sse.md
8450 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8451 avx512f_sgetexp<mode><mask_scalar_name>
8452 <round_saeonly_scalar_name> ... this.
8453 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8454 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8455 vgetexp<ssescalarmodesuffix>
8456 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8457 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8458 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8459 avx512f_vgetmant<mode><mask_scalar_name>
8460 <round_saeonly_scalar_name> ... this.
8461 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8462 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8463 vgetmant<ssescalarmodesuffix>
8464 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8465 %0<mask_scalar_operand4>, %1, %2
8466 <round_saeonly_scalar_mask_op4>, %3} ... this.
8467 * config/i386/subst.md (mask_scalar_operand4,
8468 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8469 round_saeonly_scalar_nimm_predicate): New subst attributes.
8471 2017-07-06 Julia Koval <julia.koval@intel.com>
8473 * config/i386/i386.c (ix86_erase_embedded_rounding):
8474 Remove code for old rounding pattern.
8476 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8478 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8480 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8482 * doc/sourcebuild.texi (Test Directives, Variants of
8483 dg-require-support): Add documentation for dg-require-stack-check.
8485 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8487 * config/i386/subst.md (mask_scalar, round_scalar,
8488 round_saeonly_scalar): New meta-templates.
8489 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8490 round_scalar_mask_operand3, round_scalar_mask_op3,
8491 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8492 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8493 round_saeonly_scalar_constraint,
8494 round_saeonly_scalar_prefix): New subst attribute.
8495 * config/i386/sse.md
8496 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8497 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8498 <round_scalar_name> ... this.
8499 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8500 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8501 <round_scalar_name> ... this.
8502 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8503 <sse>_vm<code><mode>3<mask_scalar_name>
8504 <round_saeonly_scalar_name> ... this.
8505 (v<plusminus_mnemonic><ssescalarmodesuffix>
8506 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8507 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8508 v<plusminus_mnemonic><ssescalarmodesuffix>
8509 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8510 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8511 (v<multdiv_mnemonic><ssescalarmodesuffix>
8512 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8513 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8514 v<multdiv_mnemonic><ssescalarmodesuffix>
8515 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8516 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8517 (v<maxmin_float><ssescalarmodesuffix>
8518 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8519 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8520 v<maxmin_float><ssescalarmodesuffix>
8521 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8522 %0<mask_scalar_operand3>, %1, %<iptr>2
8523 <round_saeonly_scalar_mask_op3>} ... this.
8525 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8527 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8528 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8530 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8531 Alan Hayward <alan.hayward@arm.com>
8532 David Sherwood <david.sherwood@arm.com>
8534 * combine.c (simplify_if_then_else): Remove "enum" before
8536 * compare-elim.c (can_eliminate_compare): Likewise.
8537 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8539 (aarch64_lookup_simd_builtin_type): Likewise.
8540 (aarch64_simd_builtin_type): Likewise.
8541 (aarch64_init_simd_builtin_types): Likewise.
8542 (aarch64_simd_expand_args): Likewise.
8543 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8545 (aarch64_reverse_mask): Likewise.
8546 (aarch64_simd_emit_reg_reg_move): Likewise.
8547 (aarch64_gen_adjusted_ldpstp): Likewise.
8548 (aarch64_ccmp_mode_to_code): Likewise.
8549 (aarch64_operands_ok_for_ldpstp): Likewise.
8550 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8551 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8553 (aarch64_min_divisions_for_recip_mul): Likewise.
8554 (aarch64_reassociation_width): Likewise.
8555 (aarch64_get_condition_code_1): Likewise.
8556 (aarch64_simd_emit_reg_reg_move): Likewise.
8557 (aarch64_simd_attr_length_rglist): Likewise.
8558 (aarch64_reverse_mask): Likewise.
8559 (aarch64_operands_ok_for_ldpstp): Likewise.
8560 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8561 (aarch64_gen_adjusted_ldpstp): Likewise.
8562 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8564 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8565 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8566 (arm_lookup_simd_builtin_type): Likewise.
8567 (arm_simd_builtin_type): Likewise.
8568 (arm_init_simd_builtin_types): Likewise.
8569 (arm_expand_builtin_args): Likewise.
8570 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8571 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8572 (ft32_setup_incoming_varargs): Likewise.
8573 (ft32_function_arg): Likewise.
8574 (ft32_function_arg_advance): Likewise.
8575 (ft32_pass_by_reference): Likewise.
8576 (ft32_arg_partial_bytes): Likewise.
8577 (ft32_valid_pointer_mode): Likewise.
8578 (ft32_addr_space_pointer_mode): Likewise.
8579 (ft32_addr_space_legitimate_address_p): Likewise.
8580 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8582 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8583 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8584 (iamcu_alignment): Likewise.
8585 (canonicalize_vector_int_perm): Likewise.
8586 (ix86_noce_conversion_profitable_p): Likewise.
8587 (ix86_mpx_bound_mode): Likewise.
8588 (ix86_operands_ok_for_move_multiple): Likewise.
8589 * config/microblaze/microblaze-protos.h
8590 (microblaze_expand_conditional_branch_reg): Likewise.
8591 * config/microblaze/microblaze.c
8592 (microblaze_expand_conditional_branch_reg): Likewise.
8593 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8595 (rs6000_reassociation_width): Likewise.
8596 (rs6000_invalid_binary_op): Likewise.
8597 (fusion_p9_p): Likewise.
8598 (emit_fusion_p9_load): Likewise.
8599 (emit_fusion_p9_store): Likewise.
8600 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8602 (riscv_hard_regno_mode_ok_p): Likewise.
8603 (riscv_address_insns): Likewise.
8604 (riscv_split_symbol): Likewise.
8605 (riscv_legitimize_move): Likewise.
8606 (riscv_function_value): Likewise.
8607 (riscv_hard_regno_nregs): Likewise.
8608 (riscv_expand_builtin): Likewise.
8609 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8610 (riscv_build_integer): Likewise.
8611 (riscv_split_integer): Likewise.
8612 (riscv_legitimate_constant_p): Likewise.
8613 (riscv_cannot_force_const_mem): Likewise.
8614 (riscv_regno_mode_ok_for_base_p): Likewise.
8615 (riscv_valid_base_register_p): Likewise.
8616 (riscv_valid_offset_p): Likewise.
8617 (riscv_valid_lo_sum_p): Likewise.
8618 (riscv_classify_address): Likewise.
8619 (riscv_legitimate_address_p): Likewise.
8620 (riscv_address_insns): Likewise.
8621 (riscv_load_store_insns): Likewise.
8622 (riscv_force_binary): Likewise.
8623 (riscv_split_symbol): Likewise.
8624 (riscv_force_address): Likewise.
8625 (riscv_legitimize_address): Likewise.
8626 (riscv_move_integer): Likewise.
8627 (riscv_legitimize_const_move): Likewise.
8628 (riscv_legitimize_move): Likewise.
8629 (riscv_address_cost): Likewise.
8630 (riscv_subword): Likewise.
8631 (riscv_output_move): Likewise.
8632 (riscv_canonicalize_int_order_test): Likewise.
8633 (riscv_emit_int_order_test): Likewise.
8634 (riscv_function_arg_boundary): Likewise.
8635 (riscv_pass_mode_in_fpr_p): Likewise.
8636 (riscv_pass_fpr_single): Likewise.
8637 (riscv_pass_fpr_pair): Likewise.
8638 (riscv_get_arg_info): Likewise.
8639 (riscv_function_arg): Likewise.
8640 (riscv_function_arg_advance): Likewise.
8641 (riscv_arg_partial_bytes): Likewise.
8642 (riscv_function_value): Likewise.
8643 (riscv_pass_by_reference): Likewise.
8644 (riscv_setup_incoming_varargs): Likewise.
8645 (riscv_print_operand): Likewise.
8646 (riscv_elf_select_rtx_section): Likewise.
8647 (riscv_save_restore_reg): Likewise.
8648 (riscv_for_each_saved_reg): Likewise.
8649 (riscv_register_move_cost): Likewise.
8650 (riscv_hard_regno_mode_ok_p): Likewise.
8651 (riscv_hard_regno_nregs): Likewise.
8652 (riscv_class_max_nregs): Likewise.
8653 (riscv_memory_move_cost): Likewise.
8654 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8655 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8656 (rl78_addr_space_address_mode): Likewise.
8657 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8659 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8660 (rs6000_reassociation_width): Likewise.
8661 (rs6000_invalid_binary_op): Likewise.
8662 (fusion_p9_p): Likewise.
8663 (emit_fusion_p9_load): Likewise.
8664 (emit_fusion_p9_store): Likewise.
8665 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8666 (ok_for_simple_move_operands): Likewise.
8667 (ok_for_simple_move_strict_operands): Likewise.
8668 (ok_for_simple_arith_logic_operands): Likewise.
8669 (visium_legitimize_reload_address): Likewise.
8670 (visium_select_cc_mode): Likewise.
8671 (output_cbranch): Likewise.
8672 (visium_split_double_move): Likewise.
8673 (visium_expand_copysign): Likewise.
8674 (visium_expand_int_cstore): Likewise.
8675 (visium_expand_fp_cstore): Likewise.
8676 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8677 (visium_function_arg): Likewise.
8678 (visium_function_arg_advance): Likewise.
8679 (visium_libcall_value): Likewise.
8680 (visium_setup_incoming_varargs): Likewise.
8681 (visium_legitimate_constant_p): Likewise.
8682 (visium_legitimate_address_p): Likewise.
8683 (visium_legitimize_address): Likewise.
8684 (visium_secondary_reload): Likewise.
8685 (visium_register_move_cost): Likewise.
8686 (visium_memory_move_cost): Likewise.
8687 (prepare_move_operands): Likewise.
8688 (ok_for_simple_move_operands): Likewise.
8689 (ok_for_simple_move_strict_operands): Likewise.
8690 (ok_for_simple_arith_logic_operands): Likewise.
8691 (visium_function_value_1): Likewise.
8692 (rtx_ok_for_offset_p): Likewise.
8693 (visium_legitimize_reload_address): Likewise.
8694 (visium_split_double_move): Likewise.
8695 (visium_expand_copysign): Likewise.
8696 (visium_expand_int_cstore): Likewise.
8697 (visium_expand_fp_cstore): Likewise.
8698 (visium_split_cstore): Likewise.
8699 (visium_select_cc_mode): Likewise.
8700 (visium_split_cbranch): Likewise.
8701 (output_cbranch): Likewise.
8702 (visium_print_operand_address): Likewise.
8703 * expmed.c (flip_storage_order): Likewise.
8704 * expmed.h (emit_cstore): Likewise.
8705 (flip_storage_order): Likewise.
8706 * genrecog.c (validate_pattern): Likewise.
8707 * hsa-gen.c (gen_hsa_addr): Likewise.
8708 * internal-fn.c (expand_arith_overflow): Likewise.
8709 * ira-color.c (allocno_copy_cost_saving): Likewise.
8710 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8711 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8712 (process_invariant_for_inheritance): Likewise.
8713 * lra-eliminations.c (move_plus_up): Likewise.
8714 * omp-low.c (lower_oacc_reductions): Likewise.
8715 * simplify-rtx.c (simplify_subreg): Likewise.
8716 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8717 (TARGET_CHKP_BOUND_MODE): Likewise..
8718 * targhooks.c (default_chkp_bound_mode): Likewise.
8719 (default_setup_incoming_vararg_bounds): Likewise.
8720 * targhooks.h (default_chkp_bound_mode): Likewise.
8721 (default_setup_incoming_vararg_bounds): Likewise.
8722 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8723 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8724 (have_whole_vector_shift): Likewise.
8725 * tree-vect-stmts.c (vectorizable_load): Likewise.
8726 * doc/tm.texi: Regenerate.
8728 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8730 Graceful degrade if Binutils PR21472 is not available.
8733 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8734 .rodata in flash test fails.
8735 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8736 * confgure: Regenerate.
8737 * config.in: Regenerate.
8738 * config/avr/avr.c (avr_asm_named_section)
8739 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8740 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8741 (avr_asm_init_sections): Same.
8743 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8745 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8746 (fma<VH:mode>4_intrinsic): Likewise.
8747 (*fmsub<VCVTF:mode>4): Likewise.
8748 (*fmsub<VH:mode>4_intrinsic): Likewise.
8750 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8753 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8754 Don't depend on "optimize > 0".
8755 (out_movhi_r_mr, out_movqi_mr_r): Same.
8756 (out_movhi_mr_r, out_movqi_r_mr): Same.
8757 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8758 io_address_operand on "optimize > 0".
8760 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8762 * tree-loop-distribution.c: Add general explanantion on the pass.
8763 (generate_loops_for_partition): Mark distributed loop.
8764 (pg_add_dependence_edges): New parameter. Handle alias data
8765 dependence specially and record it in the parameter if asked.
8766 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8767 (init_partition_graph_vertices, add_partition_graph_edge): New.
8768 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8769 (free_partition_graph_vdata, build_partition_graph): New.
8770 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8771 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8772 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8773 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8774 (version_for_distribution_p, finalize_partitions): New.
8775 (distribute_loop): Handle alias data dependence specially. Factor
8776 out loop fusion code as functions and call these functions.
8778 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8780 * tree-loop-distribution.c (classify_partition): New parameter and
8781 better handle reduction statement.
8782 (rdg_build_partitions): Revise comment.
8783 (distribute_loop): Compute statements in all partitions and pass it
8784 to classify_partition.
8786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8788 * tree-loop-distribution.c (enum partition_type): New.
8789 (struct partition): New field type.
8790 (partition_merge_into): Add parameter. Update partition type.
8791 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8792 (build_rdg_partition_for_vertex): Compute partition type.
8793 (rdg_build_partitions): Dump partition type.
8794 (distribute_loop): Update calls to partition_merge_into.
8796 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8798 * tree-loop-distribution.c (struct ddr_hasher): New.
8799 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8801 (classify_partition): Call get_data_dependence.
8802 (pg_add_dependence_edges): Ditto.
8803 (distribute_loop): Release data dependence hash table.
8805 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8807 * tree-loop-distribution.c (ref_base_address): Delete.
8808 (similar_memory_accesses): Rename ...
8809 (share_memory_accesses): ... to this. Check if partitions access
8810 the same memory reference.
8811 (distribute_loop): Call share_memory_accesses.
8813 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8815 * tree-loop-distribution.c (struct partition): New field recording
8817 (partition_alloc, partition_free): Init and release data refs.
8818 (partition_merge_into): Merge data refs.
8819 (build_rdg_partition_for_vertex): Collect data refs for partition.
8820 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8822 (distribute_loop): Remve data refs from vertice data of partition
8825 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8827 * tree-loop-distribution.c (params.h): Include header file.
8828 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8829 (datarefs_vec): New global var.
8830 (create_rdg_vertices): Use datarefs_vec directly.
8831 (free_rdg): Don't free data references.
8832 (build_rdg): Update use. Don't free data references.
8833 (distribute_loop): Compute global variable for data references.
8834 Bail out if there are too many data references.
8836 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8838 * tree-loop-distribution.c (loop_nest): New global var.
8839 (build_rdg): Use loop directly, rather than loop nest.
8840 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8842 (distribute_loop): Compute global variable loop nest. Update use.
8844 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8846 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8847 (partition_merge_into): New parameter. Dump reason for fusion.
8848 (distribute_loop): Update use of partition_merge_into.
8850 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8852 * tree-loop-distribution.c (bb_top_order_index): New.
8853 (bb_top_order_index_size, bb_top_order_cmp): New.
8854 (stmts_from_loop): Use topological order.
8855 (pass_loop_distribution::execute): Compute and release topological
8856 order for basic blocks.
8858 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8860 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8863 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8865 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
8866 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
8867 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
8868 * internal-fn.def (LOOP_DIST_ALIAS): New.
8869 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
8870 (fold_loop_internal_call): ... this.
8871 (vect_loop_dist_alias_call): New function.
8872 (set_uid_loop_bbs): Call fold_loop_internal_call.
8873 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
8876 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8879 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
8880 Require dead FLAGS_REG at the beginning of a peephole.
8882 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8885 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
8886 arguments in the call to __builtin_ia32_sbb_u32.
8887 (_subborrow_u64): Swap _X and _Y arguments in the call to
8888 __builtin_ia32_sbb_u64.
8890 2017-07-04 Jakub Jelinek <jakub@redhat.com>
8893 * tree-vrp.c (compare_assert_loc): Turn into a function template
8894 with stable template parameter. Only test if a->e is NULL,
8895 !a->e == !b->e has been verified already. Use e == NULL or
8896 e != NULL instead of e or ! e tests. If stable is true, don't use
8897 iterative_hash_expr, on the other side allow a or b or both NULL
8898 and sort the NULLs last.
8899 (process_assert_insertions): Sort using compare_assert_loc<false>
8900 instead of compare_assert_loc, later sort using
8901 compare_assert_loc<true> before calling process_assert_insertions_for
8902 in a loop. Use break instead of continue once seen NULL pointer.
8904 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8906 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8907 Cortex-R7 and Cortex-R8 processors.
8909 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8911 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8912 uninitialized while src is not.
8914 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8916 * common/config/arm/arm-common.c: Adjust include path for
8918 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8919 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8920 (arm-cpu-data.h): Likewise.
8921 (arm-cpu-cdata.h): Likewise.
8922 * config/arm/arm-cpu.h: Delete.
8923 * config/arm/arm-cpu-cdata.h: Delete.
8924 * config/arm/arm-cpu-data.h: Delete.
8926 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8928 * config/arm/arm-cpus.in (cortex-a55): New.
8929 (cortex-a75): Likewise.
8930 (cortex-a75.cortex-a55): Likewise.
8931 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8933 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8934 * config/arm/arm-cpu-cdata.h: Regenerate.
8935 * config/arm/arm-cpu-data.h: Regenerate.
8936 * config/arm/arm-cpu.h: Regenerate.
8937 * config/arm/arm-tables.opt: Regenerate.
8938 * config/arm/arm-tune.md: Regenerate.
8940 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8942 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8944 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8946 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8949 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8951 PR tree-optimization/81292
8952 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8953 full_string_p, also call adjust_related_strinfos if the adjustment
8954 is simple, otherwise invalidate related strinfos.
8956 2017-07-04 Martin Liska <mliska@suse.cz>
8959 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8960 newly created variable as DECL_IGNORED_P.
8962 2017-07-04 Martin Liska <mliska@suse.cz>
8965 * ipa-inline.c (inline_small_functions):
8966 Use xstrdup_for_dump.
8968 2017-07-04 Tom de Vries <tom@codesourcery.com>
8970 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8972 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8975 * config/darwin.c (darwin_function_switched_text_sections):
8978 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8980 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8982 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8984 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8986 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8988 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8989 min_profitable_iters, and th as inclusive lower bounds.
8990 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8991 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8992 for min_profitable_iters and min_profitable_estimate.
8993 (vect_transform_loop): Treat th as an inclusive lower bound.
8994 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8996 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8999 * config/darwin.c (darwin_function_switched_text_sections):
9000 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9001 in two pieces, and suppress the use of buf.
9003 2017-07-03 Nathan Sidwell <nathan@acm.org>
9005 * hash-table.h (hash_table_mod1): Fix indentation.
9007 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9010 * predict.c (force_edge_cold): Be more careful about propagation
9012 * profile-count.h (profile_probability::guessed,
9013 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9015 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9017 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9019 * doc/invoke.texi (rcpc architecture extension): Document it.
9021 2017-07-03 Richard Biener <rguenther@suse.de>
9023 PR tree-optimization/60510
9024 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9025 the scalar reduction PHI and use it.
9026 (vectorizable_reduction): Properly guard the single_defuse_cycle
9027 path for non-SLP reduction chains where we cannot use it.
9028 Rework reduc_def/index and vector type deduction. Rework
9029 vector operand gathering during reduction op code-gen.
9030 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9031 chains dissolve the chain and leave it to non-SLP reduction
9034 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9036 * tree-data-ref.h (dr_alignment): Declare.
9037 * tree-data-ref.c (dr_alignment): New function.
9038 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9039 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9041 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9043 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9045 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9046 and base_misalignment fields.
9047 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9048 * tree-data-ref.c: Include builtins.h.
9049 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9050 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9051 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9052 * tree-vect-data-refs.c: Include tree-cfg.h.
9053 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9054 fields instead of calculating an alignment here.
9055 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9056 innermost_loop_behavior fields.
9058 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9060 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9062 (DR_STEP_ALIGNMENT): New macro.
9063 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9064 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9065 (create_data_ref): Print it.
9066 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9067 to tell whether the step preserves vector (mis)alignment.
9068 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9069 Move the check for an integer step and generalise to all INTEGER_CST.
9070 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9071 Print the outer step alignment.
9073 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9075 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9076 with offset_alignment.
9077 (DR_ALIGNED_TO): Delete.
9078 (DR_OFFSET_ALIGNMENT): New macro.
9079 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9080 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9081 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9082 (create_data_ref): Likewise.
9083 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9084 (vect_analyze_data_refs): Likewise.
9085 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9086 creating dummy innermost behavior.
9088 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9090 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9091 with a "innermost_loop_behavior *" and refeence tree.
9092 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9093 (create_data_ref): Update call accordingly.
9094 * tree-predcom.c (find_looparound_phi): Likewise.
9096 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9098 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9099 fields with dr_wrt_vec_loop.
9100 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9101 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9102 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9103 (vect_dr_behavior): New function.
9104 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9105 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9106 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9107 track whether the step preserves the misalignment.
9108 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9109 Use vect_dr_behavior.
9110 (vect_setup_realignment): Update call accordingly.
9111 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9112 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9113 call to vect_create_addr_base_for_vector_ref.
9114 (vect_create_cond_for_align_checks): Likewise.
9115 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9116 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9117 (vect_recog_mask_conversion_pattern): Likewise.
9118 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9119 (new_stmt_vec_info): Remove redundant zeroing.
9121 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9123 * common/config/arm/arm-common.c (arm_be8_option): New function.
9124 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9125 (ISA_ARMv6): Add isa_bit_be8.
9126 * config/arm/arm.h (arm_be8_option): Add prototype.
9127 (BE8_SPEC_FUNCTION): New define.
9128 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9129 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9130 (mlittle-endian): Similarly.
9131 (mbe8, mbe32): New options.
9132 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9133 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9135 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9137 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9139 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9141 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9142 (cleanup_tree_cfg_bb): Use it.
9143 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9145 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9147 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9150 * loop-doloop.c (add_test): Update profile.
9152 2017-07-03 Martin Liska <mliska@suse.cz>
9155 * sanopt.c (rewrite_usage_of_param): New function.
9156 (sanitize_rewrite_addressable_params): Likewise.
9157 (pass_sanopt::execute): Call rewrite_usage_of_param.
9159 2017-07-03 Richard Biener <rguenther@suse.de>
9161 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9162 back to using VIEW_CONVERT_EXPR.
9164 2017-07-03 Martin Liska <mliska@suse.cz>
9167 * doc/extend.texi: Document when a resolver function is
9168 generated for target_clones.
9170 2017-07-03 Martin Liska <mliska@suse.cz>
9172 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9173 and shadow memory of auto variables which are subject of
9174 use-after-scope sanitization.
9175 (asan_expand_mark_ifn): Add do set only when is_poison.
9177 2016-07-03 Richard Biener <rguenther@suse.de>
9179 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9181 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9182 (vectorizable_reduction): Perform reduction PHI creation when
9183 visiting a reduction PHI and adjust and simplify code generation
9184 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9185 (vect_transform_loop): Visit reduction PHIs.
9186 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9187 defs into the SLP tree.
9188 (vect_build_slp_tree): Reduction defs terminate the recursion.
9189 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9191 (vect_get_vec_defs_for_stmt_copy): Export.
9192 (vect_get_vec_defs): Likewise.
9193 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9195 (vect_get_vec_defs_for_stmt_copy): Declare.
9196 (vect_get_vec_defs): Likewise.
9198 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9200 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9201 parameter with a "loop" parameter and use it instead of the
9202 loop containing DR_STMT. Don't check simple_iv when doing
9203 BB analysis. Describe the two analysis modes in the comment.
9205 2017-07-03 Tom de Vries <tom@codesourcery.com>
9207 PR tree-optimization/69468
9208 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9209 (find_same_succ_bb): Handle ignore_edge_flags.
9211 2017-07-03 Tom de Vries <tom@codesourcery.com>
9213 PR tree-optimization/81192
9214 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9216 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9218 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9220 2017-07-03 Tom de Vries <tom@codesourcery.com>
9222 PR tree-optimization/81192
9223 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9224 BB_SAME_SUCC (bb) == NULL.
9226 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9228 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9231 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9233 * dumpfile.c: Include profile-count.h
9234 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9236 (insert_cond_bb): Update profile.
9237 * tree-cfg.h (insert_cond_bb): Update prototype.
9238 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9239 * tree-dump.c: Do not include tree-cfg.
9241 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9243 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9245 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9247 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9250 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9252 * tree-complex.c (expand_complex_div_wide): update profile.
9254 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9255 Alan Hayward <alan.hayward@arm.com>
9256 David Sherwood <david.sherwood@arm.com>
9258 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9259 (CORETYPES_H): New define.
9260 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9261 (insn-modes-inline.h, s-modes-inline-h): New rules.
9262 (generated_files): Add insn-modes-inline.h.
9263 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9264 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9265 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9266 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9267 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9268 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9269 (build/gencodes.o, build/genconditions.o): Likewise.
9270 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9271 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9272 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9273 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9274 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9275 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9276 * coretypes.h: Include everything up to real.h for generators.
9277 Include insn-modes.h first. Include wide-int-print.h after
9278 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9279 * machmode.h: Don't include insn-modes.h here.
9280 * function-tests.c: Remove includes of signop.h, machmode.h,
9281 double-int.h and wide-int.h.
9283 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9285 * optc-save-gen.awk: Likewise.
9286 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9287 * godump.c: Remove include of wide-int-print.h.
9288 * pretty-print.h: Likewise.
9289 * wide-int-print.cc: Likewise.
9290 * wide-int.cc: Likewise.
9291 * hash-map-tests.c: Remove include of signop.h.
9292 * hash-set-tests.c: Likewise.
9293 * rtl-tests.c: Likewise.
9294 * mkconfig.sh: Remove include of machmode.h.
9295 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9297 (emit_insn_modes_inline_h): ...this new function. Emit the code
9298 into an insn-modes-inline.h header file, adding appropriate
9299 include guards and end comments.
9300 (emit_insn_modes_c_header): Remove include of machmode.h.
9301 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9303 (main): Handle -i flag and call emit_insn_modes_inline_h when
9306 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9308 * tree-ssa-strlen.c (strinfo): Rename the length field to
9309 nonzero_chars. Add a full_string_p field.
9310 (compare_nonzero_chars, zero_length_string_p): New functions.
9311 (get_addr_stridx): Add an offset_out parameter.
9312 Use compare_nonzero_chars.
9313 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9314 (new_strinfo): Update after above changes to strinfo.
9315 (set_endptr_and_length): Set full_string_p.
9316 (get_string_length): Update after above changes to strinfo.
9317 (unshare_strinfo): Update call to new_strinfo.
9318 (maybe_invalidate): Likewise.
9319 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9320 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9321 as a uhwi instead of an shwi. Update after above changes to
9322 strinfo and new_strinfo.
9323 (zero_length_string): Assert that chainsi contains full strings.
9324 Use zero_length_string_p. Update call to new_strinfo.
9325 (adjust_related_strinfos): Update after above changes to strinfo.
9326 Copy full_string_p from origsi.
9327 (adjust_last_stmt): Use zero_length_string_p.
9328 (handle_builtin_strlen): Update after above changes to strinfo and
9329 new_strinfo. Install the lhs as the string length if the previous
9330 entry didn't describe a full string.
9331 (handle_builtin_strchr): Update after above changes to strinfo
9333 (handle_builtin_strcpy): Likewise.
9334 (handle_builtin_strcat): Likewise.
9335 (handle_builtin_malloc): Likewise.
9336 (handle_pointer_plus): Likewise.
9337 (handle_builtin_memcpy): Likewise. Track nonzero characters
9338 that aren't necessarily followed by a nul terminator.
9339 (handle_char_store): Likewise.
9341 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9343 PR tree-optimization/80769
9344 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9345 for malloc and calloc. Document the new invariant that all related
9346 strinfos have delayed lengths or none do.
9347 (verify_related_strinfos): Move earlier in file.
9348 (set_endptr_and_length): New function, split out from...
9349 (get_string_length): ...here. Also set the lengths of related
9351 (zero_length_string): Assert that chainsi has known (rather than
9353 (adjust_related_strinfos): Likewise.
9355 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9357 PR tree-optimization/81136
9358 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9359 assert that two references with the same misalignment have the same
9360 compile-time misalignment if those compile-time misalignments
9363 2017-07-01 Andi Kleen <ak@linux.intel.com>
9365 * print-tree.c (print_node): Print all attributes.
9367 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9369 * cfg.c (scale_bbs_frequencies): New function.
9370 * cfg.h (scale_bbs_frequencies): Declare it.
9371 * cfgloopanal.c (single_likely_exit): Cleanup.
9372 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9374 (scale_loop_profile): Likewise.
9375 (loop_version): Likewise.
9376 (create_empty_loop_on_edge): Update.
9377 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9378 scale_loop_frequencies, scale_loop_profile, loopify,
9379 loop_version): Update prototypes.
9380 * modulo-sched.c (sms_schedule): Update.
9381 * predict.c (unlikely_executed_edge_p): Also check probability.
9382 (probably_never_executed_edge_p): Fix typo.
9383 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9384 * tree-parloops.c (gen_parallel_loop): Update.
9385 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9386 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9387 * tree-ssa-loop-split.c (split_loop): Update.
9388 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9389 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9390 (vect_loop_versioning): Update.
9391 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9393 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9395 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9397 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9399 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9400 to keep profile consistent.
9402 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9404 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9405 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9406 * profile-count.h (max_safe_multiplier): Make unsigned.
9407 (profile_count::guessed_zero): New.
9409 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9411 * bb-reorder.c (fix_up_crossing_landing_pad,
9412 fix_crossing_conditional_branches): Use make_single_succ_edge
9413 to keep profile consistent.
9415 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9417 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9420 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9423 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9424 the right scopes, make sure cond_jump isn't preserved between multiple
9425 iterations. Search for fallthru edge whenever there are 3+ edges and
9426 use find_fallthru_edge for it.
9428 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9430 Patch by Alexander Monakov <amonakov@ispras.ru>
9431 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9432 probabilities consistently.
9434 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9436 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9437 emit_cmp_and_jump_insns.
9439 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9442 * tree-inline.c (expand_call_inline): Combine profile statuses.
9444 2017-06-30 Andrew Pinski <apinski@cavium.com>
9446 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9447 fold_stmt returned true.
9449 2017-06-30 Nathan Sidwell <nathan@acm.org>
9451 * ggc.h (empty_string): Delete.
9452 * cfgexpand.c (expand_asm_stmt): Use plain "".
9453 * optabs.c (expand_asm_memory_barrier): Likewise.
9454 * stringpool.c (empty_string): Delete.
9455 (digit_vector, digit_string): Delete.
9456 (ggc_alloc_string): Use plain "", don't optimize single digit
9457 strings. Use ggc_alloc_atomic.
9459 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9461 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9462 comparison set and one other set, use the cost of the non-comparison
9465 2017-06-30 Nathan Sidwell <nathan@acm.org>
9467 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9470 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9472 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9473 loops. Remove now unneeded calls to gimple_switch_set_label() that
9474 just set removed labels to NULL_TREE.
9476 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9478 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9479 (set_range_info): ...here. Only call set_range_info_raw if domain
9481 (set_nonzero_bits): Call set_range_info_raw.
9482 * tree-ssanames.h (set_range_info_raw): New.
9484 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9487 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9488 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9489 of nonimmediate_operand and <store_mask_constraint> instead of m
9490 for the input operand. For V8FI iterator, always split if input
9491 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9492 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9493 <store_mask_predicate> instead of register_operand and
9494 <store_mask_constraint> instead of v for the input operand. Make
9495 sure both operands aren't MEMs for if not <mask_applied>.
9497 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9499 * lto-wrapper.c (copy_file) Close both file descriptors before
9502 2017-06-30 Martin Liska <mliska@suse.cz>
9505 * multiple_target.c (create_dispatcher_calls): Make ifunc
9506 also for function that don't have calls or are not referenced.
9508 2017-06-30 Richard Biener <rguenther@suse.de>
9510 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9511 analyze the first scalar stmt. Move vector type computation
9512 for the BB case here from ...
9513 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9514 live operation processing in the SLP case properly.
9516 2017-06-30 Richard Biener <rguenther@suse.de>
9518 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9520 2017-06-30 Martin Liska <mliska@suse.cz>
9523 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9524 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9526 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9528 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9531 2017-06-30 Richard Biener <rguenther@suse.de>
9533 PR tree-optimization/81249
9534 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9535 condition reduction result to original scalar type.
9537 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9539 * profile-count.h (enum profile_quality): Fix typos and whitespace
9542 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9544 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9545 type for branch probabilities.
9547 2017-06-29 Julian Brown <julian@codesourcery.com>
9548 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9550 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9551 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9552 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9553 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9555 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9557 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9558 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9559 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9560 CC usage from generic code to here.
9561 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9562 CC usage into the target macros.
9564 2017-06-29 Maya Rashish <coypu@sdf.org>
9566 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9569 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9571 * arm/arm-builtins.c: Include profile-count.h
9572 * except.c (sjlj_emit_function_enter): Use
9573 profile_probability::unlikely.
9575 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9577 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9578 and tocrel_offset be pointer args rather than implicitly using
9580 (legitimate_constant_pool_address_p, rs6000_emit_move,
9581 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9582 tocrel_offset and use in toc_relative_expr_p call.
9583 (print_operand, print_operand_address): Use static tocrel_base_oac
9584 and tocrel_offset_oac.
9585 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9588 2017-06-29 Maya Rashish <coypu@sdf.org>
9590 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9592 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9594 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9595 objects, take into account only the alignment of 'op0' and 'mode1' if
9598 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9600 * ccmp.c (ccmp_tree_comparison_p): New function.
9601 (ccmp_candidate_p): Update to use above function.
9602 (get_compare_parts): New function.
9603 (expand_ccmp_next): Update to use new functions.
9604 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9606 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9607 take mode as argument.
9608 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9609 * expr.c (expand_expr_real_1): Pass mode as argument.
9611 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9613 * combine.c (combine_instructions): Print insns to dump_file, together
9616 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9618 * asan.c (asan_emit_stack_protection): Update.
9619 (create_cond_insert_point): Update.
9620 * auto-profile.c (afdo_propagate_circuit): Update.
9621 * basic-block.h (struct edge_def): Turn probability to
9622 profile_probability.
9623 (EDGE_FREQUENCY): Update.
9624 * bb-reorder.c (find_traces_1_round): Update.
9625 (better_edge_p): Update.
9626 (sanitize_hot_paths): Update.
9627 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9628 (make_single_succ_edge): Update.
9629 (check_bb_profile): Update.
9630 (dump_edge_info): Update.
9631 (update_bb_profile_for_threading): Update.
9632 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9633 probabilitycount to 0.
9634 * cfgbuild.c (compute_outgoing_frequencies): Update.
9635 * cfgcleanup.c (try_forward_edges): Update.
9636 (outgoing_edges_match): Update.
9637 (try_crossjump_to_edge): Update.
9638 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9639 (expand_gimple_tailcall): Update.
9640 (construct_init_block): Use make_single_succ_edge.
9641 (construct_exit_block): Use make_single_succ_edge.
9642 * cfghooks.c (verify_flow_info): Update.
9643 (redirect_edge_succ_nodup): Update.
9644 (split_edge): Update.
9645 (account_profile_record): Update.
9646 * cfgloopanal.c (single_likely_exit): Update.
9647 * cfgloopmanip.c (scale_loop_profile): Update.
9648 (set_zero_probability): Remove.
9649 (duplicate_loop_to_header_edge): Update.
9650 * cfgloopmanip.h (loop_version): Update prototype.
9651 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9652 (force_nonfallthru_and_redirect): Update.
9653 (update_br_prob_note): Update.
9654 (rtl_verify_edges): Update.
9655 (purge_dead_edges): Update.
9656 (rtl_lv_add_condition_to_bb): Update.
9657 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9658 * cgraphunit.c (init_lowered_empty_function): Update.
9659 (cgraph_node::expand_thunk): Update.
9660 * cilk-common.c: Include profile-count.h
9661 * dojump.c (inv): Remove.
9662 (jumpifnot): Update.
9663 (jumpifnot_1): Update.
9664 (do_jump_1): Update.
9666 (do_jump_by_parts_greater_rtx): Update.
9667 (do_compare_rtx_and_jump): Update.
9668 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9669 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9670 * dwarf2cfi.c: Include profile-count.h
9671 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9672 (sjlj_emit_dispatch_table): Likewise.
9673 * explow.c: Include profile-count.h
9674 * expmed.c (emit_store_flag_force): Update.
9675 (do_cmp_and_jump): Update.
9676 * expr.c (compare_by_pieces_d::generate): Update.
9677 (compare_by_pieces_d::finish_mode): Update.
9678 (emit_block_move_via_loop): Update.
9679 (store_expr_with_bounds): Update.
9680 (store_constructor): Update.
9681 (expand_expr_real_2): Update.
9682 (expand_expr_real_1): Update.
9683 * expr.h (try_casesi, try_tablejump): Update prototypes.
9684 * gimple-pretty-print.c (dump_probability): Update.
9685 (dump_profile): New.
9686 (dump_gimple_label): Update.
9687 (dump_gimple_bb_header): Update.
9688 * graph.c (draw_cfg_node_succ_edges): Update.
9689 * hsa-gen.c (convert_switch_statements): Update.
9690 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9691 (find_if_case_1): Update.
9692 (find_if_case_2): Update.
9693 * internal-fn.c (expand_arith_overflow_result_store): Update.
9694 (expand_addsub_overflow): Update.
9695 (expand_neg_overflow): Update.
9696 (expand_mul_overflow): Update.
9697 (expand_vector_ubsan_overflow): Update.
9698 * ipa-cp.c (good_cloning_opportunity_p): Update.
9699 * ipa-split.c (split_function): Use make_single_succ_edge.
9700 * ipa-utils.c (ipa_merge_profiles): Update.
9701 * loop-doloop.c (add_test): Update.
9702 (doloop_modify): Update.
9703 * loop-unroll.c (compare_and_jump_seq): Update.
9704 (unroll_loop_runtime_iterations): Update.
9705 * lra-constraints.c (lra_inheritance): Update.
9706 * lto-streamer-in.c (input_cfg): Update.
9707 * lto-streamer-out.c (output_cfg): Update.
9708 * mcf.c (adjust_cfg_counts): Update.
9709 * modulo-sched.c (sms_schedule): Update.
9710 * omp-expand.c (expand_omp_for_init_counts): Update.
9711 (extract_omp_for_update_vars): Update.
9712 (expand_omp_ordered_sink): Update.
9713 (expand_omp_for_ordered_loops): Update.
9714 (expand_omp_for_generic): Update.
9715 (expand_omp_for_static_nochunk): Update.
9716 (expand_omp_for_static_chunk): Update.
9717 (expand_cilk_for): Update.
9718 (expand_omp_simd): Update.
9719 (expand_omp_taskloop_for_outer): Update.
9720 (expand_omp_taskloop_for_inner): Update.
9721 * omp-simd-clone.c (simd_clone_adjust): Update.
9722 * optabs.c (expand_doubleword_shift): Update.
9723 (expand_abs): Update.
9724 (emit_cmp_and_jump_insn_1): Update.
9725 (expand_compare_and_swap_loop): Update.
9726 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9727 * predict.c (predictable_edge_p): Update.
9728 (edge_probability_reliable_p): Update.
9729 (set_even_probabilities): Update.
9730 (combine_predictions_for_insn): Update.
9731 (combine_predictions_for_bb): Update.
9732 (propagate_freq): Update.
9733 (estimate_bb_frequencies): Update.
9734 (force_edge_cold): Update.
9735 * profile-count.c (profile_count::dump): Add missing space into dump.
9736 (profile_count::debug): Add newline.
9737 (profile_count::differs_from_p): Explicitly convert to unsigned.
9738 (profile_count::stream_in): Update.
9739 (profile_probability::dump): New member function.
9740 (profile_probability::debug): New member function.
9741 (profile_probability::differs_from_p): New member function.
9742 (profile_probability::differs_lot_from_p): New member function.
9743 (profile_probability::stream_in): New member function.
9744 (profile_probability::stream_out): New member function.
9745 * profile-count.h (profile_count_quality): Rename to ...
9746 (profile_quality): ... this one.
9747 (profile_probability): New.
9748 (profile_count): Update.
9749 * profile.c (compute_branch_probabilities): Update.
9750 * recog.c (peep2_attempt): Update.
9751 * sched-ebb.c (schedule_ebbs): Update.
9752 * sched-rgn.c (find_single_block_region): Update.
9753 (compute_dom_prob_ps): Update.
9754 (schedule_region): Update.
9755 * sel-sched-ir.c (compute_succs_info): Update.
9756 * stmt.c (struct case_node): Update.
9757 (do_jump_if_equal): Update.
9758 (get_outgoing_edge_probs): Update.
9759 (conditional_probability): Update.
9760 (emit_case_dispatch_table): Update.
9761 (expand_case): Update.
9762 (expand_sjlj_dispatch_table): Update.
9763 (emit_case_nodes): Update.
9764 * targhooks.c: Update.
9765 * tracer.c (better_p): Update.
9766 (find_best_successor): Update.
9767 * trans-mem.c (expand_transaction): Update.
9768 * tree-call-cdce.c: Update.
9769 * tree-cfg.c (gimple_split_edge): Upate.
9770 (move_sese_region_to_fn): Upate.
9771 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9772 * tree-eh.c (lower_resx): Upate.
9773 (cleanup_empty_eh_move_lp): Upate.
9774 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9775 * tree-inline.c (copy_edges_for_bb): Update.
9776 (copy_cfg_body): Update.
9777 * tree-parloops.c (gen_parallel_loop): Update.
9778 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9779 (gimple_gen_time_profiler): Update.
9780 * tree-ssa-dce.c (remove_dead_stmt): Update.
9781 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9782 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9783 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9784 (unloop_loops): Update.
9785 (try_peel_loop): Update.
9786 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9787 * tree-ssa-loop-split.c (connect_loops): Update.
9788 (split_loop): Update.
9789 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9790 (hoist_guard): Update.
9791 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9792 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9793 (value_replacement): Update.
9794 * tree-ssa-reassoc.c (branch_fixup): Update.
9795 * tree-ssa-tail-merge.c (replace_block_by): Update.
9796 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9797 (create_edge_and_update_destination_phis): Update.
9798 (compute_path_counts): Update.
9799 (recompute_probabilities): Update.
9800 (update_joiner_offpath_counts): Update.
9801 (freqs_to_counts_path): Update.
9802 (duplicate_thread_path): Update.
9803 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9804 (struct switch_conv_info): Update.
9805 (gen_inbound_check): Update.
9806 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9807 (vect_do_peeling): Update.
9808 (vect_loop_versioning): Update.
9809 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9810 (optimize_mask_stores): Update.
9811 * ubsan.c (ubsan_expand_null_ifn): Update.
9812 * value-prof.c (gimple_divmod_fixed_value): Update.
9813 (gimple_divmod_fixed_value_transform): Update.
9814 (gimple_mod_pow2): Update.
9815 (gimple_mod_pow2_value_transform): Update.
9816 (gimple_mod_subtract): Update.
9817 (gimple_mod_subtract_transform): Update.
9818 (gimple_ic): Update.
9819 (gimple_stringop_fixed_value): Update.
9820 (gimple_stringops_transform): Update.
9821 * value-prof.h: Update.
9823 2017-06-29 Carl Love <cel@us.ibm.com>
9825 * config/rs6000/rs6000-c.c: Add support for built-in functions
9826 vector signed int vec_signed (vector float);
9827 vector signed long long vec_signed (vector double);
9828 vector signed int vec_signed2 (vector double, vector double);
9829 vector signed int vec_signede (vector double);
9830 vector signed int vec_signedo (vector double);
9831 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9832 instruction generator.
9833 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9834 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9835 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9837 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9838 vunsignede_v2df): Add define_expands.
9839 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9840 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9841 VEC_UNSIGNEDO): Add definitions.
9842 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9843 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9844 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9845 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9846 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9847 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9848 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9849 vec_unsignede, vec_unsignedo): Add builtin defines.
9850 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9852 * doc/extend.texi: Update the built-in documentation file for the
9853 new built-in functions.
9855 2017-06-29 Richard Biener <rguenther@suse.de>
9857 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9858 reduction chains to LOOP_VINFO_REDUCTIONS.
9859 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9860 SLP reductions after processing reduction chains.
9862 2017-06-29 Nathan Sidwell <nathan@acm.org>
9864 * builtins.c (fold_builtin_FUNCTION): Use
9865 lang_hooks.decl_printable_name.
9867 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
9870 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
9871 with only one label.
9872 * stmt.c (expand_case): Assert NCASES is greater than one.
9874 2017-06-29 Richard Biener <rguenther@suse.de>
9876 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
9878 (group_case_labels): Likewise.
9879 (find_taken_edge): Push sanity checking on val to workers...
9880 (find_taken_edge_cond_expr): ... here
9881 (find_taken_edge_switch_expr): ... and here, handle cases
9882 with just a default label.
9883 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
9884 (group_case_labels): Likewise.
9885 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
9886 group_case_labels does anything cleanup the CFG again.
9888 2017-06-29 Bin Cheng <bin.cheng@arm.com>
9890 PR tree-optimization/81196
9891 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
9892 exit condition comparing two IVs.
9894 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
9896 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
9897 profile to the dummy entry at the end of the list of architectures.
9898 * config/arm/arm-cpu-cdata.h: Regenerated.
9900 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9901 Michael Collison <michael.collison@arm.com>
9904 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9906 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9907 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9908 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9909 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9910 of shift when the shift amount is masked with constant equal to
9911 the size of the mode.
9912 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9915 2017-06-29 Martin Liska <mliska@suse.cz>
9917 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9919 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9922 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9925 (edge_growth_cache_entry::edge_growth_cache_entry): New
9927 (reset_edge_growth_cache): Update to use constructor.
9929 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9931 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9932 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9933 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9935 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9937 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9938 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9940 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9942 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9943 (*-linux-uclibc*): Add t-uclibc tmake_file.
9944 * config/t-musl: New.
9945 * config/t-uclibc: New.
9947 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9949 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9951 (gen_comm_data): Emit architectural setting of arch_prof.
9952 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9954 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9955 (armv8-m.base, armv8-m.main): Likewise.
9956 * arm-protos.h (arm_build_target): Add profile field.
9957 (arch_option): Likewise.
9958 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9960 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9961 arm_active_target.profile.
9963 2017-06-28 Richard Biener <rguenther@suse.de>
9966 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9967 TYPE_OVERFLOW_WRAPS.
9968 * match.pd (negate_expr_p): Likewise.
9969 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9970 fold_build2, not fold_binary.
9972 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9974 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9975 Convert memory address to Pmode.
9976 (aarch64_print_operand): Assert MEM operands are always Pmode.
9978 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9981 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9982 Remove redundant if.
9983 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9984 * config/arm/aarch-common-protos.h
9985 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9986 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9988 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9991 * multiple_target.c (create_dispatcher_calls): Set the default
9992 clone to be static, not public.
9994 2017-06-28 Richard Biener <rguenther@suse.de>
9996 * tree-vect-loop.c (vectorizable_reduction): Move special
9997 cond reduction IV var creation ...
9998 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9999 parameter. Use STMT_VINFO_VECTYPE.
10000 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10003 2017-06-28 Martin Liska <mliska@suse.cz>
10006 * ipa-visibility.c (non_local_p): Handle visibility.
10008 2017-06-28 Martin Liska <mliska@suse.cz>
10011 * common.opt: Add IntegerRange to various options.
10012 * opt-functions.awk (integer_range_info): New function.
10013 * optc-gen.awk: Add integer_range_info to cl_options struct.
10014 * opts-common.c (decode_cmdline_option): Handle
10015 CL_ERR_INT_RANGE_ARG.
10016 (cmdline_handle_error): Likewise.
10017 * opts.c (print_filtered_help): Show valid interval in
10018 when --help is provided.
10019 * opts.h (struct cl_option): Add range_min and range_max fields.
10020 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10022 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10024 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10025 (x * C EQ/NE y * C): New transformation.
10027 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10029 * genmultilib (combination_space): Accept '+' in option names.
10031 2017-06-28 Martin Liska <mliska@suse.cz>
10034 * asan.c (instrument_derefs): Bail out inner references
10035 that are hard register variables.
10037 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10040 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10041 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10043 2017-06-28 Richard Biener <rguenther@suse.de>
10045 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10046 (vect_get_slp_defs): Adjust.
10047 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10049 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10051 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10052 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10053 (vectorizable_reduction): Adjust.
10054 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10056 (vect_get_slp_defs): Likewise.
10057 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10058 (vectorizable_bswap): Adjust.
10059 (vectorizable_call): Likewise.
10060 (vectorizable_conversion): Likewise.
10061 (vectorizable_assignment): Likewise.
10062 (vectorizable_shift): Likewise.
10063 (vectorizable_operation): Likewise.
10064 (vectorizable_store): Likewise.
10065 (vectorizable_condition): Likewise.
10066 (vectorizable_comparison): Likewise.
10068 2017-06-28 Michael Collison <michael.collison@arm.com>
10071 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10073 (arm_gen_movmemqi): Removed unused variable 'i'.
10074 Convert 'for' loop into 'while' loop.
10075 (arm_expand_prologue): Remove last unnecessary set of insn.
10076 (thumb_pop): Remove unused variable 'pushed_words'.
10077 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10079 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10081 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10082 * config/s390/s390-protos.h: Add prototype of
10083 s390_rel_address_ok_p.
10084 * config/s390/s390.c (s390_got_symbol): New function.
10085 (s390_rel_address_ok_p): New function.
10086 (legitimize_pic_address): Use s390_rel_address_ok_p.
10087 (s390_load_got): Use s390_got_symbol.
10088 (s390_option_override): Issue error if
10089 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10090 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10092 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10094 2017-06-27 Andrew Pinski <apinski@cavium.com>
10096 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10097 (X * copysign (1.0, X)): New pattern.
10098 (X * copysign (1.0, -X)): New pattern.
10099 (copysign (-1.0, CST)): New pattern.
10101 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10103 * genmultilib (combination_space): Remove variable.
10104 Validate reuse rules against regular expression for any sequence
10105 of multilib options in any order.
10107 2017-06-27 Michael Collison <michael.collison@arm.com>
10109 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10110 call aarch64_split_simd_combine.
10111 * (aarch64_combine_internal<mode>): Delete pattern.
10112 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10113 Allow register and subreg operands.
10115 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10117 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10118 specific need, just fallback on defaults.
10119 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10121 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10122 Olivier Hainque <hainque@adacore.com>
10124 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10126 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10127 targets. Pick a default if no particular attempt applied.
10128 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10131 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10133 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10134 (x86_64-wrs-vxworks7): Likewise.
10136 2017-06-27 Marek Polacek <polacek@redhat.com>
10139 * ubsan.c (instrument_null): Check get_base_address's result for null.
10141 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10143 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10145 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10147 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10148 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10149 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10150 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10151 New function types.
10152 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10153 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10154 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10155 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10156 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10157 BUILT_IN_FEUPDATEENV): New builtins.
10158 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10159 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10160 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10161 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10163 (builtin_structptr_types): Adjust size.
10164 * tree.c (builtin_structptr_types): Add four entries.
10166 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10167 Olivier Hainque <hainque@adacore.com>
10169 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10170 (TLS_SYM): New local macro, forcing reference to __tls__ on
10171 link command lines for VxWorks 7 RTPs, triggering initialization
10173 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10174 OS features TLS support, true for RTPs on VxWorks 7.
10175 * config/vxworks.c (vxworks_override_options): Setup emutls
10178 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10180 * predict.c (test_prediction_value_range): Use -1U instead of -1
10181 to avoid narrowing conversion warning.
10182 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10183 to avoid narrowing conversion warning.
10184 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10186 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10188 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10190 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10191 64bit configurations.
10192 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10193 (SIZE_TYPE): Likewise.
10194 * config/vxworks.c (vxworks_emutls_var_fields): Use
10195 long_unsigned_type_node instead of unsigned_type_node as the offset
10196 field type, which is "pointer" mode in emutls.c.
10198 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10201 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10203 PR middle-end/81207
10204 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10205 gimple_vuse copying separately from gimple_vdef copying.
10207 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10209 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10212 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10213 Olivier Hainque <hainque@adacore.com>
10215 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10216 for all vxworks7 targets.
10217 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10218 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10219 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10220 variations for VX6/VX7 and 32/64bits later on in ...
10221 (VXWORKS_LIB_SPEC): Leverage new macros.
10222 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10223 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10225 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10227 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10228 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10230 2017-06-26 Carl Love <cel@us.ibm.com>
10232 * config/rs6000/rs6000-c.c: Add support for built-in functions
10233 vector bool char vec_reve (vector bool char);
10234 vector signed char vec_reve (vector signed char);
10235 vector unsigned char vec_reve (vector unsigned char);
10236 vector bool int vec_reve (vector bool int);
10237 vector signed int vec_reve (vector signed int);
10238 vector unsigned int vec_reve (vector unsigned int);
10239 vector bool long long vec_reve (vector bool long long);
10240 vector signed long long vec_reve (vector signed long long);
10241 vector unsigned long long vec_reve (vector unsigned long long);
10242 vector bool short vec_reve (vector bool short);
10243 vector signed short vec_reve (vector signed short);
10244 vector double vec_reve (vector double);
10245 vector float vec_reve (vector float);
10246 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10247 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10248 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10249 (altivec_vreve): New pattern.
10250 * config/rs6000/altivec.h (vec_reve): New define.
10251 * doc/extend.texi (vec_rev): Update the built-in documentation file
10252 for the new built-in functions.
10254 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10256 PR tree-optimization/71815
10257 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10259 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10261 (slsr_process_phi): Likewise.
10262 (replace_uncond_cands_and_profitable_phis): Don't replace a
10263 multiply candidate with a stride of 1 (copy or cast).
10264 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10266 (lowest_cost_path): Likewise.
10267 (total_savings): Likewise.
10269 2017-06-26 Richard Biener <rguenther@suse.de>
10272 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10273 Use def_builtin_pure for all gather builtins.
10275 2017-06-26 Richard Biener <rguenther@suse.de>
10277 PR tree-optimization/81203
10278 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10279 non-dominating BBs.
10281 2017-06-26 Marek Polacek <polacek@redhat.com>
10284 * doc/invoke.texi: Document -Wmultistatement-macros.
10286 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10288 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10289 arm_neon_ok_no_float_abi effective target.
10291 2017-06-26 Richard Biener <rguenther@suse.de>
10293 PR tree-optimization/80928
10294 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10295 (copy_bbs): Set BB_DUPLICATED flag early.
10296 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10298 (execute_on_shrinking_pred): Likewise.
10299 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10300 BB_DUPLICATED blocks.
10301 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10302 iterate over all PHIs considering removal of *gsi.
10304 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10306 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10309 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10311 * config/rs6000/rs6000-string.c: (expand_block_clear,
10312 do_load_for_compare, select_block_compare_mode,
10313 compute_current_alignment, expand_block_compare,
10314 expand_strncmp_align_check, expand_strn_compare,
10315 expand_block_move, rs6000_output_load_multiple)
10316 Move functions related to string/block move/compare
10317 to a separate file.
10318 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10319 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10320 for this function which is now used in two files.
10321 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10322 * config.gcc: Add rs6000-string.o to extra_objs for
10323 targets powerpc*-*-* and rs6000*-*-*.
10325 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10328 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10329 32-bit, since indexed is not valid for DImode.
10330 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10331 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10332 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10333 (define_peephole2 for Altivec d-form store): Likewise.
10336 * multiple_target.c (create_dispatcher_calls): Only create the
10337 dispatcher call if the function is the default clone of a
10338 versioned function.
10340 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10342 PR middle-end/80902
10343 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10344 a call, force the call to not be a tail call.
10346 2017-06-23 Jeff Law <law@redhat.com>
10348 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10351 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10353 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10354 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10355 handling for early expansion of vector shifts (sl,sr,sra,rl).
10356 (builtin_function_type): Add vector shift right instructions
10357 to the unsigned argument list.
10359 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10361 rtl-optimizatoin/79286
10362 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10363 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10364 trap. PIC register plus a const unspec without offset can never trap.
10366 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10368 * tree.h (builtin_structptr_type): New type.
10369 (builtin_structptr_types): Declare new array.
10370 * tree.c (builtin_structptr_types): New array.
10371 (free_lang_data, build_common_tree_nodes): Use it.
10373 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10376 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10379 2017-06-22 Matt Turner <mattst88@gmail.com>
10381 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10382 Lake models to skylake case. Assume skylake for unknown
10383 models with clflushopt.
10385 2017-06-22 Jeff Law <law@redhat.com>
10387 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10388 frame sizes that do not satisfy aarch64_uimm12_shift.
10390 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10392 * profile-count.h (apply_probability,
10393 apply_scale, probability_in): Fix checks for zero.
10395 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10397 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10398 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10400 2016-06-22 Richard Biener <rguenther@suse.de>
10402 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10403 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10404 REDUC_MAX_EXPR support.
10405 (vectorizable_reduction): Likewise.
10406 (vect_create_epilog_for_reduction): Likewise.
10408 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10410 * match.pd (A / (1 << B) -> A >> B): New.
10411 * generic-match-head.c: Include optabs-tree.h.
10412 * gimple-match-head.c: Likewise.
10413 * optabs-tree.h (target_supports_op_p): New.
10414 * optabs-tree.c (target_supports_op_p): New.
10416 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10418 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10419 $gcc_cv_ld --help output.
10420 (gcc_cv_ld_demangle): Likewise.
10421 (gcc_cv_ld_eh_frame_hdr): Likewise.
10422 (gcc_cv_ld_pie): Likewise.
10423 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10424 (gcc_cv_ld_buildid): Likewise.
10425 (gcc_cv_ld_sysroot): Likewise.
10426 (ld_bndplt_support): Likewise.
10427 (ld_pushpopstate_support): Likewise.
10428 * configure: Regenerate.
10429 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10431 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10434 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10435 operands indexes to avoid gap between operands and match_dups.
10437 2017-06-21 Andrew Pinski <apinski@cavium.com>
10439 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10440 Increment Arith_shift and Arith_shift_reg by 1.
10441 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10443 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10444 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10445 (aarch64_strip_extend): Add new argument and test for it.
10446 (aarch64_cheap_mult_shift_p): New function.
10447 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10448 add a cost if it is true.
10449 Update calls to aarch64_strip_extend.
10450 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10452 2017-06-21 Andrew Pinski <apinski@cavium.com>
10454 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10456 (thunderxt88): Likewise.
10457 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10458 (thunderx_prefetch_tune): New variable.
10459 (thunderx2t99_prefetch_tune): Update for the correct values.
10460 (thunderxt88_tunings): New variable.
10461 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10462 generic_prefetch_tune.
10463 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10465 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10467 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10468 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10469 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10470 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10471 (aarch64_atomic_cas<mode>, GPI): Likewise.
10473 2017-06-21 Martin Liska <mliska@suse.cz>
10475 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10476 statements on cold and hot labels.
10477 * predict.c (tree_estimate_probability_bb): Remove the
10478 prediction from this place.
10480 2017-06-21 Martin Liska <mliska@suse.cz>
10482 PR tree-optimization/79489
10483 * gimplify.c (maybe_add_early_return_predict_stmt): New
10485 (gimplify_return_expr): Call the function.
10486 * predict.c (tree_estimate_probability_bb): Remove handling
10488 * predict.def: Update comment about early return predictor.
10489 * gimple-predict.h (is_gimple_predict): New function.
10490 * predict.def: Change default value of early return to 66.
10491 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10493 * passes.def: Put pass_strip_predict_hints to the beginning of
10496 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10498 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10499 FUNCTION_DECL declarations.
10500 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10502 (dwaf2out_decl): Likewise.
10503 * godump.c (go_early_global_decl): Skip call to the real debug hook
10504 for FUNCTION_DECL declarations.
10505 * passes.c (rest_of_decl_compilation): Skip call to the
10506 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10507 -fdump-go-spec is passed.
10509 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10511 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10512 Reorder for compactness.
10513 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10514 (def_builtin_pure, def_builtin_pure2): New functions.
10515 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10517 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10519 * match.pd (nop_convert): New predicate.
10520 ((A +- CST1) +- CST2): Allow some NOP conversions.
10522 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10525 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10526 with ctors/dtors if GOVD_SHARED is set.
10528 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10530 * config/aarch64/aarch64.md (movti_aarch64):
10531 Emit mov rather than orr.
10532 (movtf_aarch64): Likewise.
10533 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10534 Emit mov rather than orr.
10536 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10538 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10539 Swap alternatives, make integer dup more expensive.
10541 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10543 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10544 Return true for non-tls symbols.
10546 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10548 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10549 (cortex-a75): Likewise.
10550 (cortex-a75.cortex-a55): Likewise.
10551 * config/aarch64/aarch64-tune.md: Regenerate.
10552 * doc/invoke.texi (-mtune): Document new values for -mtune.
10554 2017-06-21 Tom de Vries <tom@codesourcery.com>
10556 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10557 stack_size feature.
10558 (Effective-Target Keywords, Other attributes): Suggest using
10559 dg-add-options stack_size feature to get stack limit in stack_size
10560 effective target documentation.
10562 2017-06-21 Julian Brown <julian@codesourcery.com>
10563 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10565 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10566 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10567 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10569 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10570 attribute type list for neon_multiply.
10571 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10572 attribute type list for neon_multiply.
10573 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10574 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10575 attribute type list for neon_multiply.
10576 * config/arm/types.md (crypto_pmull): Add.
10577 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10578 attribute type list.
10580 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10582 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10585 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10587 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10588 to make sure not to dereference a NULL cost_classes_ptr pointer.
10590 2017-06-20 Carl Love <cel@us.ibm.com>
10592 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10593 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10594 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10595 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10596 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10597 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10598 VMULOSW): New enum "unspec" values.
10599 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10600 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10601 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10602 altivec_vmulosw): New patterns.
10603 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10604 VMULOSW): Add definitions.
10606 2017-06-20 Julia Koval <julia.koval@intel.com>
10608 * config/i386/i386.c: Fix rounding expand for new pattern.
10609 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10611 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10613 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10614 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10616 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10618 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10621 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10623 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10624 architecture, then by implementer ID.
10625 * config/aarch64/aarch64-tune.md: Regenerate.
10627 2017-06-20 Richard Biener <rguenther@suse.de>
10629 PR middle-end/81097
10630 * fold-const.c (split_tree): Fold to type before negating.
10632 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10634 * diagnostic-show-locus.c
10635 (selftest::test_fixit_deletion_affecting_newline): New function.
10636 (selftest::diagnostic_show_locus_c_tests): Call it.
10638 2017-06-20 Andreas Schwab <schwab@suse.de>
10641 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10644 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10646 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10647 __ARM_FEATURE_COPROC according to support.
10649 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10651 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10652 Rewritten to avoid overflow for > 32-bit pointers.
10655 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10656 by removing enum keyword.
10657 (ubsan_type_descriptor): Likewise. Formatting fix.
10660 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10661 splitter): Require TARGET_SSE2 in the condition.
10663 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10666 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10667 for doing vector set of SFmode on ISA 3.0.
10668 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10669 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10671 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10672 SFmode value into a V4SF variable that was extracted from another
10673 V4SF variable without converting the element to double precision
10674 and back to single precision vector format.
10675 (vsx_insert_extract_v4sf_p9_2): Likewise.
10677 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10679 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10680 in UWHI to avoid undefined overflow.
10683 * ubsan.h (enum ubsan_encode_value_phase): New.
10684 (ubsan_encode_value): Change second argument to
10685 enum ubsan_encode_value_phase with default value of
10686 UBSAN_ENCODE_VALUE_GENERIC.
10687 * ubsan.c (ubsan_encode_value): Change second argument to
10688 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10689 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10690 create_tmp_var_raw instead of create_tmp_var and use a
10692 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10693 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10694 ubsan_encode_value callers.
10697 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10698 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10699 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10701 2017-06-19 Richard Biener <rguenther@suse.de>
10703 PR middle-end/81118
10704 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10705 estimates if we changed anything.
10707 2017-06-19 Richard Biener <rguenther@suse.de>
10709 PR tree-optimization/80887
10710 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10711 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10712 simplified lookups, then reset mprts_hook.
10713 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10715 (try_to_simplify): Likewise.
10717 2017-06-19 Martin Liska <mliska@suse.cz>
10720 * gimplify.c (gimplify_switch_expr):
10721 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10723 2017-06-19 Martin Liska <mliska@suse.cz>
10725 * doc/install.texi: Document that PGO runs in 4 stages.
10727 2017-06-19 Martin Liska <mliska@suse.cz>
10730 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10731 to dispatcher function name.
10732 * multiple_target.c (replace_function_decl): New function.
10733 (create_dispatcher_calls): Redirect both edges and references.
10735 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10737 * profile-count.c (profile_count::dump): Dump quality.
10738 (profile_count::differs_from_p): Update for unsigned val.
10739 * profile-count.h (profile_count_quality): New enum.
10740 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10742 2017-06-19 Richard Biener <rguenther@suse.de>
10744 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10745 struct function as arg.
10746 (estimate_numbers_of_iterations): Export overload with loop arg.
10747 (free_numbers_of_iterations_estimates_loop): Use an overload of
10748 free_numbers_of_iterations_estimates instead.
10749 * tree-cfg.c (remove_bb): Adjust.
10750 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10751 * tree-parloops.c (gen_parallel_loop): Likewise.
10752 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10754 (tree_unroll_loops_completely): Likewise.
10755 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10756 Use an overload instead and export.
10757 (estimated_loop_iterations): Adjust.
10758 (max_loop_iterations): Likewise.
10759 (likely_max_loop_iterations): Likewise.
10760 (estimate_numbers_of_iterations): Take struct function as arg
10762 (loop_exits_before_overflow): Adjust.
10763 (free_numbers_of_iterations_estimates_loop): Use an overload.
10764 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10765 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10767 2017-06-19 Richard Biener <rguenther@suse.de>
10770 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10771 RANGE_EXPR conservatively.
10773 2017-06-16 Carl Love <cel@us.ibm.com>
10775 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10776 definitions for vec_float, vec_float2, vec_floato,
10777 vec_floate built-ins.
10778 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10779 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10781 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10782 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10783 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10784 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10785 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10786 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10787 vec_floato): Add builtin defines.
10788 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10789 Update the built-in documentation file for the new built-in
10792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10794 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10795 (mthumb): Mark as the negative of -marm.
10797 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10799 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10801 (ARM Options, -mtune): Document that this accepts the same
10802 extension options as -mcpu.
10803 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10805 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10807 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10808 permitted extensions.
10810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10812 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10813 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10814 (armv8-m.main): Add option +nodsp.
10815 * config/arm/arm-cpu-cdata.h: Regenerated.
10817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10819 * config/arm/t-fuchsia: New file.
10820 * config.gcc (arm*-*-fuchsia*): Use it.
10822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10824 * config/arm/t-symbian: Rewrite for new option infrastructure.
10826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10828 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10829 (MULTILIB_REQUIRED): Likewise.
10831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10833 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10834 (MULTILIB_RESUE): Likewise.
10835 (MULTILIB_MATCHES): Likewise.
10836 (MULTLIB_REQUIRED): Likewise.
10838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10840 * config/arm/t-rtems: Rewrite for new option framework.
10842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10844 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10845 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10846 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10847 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10848 * config/arm/t-multilib: ... here.
10849 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10850 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10851 armv7-a and armv8*-a when A-profile libraries have not been built.
10852 * config/arm/t-rmprofile: Rewrite.
10854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10856 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10857 with a backslash. Remove the backslash after substituting unescaped
10859 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10863 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10864 the driver to pass through the default setting of -mfloat-abi.
10865 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
10867 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
10868 (all_feat_combs): New rule.
10869 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
10871 * config/arm/t-aprofile: Rewrite.
10873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10875 * config/arm/arm.h (FPUTYPE_AUTO): Define.
10876 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
10877 fpu is not specified by the user/command-line.
10878 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
10879 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
10880 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
10881 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
10882 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
10883 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
10885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10887 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
10888 * config/arm/t-arm-elf: Rewritten.
10890 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10892 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
10893 have some floating-point instructions.
10894 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
10895 (TARGET_MAYBE_HARD_FLOAT): New macro.
10896 * config/arm/arm-builtins.c (arm_init_builtins): Use
10897 TARGET_MAYBE_HARD_FLOAT.
10898 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10902 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10903 (configargs.h): Include it.
10904 (arm_print_hint_for_fpu_option): New function.
10905 (arm_parse_fpu_option): New function.
10906 (candidate_extension): New class.
10907 (arm_canon_for_multilib): New function.
10908 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10909 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10910 (ARCH_CANONICAL_SPECS): New macro.
10911 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10915 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10916 are set after handling multilib fragments. Set target_cpu_default2
10919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10921 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10923 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10924 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10926 (check_arch): Likewise.
10927 * config/arm/arm.c (arm_configure_build_target): Handle
10928 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10929 options in the default.
10931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10933 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10934 when an option is an alias of another.
10935 * config/arm/parsecpu.awk (optalias): New parser token.
10936 (gen_comm_data): Mark non-alias options as such. Emit entries
10937 for extension aliases.
10938 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10939 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10940 (armv6kz, armv6zk, armv6t2): Likewise.
10941 (armv7): Make vfpv3-d16 an alias.
10942 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10944 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10945 Sort in canonical order.
10946 (armv8-a): Sort in canonical order.
10947 (armv8.1-a, armv8.2-a): Likewise.
10948 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10950 (cortex-a9): Sort in canonical order.
10951 * config/arm/arm.c (selftests.h): Include it.
10952 (arm_test_cpu_arch_data): New function.
10953 (arm_run_self_tests): New function.
10954 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10955 (targetm): Move declaration to the end of the file.
10956 * arm-cpu-cdata.h: Regenerated.
10958 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10960 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10961 call to target_mode_check describing the type of option passed.
10962 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10963 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10964 arm_parse_cpu_option_name to match parameters against list of
10966 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10967 arm_arch_core_flags data structure.
10968 * config/arm/arm-cpu_cdata.h: Regenerated.
10970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10972 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10974 (arm_print_hint_for_cpu_option): Likewise.
10975 (arm_print_hint_for_arch_option): Likewise.
10976 (arm_parse_cpu_option_name): Likewise.
10977 (arm_parse_arch_option_name): Likewise.
10978 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10979 of entries in the all_fpus list.
10980 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10981 (arm_parse_cpu_option_name): Declare.
10982 (arm_parse_arch_option_name): Declare.
10983 (arm_parse_option_features): Declare.
10984 (arm_intialize_isa): Declare.
10985 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10987 (gen_comm_data): ... here. Make definitions non-static.
10988 * config/arm/arm-cpu-data.h: Regenerated.
10989 * config/arm/arm-cpu-cdata.h: Regenerated.
10991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10993 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10994 (cpu_arch_extension): New structure.
10995 (cpu_arch_option, arch_option, cpu_option): New structures.
10996 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10997 architecture types.
10998 (gen_data): Generate new format data tables.
10999 * config/arm/arm.c (cpu_tune): New structure.
11000 (cpu_option, processors): Delete.
11001 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11002 (arm_print_hint_for_cpu_option): ... this and ...
11003 (arm_print_hint_for_arch_option): ... this.
11004 (arm_parse_arch_cpu_name): Delete. Replace with ...
11005 (arm_parse_cpu_option_name): ... this and ...
11006 (arm_parse_arch_option_name): ... this.
11007 (arm_unrecognized_feature): Change type of target parameter to
11009 (arm_parse_arch_cpu_features): Delete. Replace with ...
11010 (arm_parse_option_features): ... this.
11011 (arm_configure_build_target): Rework to use new configuration data
11013 (arm_print_tune_info): Rework for new configuration data tables.
11014 * config/arm/arm-cpu-data.h: Regenerated.
11015 * config/arm/arm-cpu.h: Regenerated.
11017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11019 * Makefile.in (OBJS): Move sbitmap.o from here ...
11020 (OBJS-libcommon): ... to here.
11022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11024 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11025 (ISA_ALL_CRYPTO): New macro.
11026 (ISA_ALL_SIMD): New macro
11027 (ISA_ALL_FP): New macro.
11028 * config/arm/arm.c (fpu_bitlist): Update initializer.
11029 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11031 (arm9e): Add fpu. Add option for nofp
11032 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11033 (arm926ej-s, arm1026ej-s): Likewise.
11034 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11035 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11036 neon-fp16, neon-vfpv4, nofp and nosimd.
11037 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11038 (cortex-a8): Add fpu. Add option for nofp.
11039 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11040 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11041 (cortex-r4f): Add fpu.
11042 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11043 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11045 (cortex-r8): Likewise.
11046 (cortex-m4): Add fpu. Add option for nofp.
11047 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11048 (cortex-a17.cortex-a7): Likewise.
11049 (cortex-a32): Add fpu. Add options for crypto and nofp.
11050 (cortex-a35, cortex-a53): Likewise.
11051 (cortex-a57): Add fpu. Add option for crypto.
11052 (cortex-a72, cortex-a73): Likewise.
11053 (exynos-m1): Likewise.
11054 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11055 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11056 (cortex-m33): Add fpu. Add option for nofp.
11057 * config/arm/arm-cpu-cdata.h: Regenerated
11058 * config/arm/arm-cpu-data.h: Regenerated.
11060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11062 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11063 (armv5te, armv5tej): Likewise.
11064 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11065 (armv7): Add options fp and vfpv3-d16.
11066 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11067 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11069 (armv7ve): Likewise.
11070 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11071 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11072 (armv8-a): Add nocrypto option.
11073 (armv8.1-a, armv8.2-a): Likewise.
11074 (armv8-m.main): add options fp, fp.dp and nofp.
11076 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11078 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11080 (armv8-a+crc): Delete.
11081 (armv8.1-a): Add options simd, crypto and nofp.
11082 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11083 (armv8.2-a+fp16): Delete.
11084 (armv8-m.main): Add option dsp.
11085 (armv8-m.main+dsp): Delete.
11086 (cortex-a8): Add fpu. Add nofp option.
11087 (cortex-a9): Add fpu. Add nofp and nosimd options.
11088 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11089 link to main cpu and architecture data structures.
11090 (gen_comm_data): Only put isa attributes from the main architecture
11092 (option): New statement for architecture and CPU entries.
11093 * arm.c (struct cpu_option): New structure.
11094 (struct processors): Add entry for options.
11095 (arm_unrecognized_feature): New function.
11096 (arm_parse_arch_cpu_name): Ignore any characters after the first
11098 (arm_parse_arch_cpu_feature): New function.
11099 (arm_configure_build_target): Separate out any CPU and architecture
11100 features and parse separately. Don't error out if -mfpu=auto is
11101 used with only an architecture string.
11102 (arm_print_asm_arch_directives): New function.
11103 (arm_file_start): Call it.
11104 * config/arm/arm-cpu-cdata.h: Regenerated.
11105 * config/arm/arm-cpu-data.h: Likewise.
11106 * config/arm/arm-tables.opt: Likewise.
11108 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11110 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11111 assembler when it is not -mfpu=auto.
11113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11115 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11116 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11117 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11118 (ASM_CPU_SPEC): Rewrite.
11119 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11120 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11121 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11122 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11123 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11124 copied string is NUL-terminated. Also strip any characters prefixed
11126 (arm_rewrite_selected_arch): New function.
11127 (arm_rewrite_march): New function.
11129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11131 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11132 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11133 (march, mcpu, mtune): Convert to string-based options.
11134 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11135 (arm_parse_arch_cpu_name): New function.
11136 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11137 identify selected architecture or CPU.
11138 (arm_option_save): New function.
11139 (TARGET_OPTION_SAVE): Redefine.
11140 (arm_option_restore): Restore string options.
11141 (arm_option_print): Print string options.
11143 2017-06-16 Martin Sebor <msebor@redhat.com>
11145 PR tree-optimization/80933
11146 PR tree-optimization/80934
11147 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11148 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11149 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11150 (gimple_fold_builtin): Call them.
11152 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11154 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11155 as unlikely; update profile.
11157 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11159 * predict.c (force_edge_cold): Handle declaring edges impossible
11162 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11164 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11166 (try_unroll_loop_completely): Fix reporting.
11168 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11170 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11172 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11175 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11176 if given a non-constant argument for an intrinsic which requires a
11179 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11181 * profile.c (compare_freqs): New function.
11182 (branch_prob): Sort edge list.
11183 (find_spanning_tree): Assume that the list is priority sorted.
11185 2017-06-16 Richard Biener <rguenther@suse.de>
11187 PR tree-optimization/81090
11188 * passes.def (pass_record_bounds): Remove.
11189 * tree-pass.h (make_pass_record_bounds): Likewise.
11190 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11191 make_pass_record_bounds): Likewise.
11192 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11193 not free niter estimates at the beginning but at the end.
11194 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11196 2017-06-16 Richard Biener <rguenther@suse.de>
11198 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11199 initializer to workaround ICE in host GCC 4.8.
11201 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11203 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11205 (clone_inlined_nodes): Update.
11207 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11209 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11210 prefetch settings, and enable prefetching by default at -O3.
11212 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11214 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11215 Set flag_prefetch_loop_arrays according to tuning data.
11217 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11219 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11220 New tune structure.
11221 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11222 [Unrelated to main purpose of the patch] Place the pointer field last
11223 to enable type checking errors when tune structure are wrongly merged.
11224 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11225 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11226 (thunderx2t99_prefetch_tune): New tune constants.
11227 (tune_params *_tunings): Update all tunings (no functional change).
11228 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11229 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11230 from tunings structures.
11232 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11235 * ubsan.c (instrument_null): Add T argument, use it instead
11236 of computing it based on IS_LHS.
11237 (instrument_object_size): Likewise.
11238 (pass_ubsan::execute): Adjust instrument_null and
11239 instrument_object_size callers to pass gimple_get_lhs or
11240 gimple_assign_rhs1 result to it. Use instrument_null instead of
11241 calling get_base_address and instrument_mem_ref. Handle
11242 aggregate call arguments for object-size sanitization.
11244 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11246 PR tree-optimization/81089
11247 * tree-vrp.c (is_masked_range_test): Validate operands of
11250 2017-06-15 Martin Sebor <msebor@redhat.com>
11253 * dumpfile.c (dump_register): Avoid calling memset to initialize
11254 a class with a default ctor.
11255 * gcc.c (struct compiler): Remove const qualification.
11256 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11257 * hash-table.h: Ditto.
11258 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11260 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11261 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11263 * params.h (struct param_info): Make struct members non-const.
11264 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11265 with default initialization.
11266 * vec.h (vec_copy_construct, vec_default_construct): New helper
11268 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11269 with vec_copy_construct.
11270 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11271 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11272 * doc/invoke.texi (-Wclass-memaccess): Document.
11274 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11276 * emit-rtl.h (is_leaf): Update comment about local
11277 register allocator.
11279 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11282 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11283 for a variable to have a section before checking if the section has a
11285 Set section to.persistent if persistent attribute is set.
11286 Warn if .persistent attribute is used on an automatic variable.
11288 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11290 PR rtl-optimization/80474
11291 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11293 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11295 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11298 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11300 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11301 (casesi): Emit bounds checking as RTL.
11302 (casesi_internal_mips16_<mode>): Remove bounds checking.
11304 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11306 * config/xtensa/xtensa.c (xtensa_option_override): Append
11307 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11308 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11309 xtensa_doloop_hooks): Define unconditionally.
11310 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11312 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11313 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11314 for it in xtensa_option_override.
11315 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11316 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11318 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11320 * doc/cppopts.texi: Document '-' special value to -MF.
11322 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11324 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11325 (cortex_a53_fconst): Likewise.
11326 (cortex_a53_fpmul): Likewise.
11327 (cortex_a53_f_load_64): Likewise.
11328 (cortex_a53_f_load_many): Likewise.
11329 (cortex_a53_advsimd_alu): Likewise.
11330 (cortex_a53_advsimd_alu_q): Likewise.
11331 (cortex_a53_advsimd_mul): Likewise.
11332 (cortex_a53_advsimd_mul_q): Likewise.
11333 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11334 Add missing fmul, r2f_cvt and fconst cases.
11336 2017-06-14 Richard Biener <rguenther@suse.de>
11338 PR middle-end/81088
11339 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11341 (fold_binary_loc): When associating do not treat pre-existing
11342 TREE_OVERFLOW on literal constants as a reason to allow
11343 TREE_OVERFLOW on associated literal constants.
11345 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11347 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11348 (MASK_FEATURES): New macro.
11349 * config/sparc/sparc.c (sparc_option_override): Remove the special
11350 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11352 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11354 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11355 a division of 0 if non-call exceptions are enabled.
11357 2017-06-14 Andrew Pinski <apinski@cavium.com>
11358 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11361 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11362 Improve vector initialization code gen for only variable case.
11364 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11366 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11368 2017-06-14 Richard Biener <rguenther@suse.de>
11370 PR tree-optimization/81083
11371 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11374 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11376 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11377 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11378 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11379 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11380 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11381 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11383 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11385 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11386 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11388 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11390 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11394 * config/rs6000/t-rtems: Don't handle SPE.
11396 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11398 * config/rs6000/t-linux: Don't handle SPE.
11400 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11402 * config/rs6000/eabispe.h: Delete file.
11404 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11406 * config/rs6000/t-spe: Delete file.
11408 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11410 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11411 (rs6000_legitimate_offset_address_p): Return false for anything in
11412 V2SImode or V2SFmode.
11414 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11416 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11417 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11418 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11420 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11421 (rs6000_legitimate_offset_address_p): Ditto.
11422 (rs6000_emit_move): Ditto.
11423 (rs6000_init_builtins): Remove V4HI_type_node.
11425 2017-06-13 Martin Liska <mliska@suse.cz>
11428 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11429 (gate_asan): Likewise.
11430 * asan.h (asan_no_sanitize_address_p): Remove the function.
11431 (sanitize_flags_p): New function.
11432 * builtins.def: Fix coding style.
11433 * common.opt: Use renamed enum value.
11434 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11435 * doc/extend.texi: Document no_sanitize attribute.
11436 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11437 to SANITIZE_UNDEFINED_NONDEFAULT.
11438 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11439 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11440 Use sanitize_flags_p.
11441 * gimplify.c (gimplify_function_tree): Likewise.
11442 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11443 * opts.c (parse_no_sanitize_attribute): New function.
11444 (common_handle_option): Use renamed enum value.
11445 * opts.h (parse_no_sanitize_attribute): Declare.
11446 * tree.c (sanitize_flags_p): New function.
11447 * tree.h: Declared here.
11448 * tsan.c: Use sanitize_flags_p.
11449 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11450 (instrument_mem_ref): Likewise.
11451 (instrument_bool_enum_load): Likewise.
11452 (do_ubsan_in_current_function): Remove the function.
11453 (pass_ubsan::execute): Use sanitize_flags_p.
11454 * ubsan.h: Remove do_ubsan_in_current_function
11455 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11456 (dump_function_to_file): Use it here.
11458 2017-06-13 Martin Jambor <mjambor@suse.cz>
11460 PR tree-optimization/80803
11461 PR tree-optimization/81063
11462 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11463 (propagate_subaccesses_across_link): Enqueue subtree whenever
11464 necessary instead of relying on the caller.
11466 2017-06-13 Martin Jambor <mjambor@suse.cz>
11468 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11469 that have a first_link.
11470 (sort_and_splice_var_accesses): Do not check first_link before
11472 (subtree_mark_written_and_enqueue): Likewise.
11473 (propagate_all_subaccesses): Likewise and do not stop at first
11474 parent with a first_link.
11476 2017-06-13 Martin Jambor <mjambor@suse.cz>
11478 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11481 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11483 * match.pd: New pattern.
11485 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11487 * tree-vrp.c (is_masked_range_test): New function.
11488 (register_edge_assert_for): Determine ranges for
11491 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11493 PR tree-optimization/67328
11494 * fold-const.c (maskable_range_p): New function.
11495 (build_range_check): Generate bittests if possible.
11497 2017-06-13 Martin Liska <mliska@suse.cz>
11499 * gimple-pretty-print.c (dump_probability): Add new argument.
11500 (dump_edge_probability): Dump both probability and count.
11501 (dump_gimple_label): Likewise.
11502 (dump_gimple_bb_header): Likewise.
11504 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11507 * config/avr/avr-devices.c: Fix indentation.
11508 * config/avr/gen-avr-mmcu-specs.c: Dito.
11510 2017-06-13 Richard Biener <rguenther@suse.de>
11512 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11513 instead get vector type from stmt_info.
11514 (vectorizable_reduction): Adjust. Remove dead code.
11516 2017-06-13 Richard Biener <rguenther@suse.de>
11518 PR middle-end/81065
11519 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11520 case of C * (x * C2 + C3).
11521 (fold_addr_of_array_ref_difference): Properly fold index difference.
11523 2017-06-12 David S. Miller <davem@davemloft.net>
11526 * config/sparc/sparc.md (return expander): Emit frame blockage if
11527 function uses alloca.
11529 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11531 * combine.c (make_field_assignment): Check len rather than the mode
11532 precision when calling force_to_mode.
11534 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11536 Support multilibs and devices that see flash in RAM address range.
11539 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11540 (avr_mcu_t) <flash_pm_offset>: New field.
11541 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11542 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11543 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11544 (AVR_TINY_PM_OFFSET): Remove macro.
11545 * config/avr/avr.opt (-mshort-calls): New option.
11546 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11547 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11548 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11549 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11550 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11551 instead of avr_arch->have_jmp_call.
11552 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11553 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11554 avr_arch->flash_pm_offset to define.
11555 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11556 new field flash_pm_offset. Add entry for avrxmega3.
11557 (avr_texinfo): Add entry for avrxmega3.
11558 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11559 attiny212, attiny214,
11560 attiny412, attiny414, attiny416, attiny417,
11561 attiny814, attiny816, attiny817,
11562 attiny1614, attiny1616, attiny1617,
11563 attiny3214, attiny3216, attiny3217.
11564 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11565 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11566 (avr_print_operand_address) [AVR_TINY]: Same.
11567 (avr_asm_init_sections) <readonly_data_section>: Only patch
11568 callback if avr_arch->flash_pm_offset = 0.
11569 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11570 for rodata if avr_arch->flash_pm_offset != 0.
11571 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11572 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11573 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11574 (m_options): Append opt_rcall.
11575 (m_dirnames): Append dir_rcall.
11576 * config/avr/t-multilib: Regenerate.
11578 * configure.ac [target=avr]: Check whether avrxmega3 default
11579 linker description file works as needed.
11580 * configure: Regenerate.
11581 * doc/avr-mmcu.texi: Regenerate.
11582 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11583 <__AVR_ARCH__>: Document avrxmega3 and 103.
11584 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11585 <__AVR_SHORT_CALLS__>: Document it.
11586 <__AVR_PM_BASE_ADDRESS__>: Document it.
11587 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11588 (AVR Variable Attributes) <progmem>: Document this is
11589 not needed for avrxmega3.
11590 (AVR Named Address Spaces) <__flash>: Dito.
11592 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11594 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11596 2017-06-12 Doug Rupp <rupp@adacore.com>
11598 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11599 Append vxworks-stdint.h to the tm_file list.
11600 * config/vxworks-stdint.h: New file.
11602 2017-06-12 Martin Liska <mliska@suse.cz>
11604 PR tree-optimization/81041
11605 * tree-profile.c (gimple_gen_ic_func_profiler):
11606 Create an extra BB in profile-generate
11607 (gimple_gen_time_profiler): Likewise.
11609 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11611 PR tree-optimization/81003
11612 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11613 (update_range_test): Use it instead of force_gimple_operand_gsi.
11615 2017-06-12 Richard Biener <rguenther@suse.de>
11617 PR tree-optimization/81053
11618 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11619 with backedge value not defined in loop. Simplify def stmt
11622 2017-06-11 Tom de Vries <tom@codesourcery.com>
11625 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11627 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11628 nvptx_cannot_force_const_mem.
11630 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11632 * opts.c (finish_options): Move test for flag_split_stack after
11633 it has been initialized.
11635 2017-06-11 Jason Merrill <jason@redhat.com>
11637 * tree.h (id_equal): New.
11638 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11639 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11640 instead of strcmp of IDENTIFIER_POINTER.
11642 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11644 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11645 (mark_all_inlined_calls_cdtor): Fix formating.
11646 (inline_transform): Rescale profile before inlining.
11648 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11650 * cgraph.h (cgraph_edge::clone): Update prototype.
11651 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11652 (cgraph_node::create_clone): Update.
11653 (cgraph_node::create_version_clone): Update.
11654 * tree-inline.c (copy_bb): Update.
11655 (expand_call_inline): Update.
11657 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11659 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11660 factored out from ...
11661 (rs6000_emit_prologue): ... here.
11663 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11665 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11666 factored out from ...
11667 (rs6000_emit_prologue): ... here.
11669 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11671 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11673 (handle_missing_profiles): Fix computation of tp_first_run.
11674 (counts_to_freqs): Do not touch freqs when count is 0.
11676 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11678 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11681 2017-06-10 Tom de Vries <tom@codesourcery.com>
11683 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11684 attributes): Document signal effective target.
11686 2017-06-10 Tom de Vries <tom@codesourcery.com>
11688 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11689 Document effective target stack_size.
11691 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11693 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11694 to the edit_context if they can be auto-applied.
11696 2017-06-9 Ian Lance Taylor <iant@golang.org>
11698 * opts.c (finish_options): If -fsplit-stack, disable implicit
11699 -forder-blocks-and-partition.
11700 * doc/invoke.texi (Optimize Options): Document that when using
11701 -fsplit-stack -forder-blocks-and-partition is not implicitly
11704 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11706 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11707 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11708 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11709 * builtins.def (abort, trap, unreachable): Declare cold.
11710 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11711 * tree-core.h (ECF_COLD): New.
11712 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11713 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11715 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11717 * predict.c (unlikely_executed_stmt_p): Cleanup.
11719 2017-06-09 Richard Biener <rguenther@suse.de>
11721 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11722 model if the ref is always written to.
11724 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11726 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11728 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11730 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11733 2017-06-09 Tom de Vries <tom@codesourcery.com>
11736 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11737 "target cannot support label values" when encountering LABEL_REF.
11739 2017-06-09 Martin Liska <mliska@suse.cz>
11741 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11742 (gimple_gen_ic_func_profiler): Emit direct comparison
11743 of __gcov_indirect_call_callee with NULL.
11744 (gimple_gen_time_profiler): Change probability from
11745 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11747 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11749 * profile.c (edge_gcov_counts): Turn to pointer.
11750 (compute_branch_probabilities, compute_branch_probabilities): Update.
11751 (branch_prob): Do not clear edge_gcov_count.
11752 * profile.h (edge_gcov_counts): Turn to pointer.
11753 (edge_gcov_count): Update.
11755 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11757 * gimple.h (gimple_check_failed): Mark cold.
11759 2017-06-09 Richard Biener <rguenther@suse.de>
11761 PR tree-optimization/66623
11762 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11763 refactor check_reduction into two parts, properly computing
11764 whether we have to check reduction validity for outer loop
11767 2017-06-09 Richard Biener <rguenther@suse.de>
11769 PR tree-optimization/79483
11770 * graphite-scop-detection.c (order): New global.
11771 (get_order): Compute bb to order mapping that satisfies code
11772 generation constraints.
11773 (cmp_pbbs): New helper.
11774 (build_scops): Start domwalk at entry block, sort generated
11777 2017-06-09 Richard Biener <rguenther@suse.de>
11779 PR middle-end/81007
11780 * ipa-polymorphic-call.c
11781 (ipa_polymorphic_call_context::restrict_to_inner_class):
11782 Skip FIELD_DECLs with error_mark_node type.
11783 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11786 2017-06-09 Martin Liska <mliska@suse.cz>
11788 * predict.c (struct branch_predictor): New struct.
11789 (test_prediction_value_range): New test.
11790 (predict_c_tests): New function.
11791 * selftest-run-tests.c (selftest::run_tests): Run the function.
11792 * selftest.h: Declare new tests.
11794 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11797 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11798 gen_add3_insn did not fail.
11799 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11800 r0, construct that number in a temporary reg and add that reg to r0.
11801 If asked to put the result in r0 as well, fail.
11803 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11805 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11806 for early expansion of vec_eqv.
11808 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11810 PR middle-end/81005
11811 * ubsan.c (instrument_null): Avoid pointless code temporary.
11812 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11814 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11818 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11821 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11824 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11826 * predict.c (unlikely_executed_edge_p): Move ahead.
11827 (probably_never_executed_edge_p): Use it.
11829 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11831 PR middle-end/79988
11832 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11833 gimple_call_builtin_p call.
11835 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11837 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11838 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11839 rtl_check_failed_type2, rtl_check_failed_code1,
11840 rtl_check_failed_code2, rtl_check_failed_code_mode,
11841 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11842 rtvec_check_failed_bounds, rtl_check_failed_flag,
11843 _fatal_insn_not_found, _fatal_insn): Likewise.
11844 * tree.h (tree_contains_struct_check_failed,
11845 tree_check_failed, tree_not_check_failed,
11846 tree_class_check_failed, tree_range_check_failed,
11847 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11848 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11849 tree_operand_check_failed, omp_clause_check_failed,
11850 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11853 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11855 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11856 flag_branch_probabilities.
11857 * ipa-inline.c (edge_badness): Likewise.
11858 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11859 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11860 * predict.c (maybe_hot_frequency_p): Likewise.
11861 (probably_never_executed): Likewise.
11862 * sched-ebb.c (schedule_ebbs): Likewise.
11863 * sched-rgn.c (find_single_block_region): Likewise.
11864 * tracer.c (tail_duplicate): Likewise.
11866 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11868 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
11869 longer requires x_flag_profile_use.
11871 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11873 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
11874 instead of flag_reorder_blocks_and_partition.
11875 * dbxout.c (dbxout_function_end): Likewise.
11876 * dwarf2out.c (gen_subprogram_die): Likewise.
11877 * haifa-sched.c (sched_create_recovery_edges): Likewise.
11878 * hw-doloop.c (reorg_loops): Likewise.
11879 * varasm.c (assemble_start_function,
11880 assemble_end_function): Likewise.
11881 (decide_function_section): Do not check for
11882 flag_reorder_blocks_and_partition.
11884 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11886 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11888 (chkp_get_hard_register_fake_addr_expr): Ditto.
11889 (chkp_build_addr_expr): Add check for hard reg case.
11890 (chkp_parse_array_and_component_ref): Ditto.
11891 (chkp_find_bounds_1): Ditto.
11892 (chkp_process_stmt): Don't generate bounds store for
11895 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11897 * predict.c (maybe_hot_bb_p): Do not check profile status.
11898 (maybe_hot_edge_p): Likewise.
11899 (probably_never_executed): Check for zero counts even if profile
11901 (unlikely_executed_edge_p): New function.
11902 (unlikely_executed_stmt_p): New function.
11903 (unlikely_executed_bb_p): New function.
11904 (set_even_probabilities): Use unlikely predicates.
11905 (combine_predictions_for_bb): Likewise.
11906 (predict_paths_for_bb): Likewise.
11907 (predict_paths_leading_to_edge): Likewise.
11908 (determine_unlikely_bbs): New function.
11909 (estimate_bb_frequencies): Use it.
11910 (compute_function_frequency): Use zero counts even if profile is
11912 * profile-count.h: Fix typo.
11914 2017-08-08 Julia Koval <julia.koval@intel.com>
11916 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11917 _mm512_mask_cvtsepi16_storeu_epi8,
11918 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11919 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11920 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11921 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11922 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11923 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11924 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11925 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11926 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11927 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11928 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11929 __builtin_ia32_pmovuswb256mem_mask,
11930 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11931 __builtin_ia32_pmovwb512mem_mask): New builtins.
11933 2017-08-08 Julia Koval <julia.koval@intel.com>
11935 PR target/73350,80862
11936 * config/i386/subst.md (round): Fix round pattern.
11937 * config/i386/i386.c (ix86_erase_embedded_rounding):
11938 Fix erasing rounding for the fixed pattern.
11940 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11942 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11944 2017-06-08 Martin Liska <mliska@suse.cz>
11946 PR gcov-profile/80911
11947 * gcov.c (block_info::block_info): New constructor.
11949 2017-06-07 Carl Love <cel@us.ibm.com>
11951 * config/rs6000/rs6000-c: The return type of the following
11952 built-in functions was implemented as int not long long. Fix sign
11953 of return value for the unsigned version of vec_mulo and vec_mule.
11954 vector unsigned long long vec_bperm (vector unsigned long long,
11955 vector unsigned char)
11956 vector signed long long vec_mule (vector signed int,
11958 vector unsigned long long vec_mule (vector unsigned int,
11959 vector unsigned int)
11960 vector signed long long vec_mulo (vector signed int,
11962 vector unsigned long long vec_mulo (vector unsigned int,
11963 vector unsigned int)
11964 * doc/extend.texi: Fix the documentation for the built-in
11967 2017-06-07 Carl Love <cel@us.ibm.com>
11970 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11973 2017-06-07 Carl Love <cel@us.ibm.com>
11975 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11976 support, Generate doublehv for signed int/float for BE case only.
11978 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11980 * doc/invoke.texi (mcx16): Rewrite.
11982 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11984 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11985 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11986 *mov<mode>_softfloat, and an anonymous splitter): Use
11987 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11989 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11991 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11993 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11994 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11995 (rs6000_debug_reg_global): Adjust.
11996 (rs6000_init_hard_regno_mode_ok): Adjust.
11997 (rs6000_dbx_register_number): Adjust.
11998 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11999 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12000 Remove SPE_ACC and SPEFSCR.
12001 (REG_ALLOC_ORDER): Ditto.
12002 (FRAME_POINTER_REGNUM): Change to 111.
12003 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12004 (REG_CLASS_NAMES): Ditto.
12005 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12006 (REGISTER_NAMES): Ditto.
12007 (ADDITIONAL_REG_NAMES): Ditto.
12008 (rs6000_reg_names): Ditto.
12009 * config/rs6000/rs6000.md: Renumber some register number
12012 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12014 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12016 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12017 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12019 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12020 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12021 Delete the SPE high registers.
12022 (REG_ALLOC_ORDER): Ditto.
12023 (enum reg_class): Remove SPE_HIGH_REGS.
12024 (REG_CLASS_NAMES): Ditto.
12025 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12026 (REGISTER_NAMES): Ditto.
12027 (rs6000_reg_names): Ditto.
12028 * doc/tm.texi.in: Remove SPE as example.
12029 * doc/tm.texi: Regenerate.
12031 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12033 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12034 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12035 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12036 * config/rs6000/rs6000.md (type): Remove "brinc".
12038 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12040 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12041 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12042 * config/rs6000/linuxspe.h: Delete file.
12043 * config/rs6000/rs6000.md: Don't include spe.md.
12044 * config/rs6000/spe.h: Delete file.
12045 * config/rs6000/spe.md: Delete file.
12046 * config/rs6000/t-rs6000: Remove spe.md.
12048 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12050 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12051 (reg_or_none500mem_operand): Delete.
12052 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12053 instead of reg_or_none500mem_operand.
12055 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12057 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12058 handling of SPE flags.
12059 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12061 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12063 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12065 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12066 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12067 paired_divv2sf3): Similar.
12068 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12069 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12070 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12072 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12073 Rename the paired_* instruction patterns.
12074 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12076 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12077 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12078 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12079 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12080 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12081 PAIRED_VECTOR_MODE.
12082 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12083 (spe_func_has_64bit_regs_p): Delete.
12084 (spe_expand_predicate_builtin): Delete.
12085 (spe_expand_evsel_builtin): Delete.
12086 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12087 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12088 (invalid_e500_subreg): Delete.
12089 (rs6000_legitimize_address): Always force_reg op2 as well, for
12090 paired single memory accesses.
12091 (rs6000_member_type_forces_blk): Delete.
12092 (rs6000_spe_function_arg): Delete.
12093 (rs6000_expand_unop_builtin): Delete SPE handling.
12094 (rs6000_expand_binop_builtin): Ditto.
12095 (spe_expand_stv_builtin): Delete.
12096 (bdesc_2arg_spe): Delete.
12097 (spe_expand_builtin): Delete.
12098 (spe_expand_predicate_builtin): Delete.
12099 (spe_expand_evsel_builtin): Delete.
12100 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12101 (spe_init_builtins): Delete.
12102 (spe_func_has_64bit_regs_p): Delete.
12103 (savres_routine_name): Delete "info" parameter. Adjust callers.
12104 (rs6000_emit_stack_reset): Ditto.
12105 (rs6000_dwarf_register_span): Delete.
12106 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12107 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12108 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12110 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12112 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12113 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12114 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12115 mulv2sf3, divv2sf3): Delete expanders.
12117 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12119 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12121 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12123 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12124 * config/rs6000/rs6000.c: Ditto.
12126 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12128 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12129 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12130 comparison_operator.
12132 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12134 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12135 * config/rs6000/rs6000.opt: Ditto.
12136 * config/rs6000/t-rtems: Ditto.
12138 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12140 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12141 TARGET_E500_SINGLE by 0, simplify.
12142 * config/rs6000/rs6000.c: Ditto.
12143 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12144 (spe_build_register_parallel): Delete.
12145 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12146 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12147 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12148 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12149 (E500_CONVERT): Delete.
12150 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12154 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12155 * config/rs6000/dfp.md: Ditto.
12156 (negdd2, *negdd2_fpr): Merge.
12157 (absdd2, *absdd2_fpr): Merge.
12158 (negtd2, *negtd2_fpr): Merge.
12159 (abstd2, *abstd2_fpr): Merge.
12160 * config/rs6000/e500.h: Delete file.
12161 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12162 TARGET_FPRS by 1 and simplify.
12163 * config/rs6000/rs6000-c.c: Ditto.
12164 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12165 TARGET_DF_SPE by 0.
12166 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12168 * config/rs6000/rs6000.md: Ditto.
12169 (floatdidf2, *floatdidf2_fpr): Merge.
12170 (move_from_CR_gt_bit): Delete.
12171 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12172 (E500_CR_IOR_COMPARE): Delete.
12173 (All patterns that require !TARGET_FPRS): Delete.
12174 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12176 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12178 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12180 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12182 * graphds.c (add_edge): Intitialize edge's attached data.
12183 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12184 pointer parameter. Call pointed function on each edge during
12185 graph traversing. Skip traversing the edge when the function
12187 (graphds_dfs, graphds_scc): Ditto.
12188 (for_each_edge): New parameter. Pass the new parameter to callback
12190 * graphds.h (skip_edge_callback): New function pointer type.
12191 (graphds_dfs, graphds_scc): New function pointer parameter.
12192 (graphds_edge_callback, for_each_edge): New parameter.
12194 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12196 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12197 out code checking if runtime alias check is possible to below ...
12198 Call the new function.
12199 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12200 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12202 2017-06-07 Marek Polacek <polacek@redhat.com>
12205 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12206 TYPE_OVERFLOW_WRAPS check.
12208 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12210 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12211 if versioning is required.
12212 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12213 peeling with the check for versioning.
12215 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12217 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12218 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12219 Set true to new parameter if new ssa variable is defined.
12220 (vect_gen_vector_loop_niters): Refactor. Set range information
12221 for the new vector loop bound variable.
12222 (vect_do_peeling): Ditto.
12224 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12226 * tree-affine.c (ssa.h): Include header file.
12227 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12228 has wrapping overflow behavior.
12230 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12232 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12234 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12236 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12237 (tree_to_aff_combination): ... here.
12239 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12241 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12242 reg_pressure model function.
12243 (ivopts_global_cost_for_size): Delete.
12244 (determine_set_costs, iv_ca_recount_cost): Call new model function
12245 ivopts_estimate_reg_pressure.
12247 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12249 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12250 expensive than udiv. Remove floating point cases from mod.
12252 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12254 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12255 Increase idiv cost.
12257 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12259 * config/aarch64/aarch64.md
12260 (copysignsf3): Fix mask generation.
12262 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12264 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12266 (class dump_manager): Add register_dumps method.
12267 * dumpfile.c: Include langhooks.h.
12268 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12269 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12270 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12271 (dump_manager::dump_register): Start with 512 entries instead of 32.
12272 (dump_manager::register_dumps): New method.
12273 * toplev.c (general_init): Instead of invoking register_dumps
12274 langhook, invoke register_dumps method on the dump manager.
12275 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12278 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12280 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12281 Say that # requires an associated define_split to exist, and that
12282 the define_split must be suitable for use after register allocation.
12284 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12286 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12287 (compute_outgoing_frequencies): Also initialize zero counts.
12288 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12289 around loops; preserve more of profile when nothing changes.
12291 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12293 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12295 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12296 * config/arm/arm-cpu-cdata.h: Regenerate.
12297 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12298 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12299 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12300 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12301 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12303 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12305 * config/arm/t-rmprofile: Likewise.
12306 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12308 2017-06-06 David S. Miller <davem@davemloft.net>
12311 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12312 blockage if function uses alloca.
12314 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12316 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12317 New "uid" fields to hold pretty-print IDs of group and ref.
12318 Memory references are now identified as <group_id>:<ref_id>
12319 instead of using [random] addresses.
12320 (dump_mem_details): Simplify, no functional change.
12321 (dump_mem_ref): Simplify and make output more concise.
12322 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12323 (find_or_create_group): Initialize group uid.
12324 (record_ref): Initialize ref uid. Improve debug output.
12325 (prune_group_by_reuse, should_issue_prefetch_p,)
12326 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12327 (mark_nontemporal_store, determine_loop_nest_reuse):
12328 Improve debug output.
12330 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12332 * dbgcnt.def (prefetch): New debug counter.
12333 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12334 (schedule_prefetches): Stop issueing prefetches if debug counter
12337 2017-06-06 Tom de Vries <tom@codesourcery.com>
12339 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12340 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12342 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12344 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12345 Use aarch64_reg_or_zero predicate for operand 4.
12346 (aarch64_compare_and_swap<mode> define_insn_and_split):
12347 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12348 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12350 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12352 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12353 (arm_compute_save_core_reg_mask): This.
12354 (thumb1_compute_save_reg_mask): Rename into ...
12355 (thumb1_compute_save_core_reg_mask): This.
12356 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12357 (arm_compute_frame_layout): Likewise.
12359 2017-06-06 Richard Biener <rguenther@suse.de>
12361 PR tree-optimization/80974
12362 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12363 keep or clear leaders SSA info.
12365 2017-06-06 Tom de Vries <tom@codesourcery.com>
12367 * config/nvptx/nvptx.c (split_mode_p): New function.
12368 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12370 2017-06-06 Tom de Vries <tom@codesourcery.com>
12372 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12374 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12377 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12380 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12382 * shrink-wrap.c (handle_simple_exit): Update profile.
12383 (try_shrink_wrapping): Upate profile.
12385 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12387 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12388 (tree_guess_outgoing_edge_probabilities): New.
12389 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12390 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12392 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12394 * ipa-split.c (split_function): Initialize return bb profile.
12396 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12398 * profile.c (compute_branch_probabilities): Also initialize
12399 EXIT_BLOCK profile.
12401 2017-06-06 Richard Biener <rguenther@suse.de>
12403 PR tree-optimization/80928
12404 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12405 (vect_analyze_loop_operations): Properly guard analysis for
12407 (vect_transform_loop): Likewise.
12408 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12409 (vect_model_induction_cost): Do not cost for pure SLP.
12410 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12411 of induction in inner loop vectorization.
12412 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12413 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12414 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12416 (vect_analyze_slp_cost_1): Cost induction.
12417 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12418 (vect_get_slp_vect_defs): Likewise.
12419 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12420 (vect_transform_stmt): Handle SLP reductions.
12421 * tree-vectorizer.h (vectorizable_induction): Adjust.
12423 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12425 * config/rs6000/rs6000.c (make_resolver_func): Update
12426 init_lowered_empty_function call.
12428 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12430 * doc/invoke.texi: Document the -fprofile-abs-path option.
12431 * common.opt (fprofile-abs-path): New option.
12432 * gcov-io.h (gcov_write_filename): Declare.
12433 * gcov-io.c (gcov_write_filename): New function.
12434 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12435 * profile.c (output_location): Likewise.
12437 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12439 * shring-wrap.c: Revert accidental commit.
12441 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12443 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12445 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12447 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12449 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12450 profile in callgraph edge.
12451 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12452 (apply_scale): Likewise.
12453 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12454 Also scale profile when inlining function with zero profile.
12455 (initialize_cfun): Update exit block profile even when it is zero.
12456 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12457 when profile is read.
12459 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12461 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12462 (CLONE_*): New constants to define the processors we can generate
12463 code for with the target_clone attribute.
12464 (rs6000_clone_map): New array to identify which clone processors
12465 the current program is running on.
12466 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12467 target_clone attribute.
12468 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12469 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12470 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12471 (cpu_expand_builtin): Add support for target_clone attribute.
12472 (rs6000_valid_attribute_p): Allow "default" attribute.
12473 (get_decl_name): New debug function to simplify printing the
12474 current function name in debugging statements.
12475 (rs6000_clone_priority): New functions to support the target_clone
12476 attribute, and be able to generate code to switch between ISA 2.05
12477 through ISA 3.0 (power6 through power9).
12478 (rs6000_compare_version_priority): Likewise.
12479 (rs6000_get_function_versions_dispatcher): Likewise.
12480 (make_resolver_func): Likewise.
12481 (add_condition_to_bb): Likewise.
12482 (dispatch_function_versions): Likewise.
12483 (rs6000_generate_version_dispatcher_body): Likewise.
12484 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12485 (fusion_gpr_load_p): Fix a spacing issue.
12486 * doc/extend.texi (Common Function Attributes): Document that the
12487 PowerPC supports the target_clone attribute.
12489 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12491 * config/arm/arm.h: explain F symbol found in description of ARM
12492 register allocation in its legend.
12494 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12496 * config/mips/frame-header-opt.c: Include profile-count.h.
12497 * config/riscv/riscv.c: Include profile-count.h
12499 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12501 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12503 (sm_set_flag_if_changed): Add bbs field.
12504 (execute_sm_if_changed_flag_set): Pass BBS.
12505 (execute_sm): Update.
12507 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12509 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12512 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12514 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12515 (peephole2): New peephole2 to emit the above.
12516 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12518 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12520 * config/aarch64/aarch64.c (define_peephole2 above
12521 *sub_<shift>_<mode>): New peephole.
12523 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12525 * config/i386/i386.c (make_resolver_func): Update.
12526 * Makefile.in: Add profile-count.h and profile-count.o
12527 * auto-profile.c (afdo_indirect_call): Update to new API.
12528 (afdo_set_bb_count): Update.
12529 (afdo_propagate_edge): Update.
12530 (afdo_propagate_circuit): Update.
12531 (afdo_calculate_branch_prob): Update.
12532 (afdo_annotate_cfg): Update.
12533 * basic-block.h: Include profile-count.h
12534 (struct edge_def): Turn count to profile_count.
12535 (struct basic_block_def): Likewie.
12536 (REG_BR_PROB_BASE): Move to profile-count.h
12537 (RDIV): Move to profile-count.h
12538 * bb-reorder.c (max_entry_count): Turn to profile_count.
12539 (find_traces): Update.
12540 (rotate_loop):Update.
12541 (connect_traces):Update.
12542 (sanitize_hot_paths):Update.
12543 * bt-load.c (migrate_btr_defs): Update.
12544 * cfg.c (RDIV): Remove.
12545 (init_flow): Use alloc_block.
12546 (alloc_block): Uninitialize count.
12547 (unchecked_make_edge): Uninitialize count.
12548 (check_bb_profile): Update.
12549 (dump_edge_info): Update.
12550 (dump_bb_info): Update.
12551 (update_bb_profile_for_threading): Update.
12552 (scale_bbs_frequencies_int): Update.
12553 (scale_bbs_frequencies_gcov_type): Update.
12554 (scale_bbs_frequencies_profile_count): New.
12555 * cfg.h (update_bb_profile_for_threading): Update.
12556 (scale_bbs_frequencies_profile_count): Declare.
12557 * cfgbuild.c (compute_outgoing_frequencies): Update.
12558 (find_many_sub_basic_blocks): Update.
12559 * cfgcleanup.c (try_forward_edges): Update.
12560 (try_crossjump_to_edge): Update.
12561 * cfgexpand.c (expand_gimple_tailcall): Update.
12562 (construct_exit_block): Update.
12563 * cfghooks.c (verify_flow_info): Update.
12564 (dump_bb_for_graph): Update.
12565 (split_edge): Update.
12566 (make_forwarder_block): Update.
12567 (duplicate_block): Update.
12568 (account_profile_record): Update.
12569 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12570 (get_estimated_loop_iterations): Update.
12571 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12572 (single_likely_exit): Update.
12573 * cfgloopmanip.c (scale_loop_profile): Update.
12575 (set_zero_probability): Update.
12576 (lv_adjust_loop_entry_edge): Update.
12577 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12578 (purge_dead_edges): Update.
12579 (rtl_account_profile_record): Update.
12580 * cgraph.c (cgraph_node::create): Uninitialize count.
12581 (symbol_table::create_edge): Uninitialize count.
12582 (cgraph_update_edges_for_call_stmt_node): Update.
12583 (cgraph_edge::dump_edge_flags): Update.
12584 (cgraph_node::dump): Update.
12585 (cgraph_edge::maybe_hot_p): Update.
12586 * cgraph.h: Include profile-count.h
12587 (create_clone), create_edge, create_indirect_edge): Update.
12588 (cgraph_node): Turn count to profile_count.
12589 (cgraph_edge0: Likewise.
12590 (make_speculative, clone): Update.
12591 (create_edge): Update.
12592 (init_lowered_empty_function): Update.
12593 * cgraphclones.c (cgraph_edge::clone): Update.
12594 (duplicate_thunk_for_node): Update.
12595 (cgraph_node::create_clone): Update.
12596 * cgraphunit.c (cgraph_node::analyze): Update.
12597 (cgraph_node::expand_thunk): Update.
12598 * final.c (dump_basic_block_info): Update.
12599 * gimple-streamer-in.c (input_bb): Update.
12600 * gimple-streamer-out.c (output_bb): Update.
12601 * graphite.c (print_global_statistics): Update.
12602 (print_graphite_scop_statistics): Update.
12603 * hsa-brig.c: Include basic-block.h.
12604 * hsa-dump.c: Include basic-block.h.
12605 * hsa-gen.c (T sum_slice): Update.
12606 (convert_switch_statements):Update.
12607 * hsa-regalloc.c: Include basic-block.h.
12608 * ipa-chkp.c (chkp_produce_thunks): Update.
12609 * ipa-cp.c (struct caller_statistics): Update.
12610 (init_caller_stats): Update.
12611 (gather_caller_stats): Update.
12612 (ipcp_cloning_candidate_p): Update.
12613 (good_cloning_opportunity_p): Update.
12614 (get_info_about_necessary_edges): Update.
12615 (dump_profile_updates): Update.
12616 (update_profiling_info): Update.
12617 (update_specialized_profile): Update.
12618 (perhaps_add_new_callers): Update.
12619 (decide_about_value): Update.
12620 (ipa_cp_c_finalize): Update.
12621 * ipa-devirt.c (struct odr_type_warn_count): Update.
12622 (struct decl_warn_count): Update.
12623 (struct final_warning_record): Update.
12624 (possible_polymorphic_call_targets): Update.
12625 (ipa_devirt): Update.
12626 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12627 * ipa-icf.c (sem_function::merge): Update.
12628 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12629 * ipa-inline.c (compute_uninlined_call_time): Update.
12630 (compute_inlined_call_time): Update.
12631 (want_inline_small_function_p): Update.
12632 (want_inline_self_recursive_call_p): Update.
12633 (edge_badness): Update.
12634 (lookup_recursive_calls): Update.
12635 (recursive_inlining): Update.
12636 (inline_small_functions): Update.
12637 (dump_overall_stats): Update.
12638 (dump_inline_stats): Update.
12639 * ipa-profile.c (ipa_profile_generate_summary): Update.
12640 (ipa_propagate_frequency): Update.
12641 (ipa_profile): Update.
12642 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12643 * ipa-utils.c (ipa_merge_profiles): Update.
12644 * loop-doloop.c (doloop_modify): Update.
12645 * loop-unroll.c (report_unroll): Update.
12646 (unroll_loop_runtime_iterations): Update.
12647 * lto-cgraph.c (lto_output_edge): Update.
12648 (lto_output_node): Update.
12649 (input_node): Update.
12650 (input_edge): Update.
12651 (merge_profile_summaries): Update.
12652 * lto-streamer-in.c (input_cfg): Update.
12653 * lto-streamer-out.c (output_cfg): Update.
12654 * mcf.c (create_fixup_graph): Update.
12655 (adjust_cfg_counts): Update.
12656 (sum_edge_counts): Update.
12657 * modulo-sched.c (sms_schedule): Update.
12658 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12659 * predict.c (maybe_hot_count_p): Update.
12660 (probably_never_executed): Update.
12661 (dump_prediction): Update.
12662 (combine_predictions_for_bb): Update.
12663 (propagate_freq): Update.
12664 (handle_missing_profiles): Update.
12665 (counts_to_freqs): Update.
12666 (rebuild_frequencies): Update.
12667 (force_edge_cold): Update.
12668 * predict.h: Include profile-count.h
12669 (maybe_hot_count_p, counts_to_freqs): UPdate.
12670 * print-rtl-function.c: Do not include cfg.h
12671 * print-rtl.c: Include basic-block.h
12672 * profile-count.c: New file.
12673 * profile-count.h: New file.
12674 * profile.c (is_edge_inconsistent): Update.
12675 (correct_negative_edge_counts): Update.
12676 (is_inconsistent): Update.
12677 (set_bb_counts): Update.
12678 (read_profile_edge_counts): Update.
12679 (compute_frequency_overlap): Update.
12680 (compute_branch_probabilities): Update; Initialize and deinitialize
12682 (branch_prob): Update.
12683 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12684 (edge_gcov_count): New.
12685 (bb_gcov_count): New.
12686 * shrink-wrap.c (try_shrink_wrapping): Update.
12687 * tracer.c (better_p): Update.
12688 * trans-mem.c (expand_transaction): Update.
12689 (ipa_tm_insert_irr_call): Update.
12690 (ipa_tm_insert_gettmclone_call): Update.
12691 * tree-call-cdce.c: Update.
12692 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12693 (gimple_duplicate_sese_tail): Update.
12694 (gimple_account_profile_record): Update.
12695 (execute_fixup_cfg): Update.
12696 * tree-inline.c (copy_bb): Update.
12697 (copy_edges_for_bb): Update.
12698 (initialize_cfun): Update.
12699 (freqs_to_counts): Update.
12700 (copy_cfg_body): Update.
12701 (expand_call_inline): Update.
12702 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12703 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12704 (try_unroll_loop_completely): Update.
12705 (try_peel_loop): Update.
12706 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12707 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12708 * tree-ssa-loop-split.c (connect_loops): Update.
12709 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12710 * tree-ssa-reassoc.c (branch_fixup): Update.
12711 * tree-ssa-tail-merge.c (replace_block_by): Update.
12712 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12713 (compute_path_counts): Update.
12714 (update_profile): Update.
12715 (recompute_probabilities): Update.
12716 (update_joiner_offpath_counts): Update.
12717 (estimated_freqs_path): Update.
12718 (freqs_to_counts_path): Update.
12719 (clear_counts_path): Update.
12720 (ssa_fix_duplicate_block_edges): Update.
12721 (duplicate_thread_path): Update.
12722 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12723 (struct switch_conv_info): Update.
12724 * tree-tailcall.c (decrease_profile): Update.
12725 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12726 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12727 * value-prof.c (check_counter): Update.
12728 (gimple_divmod_fixed_value): Update.
12729 (gimple_mod_pow2): Update.
12730 (gimple_mod_subtract): Update.
12731 (gimple_ic_transform): Update.
12732 (gimple_stringop_fixed_value): Update.
12733 * value-prof.h (gimple_ic): Update.
12735 2017-06-02 Carl Love <cel@us.ibm.com>
12737 * config/rs6000/rs6000-c: Add support for built-in functions
12738 vector double vec_doublee (vector signed int);
12739 vector double vec_doublee (vector unsigned int);
12740 vector double vec_doublee (vector float);
12741 vector double vec_doubleh (vector signed int);
12742 vector double vec_doubleh (vector unsigned int);
12743 vector double vec_doubleh (vector float);
12744 vector double vec_doublel (vector signed int);
12745 vector double vec_doublel (vector unsigned int);
12746 vector double vec_doublel (vector float);
12747 vector double vec_doubleo (vector signed int);
12748 vector double vec_doubleo (vector unsigned int);
12749 vector double vec_doubleo (vector float);.
12750 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12751 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12753 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12754 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12755 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12757 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12758 vec_doublel, vec_doubleh.
12759 * doc/extend.texi: Update the built-in documentation file for the
12760 new built-in functions.
12762 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12765 * ipa-inline-analysis.c (free_growth_caches): Set
12766 edge_removal_hook_holder to NULL after removing it.
12768 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12770 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12771 comparision with zero.
12773 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12774 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12775 for early expansion of vec_min and vec_max builtins.
12776 (builtin_function_type): Add min/max unsigned variants to those
12777 identified as having unsigned arguments.
12779 2017-06-02 Olivier Hainque <hainque@adacore.com>
12781 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12783 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12785 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12786 Use VALL_F16 iterator rather than VALL.
12788 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12790 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12791 Emit CBNZ inside loop when doing a strong exchange and comparing
12792 against zero. Generate the CC flags after the loop.
12794 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12796 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12797 (dl_section_ref): New.
12798 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12799 On AIX, append an expression to subtract the size of the
12800 section length to dl_section_ref.
12802 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12804 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12805 for early expansion of vector absolute builtins.
12807 2017-06-02 Richard Biener <rguenther@suse.de>
12809 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12810 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12812 2017-06-02 Richard Biener <rguenther@suse.de>
12814 PR tree-optimization/80948
12815 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12816 stmt order as well.
12818 2017-06-02 Richard Biener <rguenther@suse.de>
12820 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12822 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12823 for inductions as relevant.
12825 2017-06-02 Richard Biener <rguenther@suse.de>
12827 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12828 (vectorizable_induction): ... this. Remove dead code.
12830 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12832 * builtins. (expand_builtin_alloca): Remove second parameter and
12833 infer its value from the first parameter instead.
12834 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12836 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12838 PR rtl-optimization/80903
12839 * loop-doloop.c (add_test): Unshare sequence.
12841 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12843 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12845 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12847 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12849 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12850 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12851 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12852 (xlogue_layout::get_stub_rtx): Make static.
12853 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12854 (xlogue_layout::compute_stub_managed_regs): Rename to...
12855 (xlogue_layout::count_stub_managed_regs): ...this.
12856 (xlogue_layout::is_stub_managed_reg): New function.
12857 (xlogue_layout::m_stub_names): Rename to...
12858 (xlogue_layout::s_stub_names): ...this, make static.
12859 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12860 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12861 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12862 xlogue_layout::s_stub_names): Instantiate statics.
12863 (stub_managed_regs): Remove.
12864 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12865 (disable_call_ms2sysv_xlogues): Rename to...
12866 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
12867 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
12869 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
12870 change after reload_completed.
12871 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
12873 (ix86_expand_prologue): Likewise.
12874 (ix86_expand_epilogue): Likewise.
12875 (ix86_expand_split_stack_prologue): Likewise.
12876 (ix86_compute_frame_layout): Remove frame parameter ...
12877 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
12878 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
12880 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
12881 (ix86_frame): Move from here ...
12882 * config/i386/i386.h (ix86_frame): ... to here.
12883 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
12884 complete ix86_frame data structure instead. Remove some_ld_name.
12886 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12888 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
12889 symbols that hold a DECL_VALUE_EXPR.
12891 2017-06-01 Martin Jambor <mjambor@suse.cz>
12893 PR tree-optimization/80898
12894 * tree-sra.c (process_subtree_disqualification): Removed.
12895 (disqualify_candidate): Do not acll
12896 process_subtree_disqualification.
12897 (subtree_mark_written_and_enqueue): New function.
12898 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12899 RHS has been disqualified and re-queue LHS if necessary. Apart
12900 from that, ignore disqualified RHS.
12902 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12904 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12905 address fetch for z10 or later.
12907 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12909 * config/arc/arc.md (tst_movb): Add guard when splitting.
12911 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12913 * config/arc/arc.c (arc_can_eliminate): Test against
12914 arc_frame_pointer_needed.
12916 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12918 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12919 to prevent store reordering.
12920 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12921 (type): Add block type.
12922 (stack_tie): Define special instruction to be used in
12925 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12927 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12928 constraint. It is not valid for the pattern.
12929 (noncommutative_binary_comparison): Likewise.
12931 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12933 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12936 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12938 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12939 be used by the reg-alloc.
12941 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12943 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12944 reg-alloc when having mul64 or mul32x16 instructions.
12945 (mulsidi3): Likewise.
12946 (umulsidi3): Likewise.
12947 (mulsi32x16): New pattern.
12948 (mulsi64): Likewise.
12949 (mulsidi64): Likewise.
12950 (umulsidi64): Likewise.
12951 (MUL32x16_REG): Define.
12952 (mul64_600): Use MUL32x16_REG.
12953 (mac64_600): Likewise.
12954 (umul64_600): Likewise.
12955 (umac64_600): Likewise.
12957 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12959 * config/arc/arc.md (mulsi3_700): Make it commutative.
12961 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12963 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12965 (*sign_extendsidi2_insn): Likewise for movstosw.
12967 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12969 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12970 the type of the input discriminant value. Convert the
12971 discriminant value of signedness vary.
12973 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12975 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12976 Add to -Wall section.
12978 2017-06-01 Richard Biener <rguenther@suse.de>
12980 PR middle-end/66313
12981 * fold-const.c (fold_plusminus_mult_expr): If the factored
12982 factor may be zero use a wrapping type for the inner operation.
12983 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12984 and handle moved defs.
12985 (process_assignment): Properly guard the unary op case. Return a
12986 tri-state indicating that moving the stmt before the call may allow
12987 to continue. Pass through to_move.
12988 (find_tail_calls): Handle moving unrelated defs before
12991 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12994 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12995 splitter result in the canonical way.
12997 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12999 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13000 also for 32bit target. Update insn attributes.
13001 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13003 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13005 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13006 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13007 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13008 (_mm_maskz_min_ss): New intrinsics.
13010 2017-05-31 Martin Liska <mliska@suse.cz>
13012 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13013 Change comment style to one we normally use.
13014 (vectorizable_reduction): Likewise.
13015 (vectorizable_induction): Likewise.
13016 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13017 (vectorizable_call): Likewise.
13018 (vectorizable_simd_clone_call): Likewise.
13019 (vectorizable_conversion): Likewise.
13020 (vectorizable_assignment): Likewise.
13021 (vectorizable_shift): Likewise.
13022 (vectorizable_operation): Likewise.
13023 (vectorizable_store): Likewise.
13024 (vectorizable_load): Likewise.
13025 * tree-vectorizer.h: Likewise.
13027 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13029 * passes.c (emergency_dump_function): New.
13030 * tree-pass.h (emergency_dump_function): Declare.
13031 * plugin.c (plugins_internal_error_function): Remove.
13032 * plugin.h (plugins_internal_error_function): Remove declaration.
13033 * toplev.c (internal_error_function): New static function. Use it...
13034 (general_init): ...here.
13036 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13038 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13039 (arc_rtx_costs): Add costs for new patterns.
13040 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13041 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13043 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13045 * tree-ssa-strlen.c (get_next_strinfo): New function.
13046 (get_stridx_plus_constant): Use it.
13047 (zero_length_string): Likewise.
13048 (adjust_related_strinfos): Likewise.
13049 (adjust_last_stmt): Likewise.
13051 2017-05-31 Richard Biener <rguenther@suse.de>
13054 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13055 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13057 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13059 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13060 loop_vinfo argument and use of dependence distance vectors.
13061 Check instead whether the two references differ only in their
13062 initial value and assume that they have the same alignment if the
13063 difference is a multiple of the vector alignment.
13064 (vect_analyze_data_refs_alignment): Update call accordingly.
13066 2017-05-31 Martin Liska <mliska@suse.cz>
13069 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13071 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13073 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13074 (create_intersect_range_checks): Move from ...
13075 * tree-data-ref.c (create_intersect_range_checks_index)
13076 (create_intersect_range_checks): ... to here.
13077 (create_runtime_alias_checks): New function factored from ...
13078 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13079 here. Call above function.
13080 * tree-data-ref.h (create_runtime_alias_checks): New function.
13082 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13084 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13085 segment length for dr_b and compute it in wide_int.
13087 2017-05-31 Richard Biener <rguenther@suse.de>
13089 PR tree-optimization/80906
13090 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13091 and pass through iv_map.
13092 (copy_bb_and_scalar_dependences): Adjust.
13093 (translate_pending_phi_nodes): Likewise.
13094 (copy_loop_close_phi_args): Handle code-generating IVs instead
13097 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13099 * diagnostic-color.c (color_dict): Add "type-diff".
13100 (parse_gcc_colors): Update comment.
13101 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13102 -fdiagnostics-show-template-tree and -fno-elide-type.
13103 (GCC_COLORS): Add type-diff to example.
13105 (-fdiagnostics-show-template-tree): New.
13106 (-fno-elide-type): New.
13107 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13108 the pp_format_decoder callback. Call any m_format_postprocessor's
13110 (pretty_printer::pretty_printer): Initialize
13111 m_format_postprocessor.
13112 (pretty_printer::~pretty_printer): Delete any
13113 m_format_postprocessor.
13114 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13115 (class format_postprocessor): New class.
13116 (struct pretty_printer::format_decoder): Document the new parameters.
13117 (struct pretty_printer::m_format_postprocessor): New field.
13118 * tree-diagnostic.c (default_tree_printer): Update for new
13119 bool and const char ** params.
13120 * tree-diagnostic.h (default_tree_printer): Likewise.
13122 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13124 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13125 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13126 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13127 rs6000_gen_cell_microcode code.
13128 (rs6000_final_prescan_insn): Delete.
13129 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13130 "warn-cell-microcode" entries.
13131 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13132 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13133 throughout. Change cc_reg_not_micro_cr0_operand to
13134 cc_reg_not_cr0_operand throughout.
13135 (*extendhi<mode>2_noload): Delete.
13136 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13137 (mwarn-cell-microcode): Delete.
13138 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13139 -mgen-cell-microcode and -mwarn-cell-microcode.
13141 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13144 * config/i386/constraints.md (Yd): New constraint.
13146 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13147 and (?Yd, r) alternatives. Update insn attributes.
13148 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13149 and (?*Yd, r) alternatives. Update insn attributes.
13150 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13152 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13154 * gimplify.c (gimplify_modify_expr): Don't create a
13155 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13158 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13160 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13162 2017-05-30 Richard Biener <rguenther@suse.de>
13164 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13165 and reduc_def fields.
13166 (STMT_VINFO_REDUC_TYPE): New define.
13167 (STMT_VINFO_REDUC_DEF): Likewise.
13168 (vect_force_simple_reduction): Adjust prototype.
13169 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13170 (vect_is_simple_reduction): Remove check_reduction argument.
13171 (vect_force_simple_reduction): Adjust and set
13172 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13173 (vectorizable_reduction): Do not re-do reduction analysis
13174 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13175 * tree-parloops.c (gather_scalar_reductions): Adjust.
13177 2017-05-30 Richard Biener <rguenther@suse.de>
13179 PR middle-end/80901
13180 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13183 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13185 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13186 Introduce unknown_misalignment parameter and remove vf.
13187 (vect_peeling_hash_get_lowest_cost):
13188 Pass unknown_misalignment parameter.
13189 (vect_enhance_data_refs_alignment):
13190 Fix unsupportable data ref treatment.
13192 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13194 * tree-vect-data-refs.c (vect_get_data_access_cost):
13195 Workaround for SLP handling.
13196 (vect_enhance_data_refs_alignment):
13197 Compute costs for doing no peeling at all, compare to the best
13198 peeling costs so far and avoid peeling if cheaper.
13200 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13202 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13203 Return peeling info and set costs to zero for unlimited cost
13205 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13206 with unknown misalignment. Compute and costs for unknown
13207 misalignment, compare them to the costs for known misalignment
13208 and choose the cheapest for peeling.
13210 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13212 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13213 (vect_get_peeling_costs_all_drs): Create function.
13214 (vect_peeling_hash_get_lowest_cost):
13215 Use vect_get_peeling_costs_all_drs.
13216 (vect_peeling_supportable): Create function.
13217 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13219 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13221 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13222 DR_HAS_NEGATIVE_STEP.
13223 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13224 (vect_enhance_data_refs_alignment): Use.
13225 (vect_duplicate_ssa_name_ptr_info): Use.
13226 * tree-vectorizer.h (dr_misalignment): Use.
13227 (known_alignment_for_access_p): Use.
13229 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13232 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13234 (has_section_name): New function.
13236 2017-05-30 Martin Liska <mliska@suse.cz>
13239 * auto-profile.c (get_function_decl_from_block): Fix
13242 2017-05-30 Richard Biener <rguenther@suse.de>
13244 PR middle-end/80876
13245 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13247 2017-05-30 Martin Liska <mliska@suse.cz>
13249 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13250 * dumpfile.h (struct dump_file_info): Remove ctors.
13252 2017-05-30 Martin Liska <mliska@suse.cz>
13254 * predict.def: Fix GNU coding style.
13256 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13258 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13259 Mark 'to' argument with ATTRIBUTE_UNUSED.
13261 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13263 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13264 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13265 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13268 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13270 * doc/install.texi (Options specification): Restore entry of
13271 --enable-sjlj-exceptions.
13273 2017-05-27 Michael Eager <eager@eagercon.com>
13276 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13278 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13280 * config/microblaze/microblaze.h
13281 (FIXED_REGISTERS): Update in macro.
13282 (CALL_USED_REGISTERS): Update in macro.
13284 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13286 * doc/install.texi: Add links to macOS binary distributions.
13288 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13292 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13294 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13296 2017-05-26 Martin Liska <mliska@suse.cz>
13298 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13300 2017-05-26 Martin Liska <mliska@suse.cz>
13302 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13303 always leading ';; '.
13304 (dump_bb_info): Likewise.
13305 (brief_dump_cfg): Likewise.
13306 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13307 * dumpfile.c: Remove usage of TDF_VERBOSE.
13308 * dumpfile.h (enum dump_kind): Likewise.
13309 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13310 * print-tree.c (debug_verbose): Remove.
13311 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13312 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13313 * tree-diagnostic.c (default_tree_printer): Replace
13314 TDF_DIAGNOSTIC with TDF_SLIM.
13316 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13318 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13319 in parameter loop, rather than loop_vinfo.
13320 (create_intersect_range_checks): Ditto.
13321 (vect_create_cond_for_alias_checks): Update call to above functions.
13323 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13325 PR tree-optimization/80815
13326 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13327 for merging runtime alias checks. Handle negative DR_STEPs.
13329 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13331 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13333 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13334 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13335 out code pruning runtime alias checks.
13336 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13337 factored out from above.
13338 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13340 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13342 (prune_runtime_alias_test_list): New decalaration.
13344 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13346 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13347 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13348 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13349 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13350 (operator==, comp_dr_with_seg_len_pair): Ditto.
13351 (vect_prune_runtime_alias_test_list): Ditto.
13353 2017-05-26 Martin Liska <mliska@suse.cz>
13356 * params.def: Bound partial-inlining-entry-probability param.
13358 2017-05-26 Marek Polacek <polacek@redhat.com>
13361 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13364 2017-05-26 Richard Biener <rguenther@suse.de>
13366 PR tree-optimization/80842
13367 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13370 2017-05-26 Richard Biener <rguenther@suse.de>
13372 PR tree-optimization/80844
13373 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13375 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13377 * doc/md.texi (Machine Constraints): Update x86 family
13378 machine constraints section to match 'config/i386/constraints.md'.
13380 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13382 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13384 2017-05-25 Nathan Sidwell <nathan@acm.org>
13386 * doc/invoke.texi (--enable-languages): Update documentation.
13388 2017-05-25 Martin Liska <mliska@suse.cz>
13390 * dumpfile.c: Add TDF_FOLDING.
13391 * dumpfile.h (enum dump_kind): Likewise.
13392 * genmatch.c (dt_simplify::gen_1): Use it.
13394 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13396 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13398 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13400 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13401 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13403 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13405 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13406 * match.pd (X == C): Rewrite it here.
13407 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13408 with_certain_nonzero_bits2): New predicates.
13409 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13411 2017-05-24 Nathan Sidwell <nathan@acm.org>
13413 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13416 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13419 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13421 * config/powerpcspe: New port. Files are copied from the rs6000
13422 port, with "rs6000" in filenames replaced by "powerpcspe".
13424 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13426 PR rtl-optimization/80754
13427 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13429 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13431 * config/sparc/sparc.md (length): Return the correct value for -mflat
13432 sibcalls to match output_sibcall.
13434 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13438 * config/rs6000/rs6000.c (struct machine_function): Add new field
13440 (rs6000_get_separate_components): Init that field, use it.
13441 (rs6000_components_for_bb): Use the field.
13443 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13445 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13447 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13449 PR middle-end/80823
13450 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13452 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13455 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13456 address against address_operand predicate.
13457 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13459 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13461 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13462 parameters passed indirectly.
13464 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13466 * config/i386/i386.md (*movdi_internal): Remove SSE4
13467 alternative 18 (?r, *v). Update insn attributes.
13468 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13469 Update insn attributes.
13470 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13471 Update insn attributes.
13472 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13473 alternative 1 (r, v). Remove isa attribute.
13474 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13475 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13476 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13478 2017-05-23 Tom de Vries <tom@codesourcery.com>
13480 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13483 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13485 * cgraphunit.c (symbol_table::process_new_functions): Update.
13486 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13487 (inline_generate_summary): Rename to ...
13488 (ipa_fn_summary_generate): ... this one.
13489 (inline_read_summary): Rename to ...
13490 (ipa_fn_summary_read): ... this one.
13491 (inline_write_summary): Rename to ...
13492 (ipa_fn_summary_write): ... this one.
13493 (inline_free_summary): Rename to ...
13494 (ipa_free_fn_summary): ... this one.
13495 (pass_data_local_fn_summary, pass_local_fn_summary,
13496 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13497 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13498 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13499 make_pass_ipa_fn_summary): New.
13500 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13501 inline_write_summary, inline_free_summary): Remove.
13502 (ipa_free_fn_summary) : New.
13503 * ipa-inline.c (ipa_inline): Update.
13504 (pass_ipa_inline): Do not generate summaries.
13505 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13507 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13508 and add pass_ipa_fn_summary.
13509 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13511 (make_pass_inline_parameters): Remove.
13513 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13515 * omp-low.c (struct omp_context): Remove "default_kind" member.
13518 * omp-offload.c (execute_oacc_device_lower): Remove the
13519 parallelism dimensions function attributes for unparallelized
13520 OpenACC kernels constructs.
13522 2017-05-23 Martin Liska <mliska@suse.cz>
13524 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13526 (cgraph_edge::make_speculative): Likewise.
13527 (cgraph_edge::resolve_speculation): Likewise.
13528 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13529 (cgraph_node::dump): Likewise.
13530 * cgraph.h: Likewise.
13531 * cgraphunit.c (analyze_functions): Likewise.
13532 (symbol_table::compile): Likewise.
13533 * ipa-cp.c (print_all_lattices): Likewise.
13534 (determine_versionability): Likewise.
13535 (initialize_node_lattices): Likewise.
13536 (ipcp_verify_propagated_values): Likewise.
13537 (estimate_local_effects): Likewise.
13538 (update_profiling_info): Likewise.
13539 (create_specialized_node): Likewise.
13540 (perhaps_add_new_callers): Likewise.
13541 (decide_about_value): Likewise.
13542 (decide_whether_version_node): Likewise.
13543 (identify_dead_nodes): Likewise.
13544 (ipcp_store_bits_results): Likewise.
13545 * ipa-devirt.c (dump_targets): Likewise.
13546 (ipa_devirt): Likewise.
13547 * ipa-icf.c (sem_item::dump): Likewise.
13548 (sem_function::equals): Likewise.
13549 (sem_variable::equals): Likewise.
13550 (sem_item_optimizer::read_section): Likewise.
13551 (sem_item_optimizer::execute): Likewise.
13552 (congruence_class::dump): Likewise.
13553 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13554 (dump_inline_summary): Likewise.
13555 (estimate_node_size_and_time): Likewise.
13556 (inline_analyze_function): Likewise.
13557 * ipa-inline-transform.c (inline_call): Likewise.
13558 * ipa-inline.c (report_inline_failed_reason): Likewise.
13559 (want_early_inline_function_p): Likewise.
13560 (edge_badness): Likewise.
13561 (update_edge_key): Likewise.
13562 (inline_small_functions): Likewise.
13563 * ipa-profile.c (ipa_profile): Likewise.
13564 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13565 (ipa_make_edge_direct_to_target): Likewise.
13566 (remove_described_reference): Likewise.
13567 (ipa_impossible_devirt_target): Likewise.
13568 (propagate_controlled_uses): Likewise.
13569 (ipa_print_node_params): Likewise.
13570 (ipcp_transform_function): Likewise.
13571 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13572 (propagate_pure_const): Likewise.
13573 * ipa-reference.c (generate_summary): Likewise.
13574 (read_write_all_from_decl): Likewise.
13575 (propagate): Likewise.
13576 (ipa_reference_read_optimization_summary): Likewise.
13577 * ipa-utils.c (ipa_merge_profiles): Likewise.
13578 * ipa.c (walk_polymorphic_call_targets): Likewise.
13579 (symbol_table::remove_unreachable_nodes): Likewise.
13580 (ipa_single_use): Likewise.
13581 * passes.c (execute_todo): Likewise.
13582 * predict.c (drop_profile): Likewise.
13583 * symtab.c (symtab_node::get_dump_name): New function.
13584 (symtab_node::dump_name): Likewise.
13585 (symtab_node::dump_asm_name): Likewise.
13586 (symtab_node::dump_references): Likewise.
13587 (symtab_node::dump_referring): Likewise.
13588 (symtab_node::dump_base): Likewise.
13589 (symtab_node::debug_symtab): Likewise.
13590 * tree-sra.c (convert_callers_for_node): Likewise.
13591 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13592 * value-prof.c (init_node_map): Likewise.
13594 2017-05-23 Martin Liska <mliska@suse.cz>
13596 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13597 and symtab_node::debug_symtab to symbol_table::debug.
13598 * cgraphunit.c (analyze_functions): Use the renamed function.
13599 (symbol_table::compile): Likewise.
13600 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13601 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13602 * passes.c (execute_todo): Likewise.
13603 * symtab.c (symbol_table::dump): New function.
13604 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13606 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13608 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13609 that nonconst implies exec.
13611 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13613 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13614 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13615 (inline_edge_summary_vec): Turn into ...
13616 (ipa_call_summaries): ... this one.
13617 (redirect_to_unreachable, edge_set_predicate,
13618 evaluate_properties_for_edge, inline_summary_alloc,
13619 reset_ipa_call_summary, reset_inline_summary,
13620 inline_summary_t::duplicate): Update.
13621 (inline_edge_duplication_hook): Turn to ...
13622 (ipa_call_summary_t::duplicate): ... this one.
13623 (inline_edge_removal_hook): Turn to ...
13624 (ipa_call_summary_t::remove): ... this one.
13625 (dump_inline_edge_summary): Turn to ...
13626 (dump_ipa_call_summary): ... this one.
13627 (estimate_function_body_sizes): Update.
13628 (inline_update_callee_summaries): Update.
13629 (remap_edge_change_prob): Update.
13630 (remap_edge_summaries): Update.
13631 (inline_merge_summary): Update.
13632 (do_estimate_edge_time): Update.
13633 (inline_generate_summary): Update.
13634 (inline_read_section): Update.
13635 (inline_read_summary): Update.
13636 (inline_free_summary): Update.
13637 * ipa-inline.c (can_inline_edge_p): Update.
13638 (compute_inlined_call_time): Update.
13639 (want_inline_small_function_p): Update.
13640 (edge_badness): Update.
13641 (early_inliner): Update.
13642 * ipa-inline.h (inline_edge_summary): Turn to ...
13643 (ipa_call_summary): ... this one.
13644 (ipa_call_summary_t): New class.
13645 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13646 (ipa_call_summaries): New.
13647 (inline_edge_summary): Remove.
13648 (estimate_edge_growth): Update.
13649 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13650 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13651 * ipa-split.c (execute_split_functions): Update.
13652 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13654 2017-05-23 Tom de Vries <tom@codesourcery.com>
13656 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13657 attributes): Document rdrand effective target.
13659 2017-05-23 Tom de Vries <tom@codesourcery.com>
13661 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13662 attributes): Sort alphabetically.
13664 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13666 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13668 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13671 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13672 V2DF/V2DI splat into two separate patterns, one that handles
13673 registers, and the other that only handles memory. Drop support
13674 for splatting from a GPR on ISA 2.07 and then splitting the
13675 splat into direct move and splat.
13676 (vsx_splat_<mode>_reg): Likewise.
13677 (vsx_splat_<mode>_mem): Likewise.
13679 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13681 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13683 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13685 PR middle-end/80809
13686 * omp-low.c (finish_taskreg_remap): New function.
13687 (finish_taskreg_scan): If unit size of ctx->record_type
13688 is non-constant, unshare the size expression and replace
13689 decls in it with possible outer var refs.
13691 PR middle-end/80809
13692 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13693 GOVD_SHARED rather than GOVD_PRIVATE with it.
13694 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13695 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13697 PR middle-end/80853
13698 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13699 as last argument to build_outer_var_ref for pointer bases of array
13700 section reductions.
13702 2017-05-19 Martin Sebor <msebor@redhat.com>
13704 * print-tree.c (print_node): Print DECL_READ_P flag.
13706 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13708 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13709 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13710 * cgraph.c: Likewise.
13711 * cgraphunit.c: Likewise.
13712 * gengtype.c: Likewise.
13713 * ipa-cp.c: Likewise.
13714 * ipa-devirt.c: Likewise.
13715 * ipa-icf.c: Likewise.
13716 * ipa-predicate.c: Likewise.
13717 * ipa-profile.c: Likewise.
13718 * ipa-prop.c: Likewise.
13719 * ipa-split.c: Likewise.
13721 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13722 edge_predicate_pool, dump_inline_hints,
13723 inline_summary::account_size_time, redirect_to_unreachable,
13724 edge_set_predicate, set_hint_predicate,
13725 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13726 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13727 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13728 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13729 ipa_call_summary_t::remove, initialize_growth_caches,
13730 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13731 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13732 mark_modified, unmodified_parm_1, unmodified_parm,
13733 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13734 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13735 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13736 will_be_nonconstant_predicate, record_modified_bb_info,
13737 get_minimal_bb, record_modified, param_change_prob,
13738 phi_result_unknown_predicate, predicate_for_phi_result,
13739 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13740 estimate_function_body_sizes, compute_inline_parameters,
13741 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13742 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13743 inline_update_callee_summaries, remap_edge_change_prob,
13744 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13745 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13746 inline_analyze_function, inline_summary_t::insert,
13747 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13748 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13749 inline_free_summary): Move to ipa-fnsummary.h
13750 (predicate_t): Remove.
13751 * ipa-fnsummary.c: New file.
13752 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13753 (enum inline_hints_vals, inline_hints, agg_position_info,
13754 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13755 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13756 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13757 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13758 inline_read_summary, inline_write_summary, inline_free_summary,
13759 inline_analyze_function, initialize_inline_failed,
13760 inline_merge_summary, inline_update_overall_summary,
13761 compute_inline_parameters): Move to ipa-fnsummary.h
13762 * ipa-fnsummary.h: New file.
13763 * ipa-inline-transform.h: Include ipa-inline.h.
13764 * ipa-inline.c: LIkewise.
13766 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13768 * ipa-inline.c (edge_badness): Use inlined_time instead of
13769 inline_summaries->get.
13771 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13773 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13775 2017-05-22 Nathan Sidwell <nathan@acm.org>
13777 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13778 (fdump-lang): Document 'raw' option.
13779 * dumpfile.h (TDI_tu): Delete.
13780 * dumpfile.c (dump_files): Remove translation-unit.
13781 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13783 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13785 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13786 command option from $(AWK) call.
13787 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13788 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13789 [FORMAT]: Remove handling of variable.
13790 * config/avr/t-multilib: Regenerate.
13792 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13794 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13796 (dump_inline_summary): Do not print self_time.
13797 (estimate_function_body_sizes): Do not set self_time.
13798 (compute_inline_parameters): Likewise.
13799 (inline_read_section, inline_write_summary): Do not stream self_time.
13800 * ipa-inline.h (inline_summary): Drop self_time.
13802 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13804 * ipa-inline-analysis.c (account_size_time): Rename to ...
13805 (inline_summary::account_size_time): ... this one.
13806 (reset_ipa_call_summary): Turn to ...
13807 (ipa_call_summary::reset): ... this one.
13808 (reset_inline_summary): Turn to ...
13809 (inline_summary::reset): ... this one.
13810 (inline_summary_t::remove): Update.
13811 (inline_summary_t::duplicate): Update.
13812 (ipa_call_summary_t::remove): Update.
13813 (dump_inline_summary): Update.
13814 (estimate_function_body_sizes): Update.
13815 (compute_inline_parameters): Update.
13816 (estimate_node_size_and_time): Update.
13817 (inline_merge_summary): Update.
13818 (inline_update_overall_summary): Update.
13819 (inline_read_section): Update.
13820 (inline_write_summary): Update.
13821 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13822 add account_size_time and reset member functions.
13823 (ipa_call_summary): Add reset function.
13824 * ipa-predicate.h (predicate::operator &): Constify.
13826 2017-05-22 Richard Biener <rguenther@suse.de>
13828 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13830 2017-05-19 Jason Merrill <jason@redhat.com>
13832 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13834 2017-05-19 Marek Polacek <polacek@redhat.com>
13837 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13838 TYPE_OVERFLOW_WRAPS checks.
13840 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13842 * tree-core.h (enum omp_clause_default_kind): Add
13843 "OMP_CLAUSE_DEFAULT_PRESENT".
13844 * tree-pretty-print.c (dump_omp_clause): Handle it.
13845 * gimplify.c (enum gimplify_omp_var_data): Add
13846 "GOVD_MAP_FORCE_PRESENT".
13847 (gimplify_adjust_omp_clauses_1): Map it to
13848 "GOMP_MAP_FORCE_PRESENT".
13849 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13851 * gimplify.c (oacc_default_clause): Clarify.
13853 2017-05-19 Nathan Sidwell <nathan@acm.org>
13855 LANG_HOOK_REGISTER_DUMPS
13856 * toplev.c (general_init): Call register dump lang hook.
13857 * doc/invoke.texi: Document -fdump-lang option family.
13858 * dumpfile.c (dump_files): Remove class dump here.
13859 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13860 * dumpfile.h (tree_dump_index): Remove TDI_class.
13861 * langhooks-def.h (lhd_register_dumps): Declare.
13862 (LANG_HOOKS_REGISTER_DUMPS): Define.
13863 (LANG_HOOKS_INITIALIZER): Add it.
13864 * langhooks.c (lhd_register_dumps): Define.
13865 * langhooks.h (struct lang_hooks): Add register_dumps.
13867 2017-05-19 Nathan Sidwell <nathan@acm.org>
13869 * context.h (context::set_passes): New.
13870 * context.c (context::context): Do not create pass manager.
13871 * toplev.c (general_init): Create pass manager here.
13873 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
13875 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
13876 use this splitter if two add or or instructions would also work for
13877 the constant we want to generate.
13879 2017-05-19 Richard Biener <rguenther@suse.de>
13882 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
13883 predicate evaluation.
13885 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13887 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
13889 * ipa-inline.c (want_inline_small_function_p): Do not cast to
13892 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13894 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13895 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13896 (inline_edge_summary_vec): Turn into ...
13897 (ipa_call_summaries): ... this one.
13898 (redirect_to_unreachable, edge_set_predicate,
13899 evaluate_properties_for_edge, inline_summary_alloc,
13900 reset_ipa_call_summary, reset_inline_summary,
13901 inline_summary_t::duplicate): Update.
13902 (inline_edge_duplication_hook): Turn to ...
13903 (ipa_call_summary_t::duplicate): ... this one.
13904 (inline_edge_removal_hook): Turn to ...
13905 (ipa_call_summary_t::remove): ... this one.
13906 (dump_inline_edge_summary): Turn to ...
13907 (dump_ipa_call_summary): ... this one.
13908 (estimate_function_body_sizes): Update.
13909 (inline_update_callee_summaries): Update.
13910 (remap_edge_change_prob): Update.
13911 (remap_edge_summaries): Update.
13912 (inline_merge_summary): Update.
13913 (do_estimate_edge_time): Update.
13914 (inline_generate_summary): Update.
13915 (inline_read_section): Update.
13916 (inline_read_summary): Update.
13917 (inline_free_summary): Update.
13918 * ipa-inline.c (can_inline_edge_p): Update.
13919 (compute_inlined_call_time): Update.
13920 (want_inline_small_function_p): Update.
13921 (edge_badness): Update.
13922 (early_inliner): Update.
13923 * ipa-inline.h (inline_edge_summary): Turn to ...
13924 (ipa_call_summary): ... this one.
13925 (ipa_call_summary_t): New class.
13926 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13927 (ipa_call_summaries): New.
13928 (inline_edge_summary): Remove.
13929 (estimate_edge_growth): Update.
13930 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13931 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13932 * ipa-split.c (execute_split_functions): Update.
13933 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13935 2017-05-19 Richard Biener <rguenther@suse.de>
13937 PR middle-end/80764
13938 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13940 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13942 * config/rs6000/rs6000.c (struct machine_function): Add field
13943 fpr_is_wrapped_separately.
13944 (rs6000_get_separate_components): Use 64 components. Handle the
13945 new FPR components.
13946 (rs6000_components_for_bb): Handle the FPR components.
13947 (rs6000_emit_prologue_components): Handle the FPR components.
13948 (rs6000_emit_epilogue_components): Handle the FPR components.
13949 (rs6000_set_handled_components): Handle the FPR components.
13950 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13951 that are already separately shrink-wrapped.
13952 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13953 that are already separately shrink-wrapped.
13955 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13958 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13961 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13962 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13963 cases where the register allocator uses a move and an offsettable
13964 memory operation to/from a FPR register on ISA 2.06/2.07.
13965 (define_peephole2 for Altivec d-form store): Likewise.
13967 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13970 * config/i386/mmx.md (*mov<mode>_internal): Enable
13971 alternatives 11, 12, 13 and 14 also for 32bit targets.
13972 Remove alternatives 15, 16, 17 and 18.
13973 * config/i386/sse.md (vec_concatv2di): Change
13974 alternative (!x, *y) to (x, ?!*Yn).
13976 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13978 * dumpfile.h (enum dump_kind): Remove stray comma.
13980 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13982 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13983 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13984 predicate::num_conditions
13985 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13986 (CHANGED): turn into predicate::changed.
13987 (agg_position_info): Move to ipa-predicate.h
13988 (add_condition, predicate::add_clause, predicate::operator &=,
13989 predicate::or_with, predicate::evaluate, predicate::probability,
13990 dump_condition, dump_clause, predicate::dump,
13991 predicate::remap_after_duplication, predicate::remap_after_inlining,
13992 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13993 (evaluate_conditions_for_known_args): Update.
13994 (set_cond_stmt_execution_predicate): Update.
13995 * ipa-inline.h: Include ipa-predicate.h
13996 (condition, inline_param_summary, conditions, agg_position_info,
13997 predicate): Move to ipa-predicate.h
13998 * ipa-predicate.c: New file.
13999 * ipa-predicate.h: New file.
14001 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14003 * final.c (leaf_function_p): Check we are not in a sequence.
14005 2017-05-18 Martin Liska <mliska@suse.cz>
14007 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14008 * dumpfile.c (dump_register): Use new enum dump_kind.
14009 (get_dump_file_name): Likewise.
14010 (dump_enable_all): Likewise.
14011 (dump_switch_p_1): Likewise.
14012 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14013 * dumpfile.h (enum dump_kind): New enum type.
14014 (struct dump_file_info): Create constructor and
14015 format fields and comments.
14016 * passes.c (pass_manager::register_one_dump_file):
14018 * statistics.c (statistics_early_init): Likewise.
14019 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14020 TDF_TREE with TDF_SLIM.
14021 (gather_memory_references_ref): Likewise.
14023 2017-05-18 Martin Liska <mliska@suse.cz>
14025 * vec.h (struct vnull): Use it.
14027 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14029 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14030 (true_predicate, false_predicate, true_predicate_p,
14031 false_predicate_p): Remove.
14032 (single_cond_predicate, not_inlined_predicate): Turn to member function
14034 (add_condition): Update.
14035 (add_clause): Turn to...
14036 (predicate::add_clause): ... this one; update; allow passing NULL
14038 (and_predicates): Turn to ...
14039 (predicate::operator &=): ... this one.
14040 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14041 (or_predicates): Turn to ...
14042 (predicate::or_with): ... this one.
14043 (evaluate_predicate): Turn to ...
14044 (predicate::evaluate): ... this one.
14045 (predicate_probability): Turn to ...
14046 (predicate::probability): ... this one.
14047 (dump_condition): Update.
14048 (dump_predicate): Turn to ...
14049 (predicate::dump): ... this one.
14050 (account_size_time): Update.
14051 (edge_set_predicate): Update.
14052 (set_hint_predicate): UPdate.
14053 (evaluate_conditions_for_known_args): Update.
14054 (evaluate_properties_for_edge): Update.
14055 (remap_predicate_after_duplication): Turn to...
14056 (predicate::remap_after_duplication): ... this one.
14057 (remap_hint_predicate_after_duplication): Update.
14058 (inline_summary_t::duplicate): UPdate.
14059 (dump_inline_edge_summary): Update.
14060 (dump_inline_summary): Update.
14061 (set_cond_stmt_execution_predicate): Update.
14062 (set_switch_stmt_execution_predicate): Update.
14063 (compute_bb_predicates): Update.
14064 (will_be_nonconstant_expr_predicate): Update.
14065 (will_be_nonconstant_predicate): Update.
14066 (phi_result_unknown_predicate): Update.
14067 (predicate_for_phi_result): Update.
14068 (array_index_predicate): Update.
14069 (estimate_function_body_sizes): Update.
14070 (estimate_node_size_and_time): Update.
14071 (estimate_ipcp_clone_size_and_time): Update.
14072 (remap_predicate): Rename to ...
14073 (predicate::remap_after_inlining): ... this one.
14074 (remap_hint_predicate): Update.
14075 (inline_merge_summary): Update.
14076 (inline_update_overall_summary): Update.
14077 (estimate_size_after_inlining): Update.
14078 (read_predicate): Rename to ...
14079 (predicate::stream_in): ... this one.
14080 (read_inline_edge_summary): Update.
14081 (write_predicate): Rename to ...
14082 (predicate::stream_out): ... this one.
14083 (write_inline_edge_summary): Update.
14084 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14085 (clause_t): Turn to uint32_t
14086 (predicate): Turn to class; implement constructor and operators
14088 (size_time_entry): Update.
14089 (inline_summary): Update.
14090 (inline_edge_summary): Update.
14092 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14094 * fold-const.c (fold_binary_loc): Move transformation...
14095 * match.pd (C - X CMP X): ... here.
14097 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14099 * config/sparc/sparc.c (sparc_option_override): Set function
14100 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14101 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14103 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14105 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14107 2017-05-18 Marek Polacek <polacek@redhat.com>
14110 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14111 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14112 gimple_assign_load_p.
14114 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14116 PR middle-end/80692
14117 * real.c (do_compare): Give decimal_do_compare preference over
14118 comparing just the signs.
14120 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14122 * doc/md.texi (Canonicalization of Instructions): Describe the
14123 canonical form of instructions that inherently set a condition
14126 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14128 PR middle-end/80775
14129 * tree-cfg.c: Move deletion of unreachable case statements to after
14130 the merging of consecutive case labels.
14132 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14134 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14135 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14136 restoring of callee-saved registers.
14138 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14140 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14141 * config/visium/visium.c (single_set_and_flags): Likewise.
14142 * config/visium/visium.md (Substitutions): Likewise.
14144 2017-05-17 Martin Liska <mliska@suse.cz>
14146 * cfg.c: Introduce dump_flags_t type and
14147 use it instead of int type.
14149 * cfghooks.c: Likewise.
14150 * cfghooks.h (struct cfg_hooks): Likewise.
14151 * cfgrtl.c: Likewise.
14152 * cfgrtl.h: Likewise.
14153 * cgraph.c (cgraph_node::get_body): Likewise.
14154 * coretypes.h: Likewise.
14155 * domwalk.c: Likewise.
14156 * domwalk.h: Likewise.
14157 * dumpfile.c (struct dump_option_value_info): Likewise.
14158 (dump_enable_all): Likewise.
14159 (dump_switch_p_1): Likewise.
14160 (opt_info_switch_p): Likewise.
14161 * dumpfile.h (enum tree_dump_index): Likewise.
14162 (struct dump_file_info): Likewise.
14163 * genemit.c: Likewise.
14164 * generic-match-head.c: Likewise.
14165 * gengtype.c (open_base_files): Likewise.
14166 * gimple-pretty-print.c: Likewise.
14167 * gimple-pretty-print.h: Likewise.
14168 * graph.c (print_graph_cfg): Likewise.
14169 * graphite-scop-detection.c (dot_all_sese): Likewise.
14170 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14171 * loop-unroll.c (report_unroll): Likewise.
14172 * passes.c (pass_manager::register_one_dump_file): Likewise.
14173 * print-tree.c: Likewise.
14174 * statistics.c: Likewise.
14175 * tree-cfg.c: Likewise.
14176 * tree-cfg.h: Likewise.
14177 * tree-dfa.c: Likewise.
14178 * tree-dfa.h: Likewise.
14179 * tree-dump.c (dump_function): Likewise.
14180 * tree-dump.h (struct dump_info): Likewise.
14181 * tree-pretty-print.c: Likewise.
14182 * tree-pretty-print.h: Likewise.
14183 * tree-ssa-live.c: Likewise.
14184 * tree-ssa-live.h: Likewise.
14185 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14186 * tree-vect-loop.c: Likewise.
14187 * tree-vect-slp.c: Likewise.
14189 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14190 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14192 PR tree-optimization/80457
14193 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14194 of all arguments to a statement as scalar_to_vec operations.
14195 (vectorizable_call): Adjust call to vect_model_simple_cost for
14197 (vectorizable_conversion): Likewise.
14198 (vectorizable_assignment): Likewise.
14199 (vectorizable_shift): Likewise.
14200 (vectorizable_operation): Likewise.
14201 (vectorizable_comparison): Likewise.
14202 (vect_is_simple_cond): Record the def types for operands.
14203 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14204 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14205 for statement argument count.
14207 2017-05-16 Carl Love <cel@us.ibm.com>
14209 * config/rs6000/rs6000-c: Add support for built-in functions
14210 vector unsigned long long vec_bperm (vector unsigned long long,
14211 vector unsigned char)
14212 vector signed long long vec_mule (vector signed int,
14214 vector unsigned long long vec_mule (vector unsigned int,
14215 vector unsigned int)
14216 vector signed long long vec_mulo (vector signed int,
14218 vector unsigned long long vec_mulo (vector unsigned int,
14219 vector unsigned int)
14220 vector signed char vec_sldw (vector signed char,
14221 vector signed char,
14223 vector unsigned char vec_sldw (vector unsigned char,
14224 vector unsigned char,
14226 vector signed short vec_sldw (vector signed short,
14227 vector signed short,
14229 vector unsigned short vec_sldw (vector unsigned short,
14230 vector unsigned short,
14232 vector signed int vec_sldw (vector signed int,
14235 vector unsigned int vec_sldw (vector unsigned int,
14236 vector unsigned int,
14238 vector signed long long vec_sldw (vector signed long long,
14239 vector signed long long,
14241 vector unsigned long long vec_sldw (vector unsigned long long,
14242 vector unsigned long long,
14244 * config/rs6000/rs6000-c: Add support for built-in functions
14245 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14246 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14247 * doc/extend.texi: Update the built-in documentation for the
14248 new built-in functions.
14250 2017-05-16 Marek Polacek <polacek@redhat.com>
14254 * tree.c (save_expr): Don't fold the expression.
14256 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14258 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14259 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14260 and (?*y,m). Update insn attributes.
14262 2017-05-16 Martin Liska <mliska@suse.cz>
14264 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14265 flags argument of print_gimple_stmt, print_gimple_expr,
14266 print_generic_stmt and print_generic_expr.
14267 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14268 * coretypes.h: Likewise.
14269 * except.c (dump_eh_tree): Likewise.
14270 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14271 * gimple-pretty-print.h: Likewise.
14272 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14273 (backprop::push_to_worklist): Likewise.
14274 (backprop::pop_from_worklist): Likewise.
14275 (backprop::process_use): Likewise.
14276 (backprop::intersect_uses): Likewise.
14277 (note_replacement): Likewise.
14278 * gimple-ssa-store-merging.c
14279 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14280 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14281 (pass_store_merging::execute): Likewise.
14282 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14283 (ssa_base_cand_dump_callback): Likewise.
14284 (dump_incr_vec): Likewise.
14285 (replace_refs): Likewise.
14286 (replace_mult_candidate): Likewise.
14287 (create_add_on_incoming_edge): Likewise.
14288 (create_phi_basis): Likewise.
14289 (insert_initializers): Likewise.
14290 (all_phi_incrs_profitable): Likewise.
14291 (introduce_cast_before_cand): Likewise.
14292 (replace_one_candidate): Likewise.
14293 * gimplify.c (gimplify_expr): Likewise.
14294 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14295 (set_rename): Likewise.
14296 (rename_uses): Likewise.
14297 (copy_loop_phi_nodes): Likewise.
14298 (add_close_phis_to_merge_points): Likewise.
14299 (copy_loop_close_phi_args): Likewise.
14300 (copy_cond_phi_args): Likewise.
14301 (graphite_copy_stmts_from_block): Likewise.
14302 (translate_pending_phi_nodes): Likewise.
14303 * graphite-poly.c (print_pdr): Likewise.
14304 (dump_gbb_cases): Likewise.
14305 (dump_gbb_conditions): Likewise.
14306 (print_scop_params): Likewise.
14307 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14308 (build_cross_bb_scalars_use): Likewise.
14309 (gather_bbs::before_dom_children): Likewise.
14310 * hsa-dump.c (dump_hsa_immed): Likewise.
14311 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14312 (get_replacement_map): Likewise.
14313 * ipa-inline-analysis.c (dump_condition): Likewise.
14314 (estimate_function_body_sizes): Likewise.
14315 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14316 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14317 * ipa-prop.c (ipa_dump_param): Likewise.
14318 (ipa_print_node_jump_functions_for_edge): Likewise.
14319 (ipa_modify_call_arguments): Likewise.
14320 (ipa_modify_expr): Likewise.
14321 (ipa_dump_param_adjustments): Likewise.
14322 (ipa_dump_agg_replacement_values): Likewise.
14323 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14324 * ipa-pure-const.c (check_stmt): Likewise.
14325 (pass_nothrow::execute): Likewise.
14326 * ipa-split.c (execute_split_functions): Likewise.
14327 * omp-offload.c (dump_oacc_loop_part): Likewise.
14328 (dump_oacc_loop): Likewise.
14329 * trans-mem.c (tm_log_emit): Likewise.
14330 (tm_memopt_accumulate_memops): Likewise.
14331 (dump_tm_memopt_set): Likewise.
14332 (dump_tm_memopt_transform): Likewise.
14333 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14334 (print_loop): Likewise.
14335 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14336 (chkp_gather_checks_info): Likewise.
14337 (chkp_get_check_result): Likewise.
14338 (chkp_remove_check_if_pass): Likewise.
14339 (chkp_use_outer_bounds_if_possible): Likewise.
14340 (chkp_reduce_bounds_lifetime): Likewise.
14341 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14342 (chkp_mark_completed_bounds): Likewise.
14343 (chkp_register_incomplete_bounds): Likewise.
14344 (chkp_mark_invalid_bounds): Likewise.
14345 (chkp_maybe_copy_and_register_bounds): Likewise.
14346 (chkp_build_returned_bound): Likewise.
14347 (chkp_get_bound_for_parm): Likewise.
14348 (chkp_build_bndldx): Likewise.
14349 (chkp_get_bounds_by_definition): Likewise.
14350 (chkp_generate_extern_var_bounds): Likewise.
14351 (chkp_get_bounds_for_decl_addr): Likewise.
14352 * tree-chrec.c (chrec_apply): Likewise.
14353 * tree-data-ref.c (dump_data_reference): Likewise.
14354 (dump_subscript): Likewise.
14355 (dump_data_dependence_relation): Likewise.
14356 (analyze_overlapping_iterations): Likewise.
14357 * tree-inline.c (expand_call_inline): Likewise.
14358 (tree_function_versioning): Likewise.
14359 * tree-into-ssa.c (dump_defs_stack): Likewise.
14360 (dump_currdefs): Likewise.
14361 (dump_names_replaced_by): Likewise.
14362 (dump_update_ssa): Likewise.
14363 (update_ssa): Likewise.
14364 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14365 * tree-parloops.c (build_new_reduction): Likewise.
14366 (try_create_reduction_list): Likewise.
14367 (ref_conflicts_with_region): Likewise.
14368 (oacc_entry_exit_ok_1): Likewise.
14369 (oacc_entry_exit_single_gang): Likewise.
14370 * tree-pretty-print.h: Likewise.
14371 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14372 (get_scalar_evolution): Likewise.
14373 (add_to_evolution): Likewise.
14374 (get_loop_exit_condition): Likewise.
14375 (analyze_evolution_in_loop): Likewise.
14376 (analyze_initial_condition): Likewise.
14377 (analyze_scalar_evolution): Likewise.
14378 (instantiate_scev): Likewise.
14379 (number_of_latch_executions): Likewise.
14380 (gather_chrec_stats): Likewise.
14381 (final_value_replacement_loop): Likewise.
14382 (scev_const_prop): Likewise.
14383 * tree-sra.c (dump_access): Likewise.
14384 (disqualify_candidate): Likewise.
14385 (create_access): Likewise.
14386 (reject): Likewise.
14387 (maybe_add_sra_candidate): Likewise.
14388 (create_access_replacement): Likewise.
14389 (analyze_access_subtree): Likewise.
14390 (analyze_all_variable_accesses): Likewise.
14391 (sra_modify_assign): Likewise.
14392 (initialize_constant_pool_replacements): Likewise.
14393 (find_param_candidates): Likewise.
14394 (decide_one_param_reduction): Likewise.
14395 (replace_removed_params_ssa_names): Likewise.
14396 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14397 * tree-ssa-copy.c (dump_copy_of): Likewise.
14398 (copy_prop_visit_cond_stmt): Likewise.
14399 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14400 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14401 (record_equivalences_from_stmt): Likewise.
14402 * tree-ssa-dse.c (compute_trims): Likewise.
14403 (delete_dead_call): Likewise.
14404 (delete_dead_assignment): Likewise.
14405 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14406 (forward_propagate_into_cond): Likewise.
14407 (pass_forwprop::execute): Likewise.
14408 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14409 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14411 (move_computations_worker): Likewise.
14412 (execute_sm): Likewise.
14413 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14414 (remove_exits_and_undefined_stmts): Likewise.
14415 (remove_redundant_iv_tests): Likewise.
14416 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14417 (adjust_iv_update_pos): Likewise.
14418 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14419 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14420 (value_replacement): Likewise.
14421 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14422 * tree-ssa-pre.c (print_pre_expr): Likewise.
14423 (get_representative_for): Likewise.
14424 (create_expression_by_pieces): Likewise.
14425 (insert_into_preds_of_block): Likewise.
14426 (eliminate_insert): Likewise.
14427 (eliminate_dom_walker::before_dom_children): Likewise.
14428 (eliminate): Likewise.
14429 (remove_dead_inserted_code): Likewise.
14430 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14431 * tree-ssa-reassoc.c (get_rank): Likewise.
14432 (eliminate_duplicate_pair): Likewise.
14433 (eliminate_plus_minus_pair): Likewise.
14434 (eliminate_not_pairs): Likewise.
14435 (undistribute_ops_list): Likewise.
14436 (eliminate_redundant_comparison): Likewise.
14437 (update_range_test): Likewise.
14438 (optimize_range_tests_var_bound): Likewise.
14439 (optimize_vec_cond_expr): Likewise.
14440 (rewrite_expr_tree): Likewise.
14441 (rewrite_expr_tree_parallel): Likewise.
14442 (linearize_expr): Likewise.
14443 (break_up_subtract): Likewise.
14444 (linearize_expr_tree): Likewise.
14445 (attempt_builtin_powi): Likewise.
14446 (attempt_builtin_copysign): Likewise.
14447 (transform_stmt_to_copy): Likewise.
14448 (transform_stmt_to_multiply): Likewise.
14449 (dump_ops_vector): Likewise.
14450 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14451 (print_scc): Likewise.
14452 (set_ssa_val_to): Likewise.
14453 (visit_reference_op_store): Likewise.
14454 (visit_use): Likewise.
14455 (sccvn_dom_walker::before_dom_children): Likewise.
14456 (run_scc_vn): Likewise.
14457 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14459 (expr_hash_elt::print): Likewise.
14460 (const_and_copies::pop_to_marker): Likewise.
14461 (const_and_copies::record_const_or_copy_raw): Likewise.
14462 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14463 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14464 (dump_predicates): Likewise.
14465 (find_uninit_use): Likewise.
14466 (warn_uninitialized_phi): Likewise.
14467 (pass_late_warn_uninitialized::execute): Likewise.
14468 * tree-ssa.c (verify_vssa): Likewise.
14469 (verify_ssa): Likewise.
14470 (maybe_optimize_var): Likewise.
14471 * tree-vrp.c (dump_value_range): Likewise.
14472 (dump_all_value_ranges): Likewise.
14473 (dump_asserts_for): Likewise.
14474 (register_edge_assert_for_2): Likewise.
14475 (vrp_visit_cond_stmt): Likewise.
14476 (vrp_visit_switch_stmt): Likewise.
14477 (vrp_visit_stmt): Likewise.
14478 (vrp_visit_phi_node): Likewise.
14479 (simplify_cond_using_ranges_1): Likewise.
14480 (fold_predicate_in): Likewise.
14481 (evrp_dom_walker::before_dom_children): Likewise.
14482 (evrp_dom_walker::push_value_range): Likewise.
14483 (evrp_dom_walker::pop_value_range): Likewise.
14484 (execute_early_vrp): Likewise.
14486 2017-05-16 Richard Biener <rguenther@suse.de>
14488 * dwarf2out.c (loc_list_from_tree_1): Do not create
14489 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14491 2017-05-16 Richard Biener <rguenther@suse.de>
14493 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14495 (note_variable_value_in_expr): If we resolved the decl ref
14496 do not push to the stack.
14498 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14500 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14501 operations in fast-math mode.
14502 (vaddq_f16): Likewise.
14503 (vmul_f16): Likewise.
14504 (vmulq_f16): Likewise.
14505 (vsub_f16): Likewise.
14506 (vsubq_f16): Likewise.
14507 * config/arm/neon.md (add<mode>3): New.
14509 (fma:<VH:mode>3): New. Also remove outdated comment.
14512 2017-05-16 Martin Liska <mliska@suse.cz>
14516 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14517 (odr_types_equivalent_p): Likewise.
14519 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14521 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14523 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14526 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14527 non-interunit SSE move alternatives with '?'.
14528 (zero-extendsidi peephole2): New peephole to skip intermediate
14529 general register in SSE zero-extend sequence.
14531 2017-05-15 Jeff Law <law@redhat.com>
14533 * reorg.c (relax_delay_slots): Create a new variable to hold
14534 the temporary target rather than clobbering TARGET_LABEL.
14536 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14537 missing argument to extract_bit_field call.
14538 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14540 2017-05-15 Martin Liska <mliska@suse.cz>
14543 * gcc.c (process_command): Do not allow empty argument of -o option.
14545 2017-05-15 Renlin Li <renlin.li@arm.com>
14547 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14548 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14549 * config/aarch64/constraints.md (Usf): Add long call check.
14550 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14551 (call_value): Likewise.
14552 (sibcall): Likewise.
14553 (sibcall_value): Likewise.
14555 (call_value_insn): New.
14556 (sibcall_insn): Update rtx pattern.
14557 (sibcall_value_insn): Likewise.
14558 (call_internal): Remove.
14559 (call_value_internal): Likewise.
14560 (sibcall_internal): Likewise.
14561 (sibcall_value_internal): Likewise.
14562 (call_reg): Likewise.
14563 (call_symbol): Likewise.
14564 (call_value_reg): Likewise.
14565 (call_value_symbol): Likewise.
14567 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14570 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14572 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14574 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14575 compatible with CCGOCmode and with CCZmode.
14577 2017-05-14 Martin Sebor <msebor@redhat.com>
14579 PR middle-end/77671
14580 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14581 (gimple_fold_builtin_snprintf): Same.
14582 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14583 (gimple_fold_builtin_snprintf): Same.
14584 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14585 of character types.
14586 (is_call_safe): New function.
14587 (try_substitute_return_value): Call it.
14588 (try_simplify_call): New function.
14589 (pass_sprintf_length::handle_gimple_call): Call it.
14591 2017-05-14 Martin Sebor <msebor@redhat.com>
14593 PR middle-end/80669
14594 * builtins.c (expand_builtin_stpncpy): Simplify.
14596 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14598 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14599 * config/i386/i386.h
14600 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14601 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14602 (struct machine_function): Add new members call_ms2sysv,
14603 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14604 (struct machine_frame_state): New fields sp_realigned and
14605 sp_realigned_offset.
14606 * config/i386/i386.c
14607 (enum xlogue_stub): New enum.
14608 (enum xlogue_stub_sets): New enum.
14609 (class xlogue_layout): New class.
14610 (struct ix86_frame): New fields stack_realign_allocate_offset,
14611 stack_realign_offset and outlined_save_offset. Modify comments to
14612 detail stack layout when using out-of-line stubs.
14613 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14614 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14615 -mcall-ms2sysv-xlogues.
14616 (stub_managed_regs): New static variable.
14617 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14618 registers managed by out-of-line stub.
14619 (disable_call_ms2sysv_xlogues): New function.
14620 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14621 m->call_ms2sysv when appropriate and compute frame layout for
14623 (sp_valid_at, fp_valid_at): New inline functions.
14624 (choose_basereg): New function.
14625 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14627 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14628 Use align parameter of choose_baseaddr to generated aligned SSE movs
14630 (pro_epilogue_adjust_stack): Modify to track
14631 machine_frame_state::sp_realigned.
14632 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14633 (ix86_nsaved_sseregs): Likewise.
14634 (ix86_emit_save_regs): Likewise.
14635 (ix86_emit_save_regs_using_mov): Likewise.
14636 (ix86_emit_save_sse_regs_using_mov): Likewise.
14637 (get_scratch_register_on_entry): Likewise.
14638 (gen_frame_set): New function.
14639 (gen_frame_load): Likewise.
14640 (gen_frame_store): Likewise.
14641 (emit_outlined_ms2sysv_save): Likewise.
14642 (emit_outlined_ms2sysv_restore): Likewise.
14643 (ix86_expand_prologue): Modify stack re-alignment code and call
14644 emit_outlined_ms2sysv_save when appropriate.
14645 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14646 parameter rtx_insn *insn, which allows the function to be used to only
14647 generate the notes.
14648 (ix86_expand_epilogue): Modify validity checks of frame and stack
14649 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14650 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14651 * config/i386/predicates.md
14652 (save_multiple): New predicate.
14653 (restore_multiple): Likewise.
14654 * config/i386/sse.md
14655 (save_multiple<mode>): New pattern.
14656 (save_multiple_realign<mode>): Likewise.
14657 (restore_multiple<mode>): Likewise.
14658 (restore_multiple_and_return<mode>): Likewise.
14659 (restore_multiple_leave_return<mode>): Likewise.
14660 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14662 2017-05-14 Julia Koval <julia.koval@intel.com>
14664 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14665 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14666 (__builtin_ia32_xsetbv): New builtins.
14667 * config/i386/i386.c (ix86_expand_special_args_builtin):
14669 (ix86_expand_builtin): Special expand for new intrinsics.
14670 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14671 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14672 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14674 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14676 * cfganal.c (inverted_post_order_compute): Change argument type
14678 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14679 * df-core.c (rest_of_handle_df_initialize): Adjust.
14680 (rest_of_handle_df_finish): Likewise.
14681 (df_analyze_1): Likewise.
14682 (df_analyze): Likewise.
14683 (loop_inverted_post_order_compute): Change argument to be a vec *.
14684 (df_analyze_loop): Adjust.
14685 (df_get_n_blocks): Likewise.
14686 (df_get_postorder): Likewise.
14687 * df.h (struct df_d): Change field to be a vec.
14688 * lcm.c (compute_laterin): Adjust.
14689 (compute_available): Likewise.
14690 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14691 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14692 * tree-ssa-pre.c (compute_antic): Likewise.
14694 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14696 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14697 (depth_first_search::depth_first_search): Change structure init
14698 function to this constructor.
14699 (depth_first_search::add_bb): Rename function to this member.
14700 (depth_first_search::execute): Likewise.
14701 (flow_dfs_compute_reverse_finish): Adjust.
14703 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14705 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14706 (longest_simple_path): Likewise.
14707 * shrink-wrap.c (spread_components): Likewise.
14708 (disqualify_problematic_components): Likewise.
14709 (emit_common_heads_for_components): Likewise.
14710 (emit_common_tails_for_components): Likewise.
14711 (insert_prologue_epilogue_for_components): Likewise.
14713 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14715 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14718 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14720 * df-core.c (df_set_blocks): Start using auto_bitmap.
14721 (df_compact_blocks): Likewise.
14722 * df-problems.c (df_rd_confluence_n): Likewise.
14723 * df-scan.c (df_insn_rescan_all): Likewise.
14724 (df_process_deferred_rescans): Likewise.
14725 (df_update_entry_block_defs): Likewise.
14726 (df_update_exit_block_uses): Likewise.
14727 (df_entry_block_bitmap_verify): Likewise.
14728 (df_exit_block_bitmap_verify): Likewise.
14729 (df_scan_verify): Likewise.
14730 * lra-constraints.c (lra_constraints): Likewise.
14731 (undo_optional_reloads): Likewise.
14732 (lra_undo_inheritance): Likewise.
14733 * lra-remat.c (calculate_gen_cands): Likewise.
14734 (do_remat): Likewise.
14735 * lra-spills.c (assign_spill_hard_regs): Likewise.
14736 (spill_pseudos): Likewise.
14737 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14738 (bitmap_set_subtract_values): Likewise.
14740 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14742 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14743 management with auto_bitmap.
14744 (fix_inter_tick): Likewise.
14745 (fix_recovery_deps): Likewise.
14746 * ira.c (add_store_equivs): Likewise.
14747 (find_moveable_pseudos): Likewise.
14748 (split_live_ranges_for_shrink_wrap): Likewise.
14749 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14750 (rtx_reuse_manager::seen_def_p): Likewise.
14751 (rtx_reuse_manager::set_seen_def): Likewise.
14752 * print-rtl.h (class rtx_reuse_manager): Likewise.
14754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14756 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14758 (migrate_btr_def): Likewise.
14759 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14760 * df-core.c (loop_post_order_compute): Likewise.
14761 (loop_inverted_post_order_compute): Likewise.
14762 * hsa-common.h: Likewise.
14763 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14764 * init-regs.c (initialize_uninitialized_regs): Likewise.
14765 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14766 (inline_small_functions): Likewise.
14767 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14768 * ira.c (combine_and_move_insns): Likewise.
14769 (build_insn_chain): Likewise.
14770 * loop-invariant.c (find_invariants): Likewise.
14771 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14772 * predict.c (tree_predict_by_opcode): Likewise.
14773 (predict_paths_leading_to): Likewise.
14774 (predict_paths_leading_to_edge): Likewise.
14775 (estimate_loops_at_level): Likewise.
14776 (estimate_loops): Likewise.
14777 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14778 (spread_components): Likewise.
14779 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14780 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14781 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14782 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14783 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14784 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14785 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14786 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14787 (mark_threaded_blocks): Likewise.
14788 (thread_through_all_blocks): Likewise.
14789 * tree-ssa.c (verify_ssa): Likewise.
14790 (execute_update_addresses_taken): Likewise.
14791 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14793 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14795 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14797 (post_order_compute): Likewise.
14798 (inverted_post_order_compute): Likewise.
14799 (pre_and_rev_post_order_compute_fn): Likewise.
14801 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14803 * genrecog.c (int_set::int_set): Explicitly construct our
14804 auto_vec base class.
14805 * vec.h (auto_vec::auto_vec): New constructor.
14807 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14809 * bitmap.h (class auto_bitmap): New constructor taking
14810 bitmap_obstack * argument.
14812 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14814 * bitmap.h (class auto_bitmap): Change type of m_bits to
14815 bitmap_head, and adjust ctor / dtor and member operators.
14817 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14819 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14820 when returned register mode doesn't match original mode.
14822 2017-05-12 Jeff Law <law@redhat.com>
14823 Jakub Jelinek <jakub@redhat.com>
14825 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14826 we look for cc setter after the compare-elim changes.
14827 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14828 within the vector to match what compare-elim now expects.
14829 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14830 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14832 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14833 after the compare-elim changes.
14834 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14835 the vector to match what compare-elim now expects.
14836 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14837 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14838 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14839 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14840 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14842 * config/visium/visium.c (single_set_and_flags): Fix where
14843 we look for cc setter after the compare-elim changes.
14844 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14845 with the vector to match what compare-elim now expects.
14846 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14847 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14848 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14849 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14850 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14851 (neg<mode>2_insn_set_overflow): Likewise.
14853 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14855 PR middle-end/79794
14856 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14857 maybe_expand_insn call, set ops[0].target. If still set after call,
14858 set alt_rtl. Add extra arg to recursive calls.
14859 (extract_bit_field): Add alt_rtl argument. Pass to
14861 * expmed.h (extract_bit_field): Fix prototype.
14862 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14863 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14864 to extract_bit_field_calls.
14865 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
14866 Pass alt_rtl to extract_bit_field calls.
14867 * calls.c (store_unaligned_arguments_into_psuedos)
14868 load_register_parameters): Pass extra NULL to extract_bit_field calls.
14869 * optabs.c (maybe_legitimize_operand): Clear op->target when call
14871 * optabs.h (struct expand_operand): Add target bitfield.
14873 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14875 * compare-elim.c (try_eliminate_compare): Canonicalize
14876 operation with embedded compare to
14877 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
14878 (set (reg) (operation)].
14880 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
14882 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14885 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
14886 cost of adding a carry flag for ADC instruction.
14887 [case MINUS]: Ignore the cost of subtracting a carry flag
14888 for SBB instruction.
14890 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
14892 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
14894 * config/rs6000/bmiintrin.h: New file.
14895 * config/rs6000/bmi2intrin.h: New file.
14896 * config/rs6000/x86intrin.h: New file.
14898 2017-05-12 Jeff Law <law@redhat.com>
14900 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14903 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14905 PR middle-end/80707
14906 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14908 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14910 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14911 early expansion of vector divide builtins.
14912 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14913 builtins identified as having unsigned arguments.
14915 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14917 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14918 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14919 expansion of vector logical operations (and, andc, or, xor,
14922 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14924 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14925 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14927 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14929 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14930 early GIMPLE expansion of vector multiplies.
14932 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14934 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14935 TARGET_HAVE_MOVT conditional.
14936 (movt splitter): Likewise.
14938 2017-05-12 Richard Biener <rguenther@suse.de>
14940 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14941 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14942 Fold all stmts not inplace.
14944 2017-05-12 Richard Biener <rguenther@suse.de>
14946 PR tree-optimization/80713
14947 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14948 inserted_exprs bit for not removed stmts.
14950 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14952 PR middle-end/69921
14953 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14954 parallelized" attribute for parallelized OpenACC kernels.
14955 * omp-offload.c (execute_oacc_device_lower): Use it.
14957 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14958 Set "oacc kernels" attribute.
14959 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14960 parameter. Adjust all users.
14961 (oacc_fn_attrib_kernels_p): Remove function.
14962 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14963 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14964 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14965 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14966 assert "oacc kernels" attribute is set.
14968 2017-05-11 Carl Love <cel@us.ibm.com>
14970 * config/rs6000/rs6000-c: Add support for built-in functions
14971 vector unsigned char vec_popcnt (vector signed char)
14972 vector unsigned char vec_popcnt (vector unsigned char)
14973 vector unsigned short vec_popcnt (vector signed short)
14974 vector unsigned short vec_popcnt (vector unsigned short)
14975 vector unsigned int vec_popcnt (vector signed int)
14976 vector unsigned int vec_popcnt (vector unsigned int)
14977 vector unsigned long long vec_popcnt (vector signed long long)
14978 vector unsigned long long vec_popcnt (vector unsigned long long)
14979 vector signed long long vec_slo (vector signed long long,
14980 vector signed char)
14981 vector signed long long vec_slo (vector signed long long,
14982 vector unsigned char)
14983 vector unsigned long long vec_slo (vector unsigned long long,
14984 vector signed char)
14985 vector unsigned long long vec_slo (vector unsigned long long,
14986 vector unsigned char)
14987 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14988 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14989 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14990 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14991 * doc/extend.texi: Update the built-in documentation file for the
14992 new built-in functions.
14994 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14996 * attribs.h (sorted_attr_string): Move machine independent
14997 functions for target clone support from the i386 port to common
14998 code. Rename ix86_function_versions to common_function_versions.
14999 Rename make_name to make_unique_name.
15000 (common_function_versions): Likewise.
15001 (make_unique_name): Likewise.
15002 (make_dispatcher_decl): Likewise.
15003 (is_function_default_version): Likewise.
15004 * attribs.c (attr_strcmp): Likewise.
15005 (sorted_attr_string): Likewise.
15006 (common_function_versions): Likewise.
15007 (make_unique_name): Likewise.
15008 (make_dispatcher_decl): Likewise.
15009 (is_function_default_version): Likewise.
15010 * config/i386/i386.c (attr_strcmp): Likewise.
15011 (sorted_attr_string): Likewise.
15012 (ix86_function_versions): Likewise.
15013 (make_name): Likewise.
15014 (make_dispatcher_decl): Likewise.
15015 (is_function_default_version): Likewise.
15016 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15018 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15021 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15022 Account for direct move costs for vec_construct of integer
15025 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15028 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15029 (UNSPEC_STX_ATOMIC): Ditto.
15030 (loaddi_via_sse): New insn.
15031 (storedi_via_sse): Ditto.
15032 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15033 Update corresponding peephole2 patterns.
15034 (atomic_storedi_fpu): Ditto.
15036 2017-05-11 Julia Koval <julia.koval@intel.com>
15038 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15039 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15041 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15042 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15043 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15045 2017-05-11 Nathan Sidwell <nathan@acm.org>
15047 * graphite-poly.c: Include dumpfile.h.
15049 * dumpfle.h (dump_function): Declare here ...
15050 * tree-dump.h (dump_function): ... not here.
15051 * dumpfile.c: #include tree-cfg.h.
15052 (dump_function): Move here from ...
15053 * tree-dump.c (dump_function): ... here.
15054 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15055 * graphite-poly.c: Don't include tree-dump.h.
15056 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15057 * print-tree.c: Likewise.
15058 * stor-layout.c: Likewise.
15059 * tree-nested.c: Likewise.
15061 * dumpfile.c (dump_start): Use TDF_FLAGS.
15062 (dump_enable_all): Fix TDF_KIND check thinko.
15064 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15066 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15067 array entries to represent two legal parameterizations of the
15068 overloaded __builtin_cmpb function, as represented by the
15069 P6_OV_BUILTIN_CMPB constant.
15070 (altivec_resolve_overloaded_builtin): Add special case handling
15071 for the __builtin_cmpb function, as represented by the
15072 P6_OV_BUILTIN_CMPB constant.
15073 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15074 (BU_P6_64BIT_2): New macro.
15075 (BU_P6_OVERLOAD_2): New macro
15076 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15077 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15078 (CMPB): Add overload support to represent both 32-bit and 64-bit
15079 compare-bytes function.
15080 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15081 support for TARGET_CMPB.
15082 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15083 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15084 documentation of the __builtin_cmpb overloaded built-in function.
15086 2017-05-11 Richard Biener <rguenther@suse.de>
15088 PR tree-optimization/80705
15089 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15090 bases are not vectorizable.
15092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15094 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15095 when counting register pressure.
15097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15099 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15100 (iv_ca_more_deps): Renamed to ...
15101 (iv_ca_compare_deps): ... this.
15102 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15104 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15106 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15108 (determine_group_iv_costs): ... here.
15109 (find_inv_vars_cb): Record inv var if it's not recorded before.
15111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15113 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15114 (get_shiftadd_cost): Ditto.
15116 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15118 * tree-ssa-address.c: Include header file.
15119 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15121 (add_to_parts): Refactor.
15122 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15123 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15126 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15128 PR tree-optimization/53090
15129 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15131 (extract_cond_operands): Detect condition with IV on both sides
15132 and return COMP_IV_EXPR_2.
15133 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15134 (rewrite_use_compare): Simplify by removing call to function
15135 extract_cond_operands.
15137 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15139 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15140 (extract_cond_operands): Detect condition comparing against non-
15141 invariant bound and return appropriate enum value.
15142 (find_interesting_uses_cond): Update use of extract_cond_operands.
15143 Handle its return value accordingly.
15144 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15146 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15148 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15149 nonlinear iv_use computation in loop invariant sensitive way.
15151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15153 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15154 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15156 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15158 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15159 (dump_cand): Support iv_cand.inv_exprs.
15160 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15162 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15165 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15167 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15169 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15170 as local function. Include necessary header files.
15171 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15173 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15175 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15179 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15180 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15181 RSHIFT_EXPR and BIT_NOT_EXPR.
15183 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15185 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15186 (adjust_setup_cost): New parameter supporting round up adjustment.
15187 (struct address_cost_data): Delete.
15188 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15189 (split_address_cost, ptr_difference_cost): Delete.
15190 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15191 (struct ainc_cost_data): New struct.
15192 (get_address_cost_ainc): New function.
15193 (get_address_cost, get_computation_cost): Reimplement.
15194 (determine_group_iv_cost_address): Record inv_expr for all uses of
15196 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15197 (iv_ca_has_deps): Reimplemented to ...
15198 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15200 (iv_ca_extend): Call iv_ca_more_deps.
15202 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15204 * tree-ssa-address.c (struct mem_address): Move to header file.
15205 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15206 * tree-ssa-address.h (struct mem_address): Move from C file.
15207 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15209 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15211 * tree-affine.h (aff_combination_type): New interface.
15212 (aff_combination_zero_p): Remove static.
15213 (aff_combination_const_p): New interface.
15214 (aff_combination_singleton_var_p): New interfaces.
15216 2017-05-11 Richard Biener <rguenther@suse.de>
15218 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15219 Skip unreachable blocks and destinations.
15220 (eliminate): Move stmt removal and fixup ...
15221 (fini_eliminate): ... here. Skip inserted exprs.
15222 (pass_pre::execute): Move fini_pre after fini_eliminate.
15223 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15224 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15225 PRE to get rid of dead code that has invalid SSA form and
15226 split critical edges again.
15228 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15230 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15232 2017-05-11 Richard Biener <rguenther@suse.de>
15234 * passes.c (execute_function_todo): Verify loops if they are
15235 said to be up-to-date.
15236 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15237 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15239 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15242 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15243 handle calling assemble_external ourself.
15246 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15247 modes with zero size. Enhance comment.
15249 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15251 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15252 built-ins for vec_xl and vec_xst with short and char pointer
15255 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15257 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15258 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15259 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15260 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15261 (_mm_maskz_min_round_ss): New intrinsics.
15262 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15263 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15264 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15265 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15266 (__builtin_ia32_minss_mask_round): New builtins.
15267 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15268 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15269 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15271 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15272 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15274 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15277 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15279 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15280 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15281 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15282 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15283 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15284 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15285 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15286 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15287 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15288 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15289 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15290 (__builtin_ia32_mulss_mask_round): New builtins.
15291 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15292 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15293 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15295 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15296 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15298 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15301 2017-05-10 Julia Koval <julia.koval@intel.com>
15303 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15304 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15305 (_mm256_setr_m128i): New intrinsics.
15307 2017-05-10 Julia Koval <julia.koval@intel.com>
15309 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15310 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15311 (_mm_maskz_rcp14_ss): New intrinsics.
15312 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15313 (__builtin_ia32_rcp14ss_mask): New builtins.
15314 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15316 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15318 PR tree-optimization/51513
15319 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15320 (assert_unreachable_fallthru_edge_p): Use it.
15321 (group_case_labels_stmt): Likewise.
15322 * tree-cfg.h: Prototype it.
15323 * stmt.c: Include cfghooks.h and tree-cfg.h.
15324 (emit_case_dispatch_table) <gap_label>: New local variable.
15325 Use it to fill dispatch table gaps.
15326 Test for default_label before updating probabilities.
15327 (expand_case) <default_label>: Remove unneeded initialization.
15328 Test for unreachable default case statement and remove its edge.
15329 Set default_label accordingly.
15330 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15332 2017-05-10 Carl Love <cel@us.ibm.com>
15334 * config/rs6000/rs6000-c: Add support for built-in functions
15335 vector signed char vec_neg (vector signed char)
15336 vector signed short int vec_neg (vector short int)
15337 vector signed int vec_neg (vector signed int)
15338 vector signed long long vec_neg (vector signed long long)
15339 vector float vec_neg (vector float)
15340 vector double vec_neg (vector double)
15341 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15343 * config/rs6000/altivec.h: Add define for vec_neg
15344 * doc/extend.texi: Update the built-in documentation for the
15345 new built-in functions.
15347 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15349 PR tree-optimization/77644
15350 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15352 2017-05-10 Nathan Sidwell <nathan@acm.org>
15354 * dumpfile.h (TDI_lang_all): New.
15355 (TDF_KIND): New. Renumber others
15356 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15358 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15360 (get_dump_file_name): Adjust suffix generation.
15361 (dump_enable_all): Use TDF_KIND.
15362 * doc/invoke.texi (-fdump-lang-all): Document.
15364 * dumpfile.h: Tabify.
15366 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15369 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15370 Move member access before delete.
15372 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15374 * tree-inline.c (expand_call_inline): Split block at stmt
15377 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15380 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15381 are now unused after splitting mov{sf,sd}_hardfloat.
15382 (f32_lr2): Likewise.
15383 (f32_lm): Likewise.
15384 (f32_lm2): Likewise.
15385 (f32_li): Likewise.
15386 (f32_li2): Likewise.
15387 (f32_lv): Likewise.
15388 (f32_sr): Likewise.
15389 (f32_sr2): Likewise.
15390 (f32_sm): Likewise.
15391 (f32_sm2): Likewise.
15392 (f32_si): Likewise.
15393 (f32_si2): Likewise.
15394 (f32_sv): Likewise.
15395 (f32_dm): Likewise.
15396 (f32_vsx): Likewise.
15397 (f32_av): Likewise.
15398 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15399 For movsf, order stores so the VSX stores occur before the GPR
15400 store which encourages the register allocator to use a traditional
15401 FPR instead of a GPR. For movsd, order the stores so that the GPR
15402 store comes before the VSX stores to allow the power6 to work.
15403 This is due to the power6 not having a 32-bit integer store
15404 instruction from a FPR.
15405 (movsf_hardfloat): Likewise.
15406 (movsd_hardfloat): Likewise.
15408 2017-05-09 Martin Sebor <msebor@redhat.com>
15410 PR translation/80280
15411 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15414 PR translation/80280
15415 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15416 data member added in r247778.
15417 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15419 2017-05-09 Nathan Sidwell <nathan@acm.org>
15421 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15423 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15426 2017-05-09 Marek Polacek <polacek@redhat.com>
15428 * doc/invoke.texi: Fix typo.
15430 2017-05-09 Richard Biener <rguenther@suse.de>
15432 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15433 (vrp_val_is_min): Likewise.
15434 (set_value_range_to_value): Likewise.
15435 (set_value_range_to_nonnegative): Likewise.
15436 (gimple_assign_nonzero_p): Likewise.
15437 (gimple_stmt_nonzero_p): Likewise.
15438 (vrp_int_const_binop): Likewise. Remove unreachable case.
15439 (adjust_range_with_scev): Adjust comments.
15440 (compare_range_with_value): Likewise.
15441 (extract_range_from_phi_node): Likewise.
15442 (test_for_singularity): Likewise.
15444 2017-05-09 Richard Biener <rguenther@suse.de>
15446 * tree-vrp.c (get_single_symbol): Add assert that we don't
15447 get overflowed constants as invariant part.
15448 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15449 checks. Use wi::cmp instead of recursing for integer constants.
15450 (compare_values): Just ignore whether we assumed undefined
15451 overflow instead of failing the compare.
15452 (extract_range_for_var_from_comparison_expr): Add comment before the
15453 TREE_NO_WARNING sets.
15454 (test_for_singularity): Likewise.
15455 (extract_range_from_comparison): Do not disable optimization
15456 when we assumed undefined overflow.
15457 (extract_range_basic): Remove init of unused var.
15459 2017-05-09 Richard Biener <rguenther@suse.de>
15461 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15462 (extract_range_from_multiplicative_op_1): Adjust.
15463 (extract_range_from_binary_expr_1): Use int_const_binop.
15465 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15468 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15469 rs6000_store_data_bypass_p in seven define_bypass directives and
15470 in several comments.
15471 * config/rs6000/rs6000-protos.h: Add prototype for
15472 rs6000_store_data_bypass_p function.
15473 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15474 function implements slightly different (rs6000-specific) semantics
15475 than store_data_bypass_p, returning false rather than aborting
15476 with assertion error when arguments do not satisfy the
15477 requirements of store data bypass.
15478 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15479 rs6000_store_data_bypass_p.
15481 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15483 * config/xtensa/xtensa-protos.h
15484 (xtensa_initial_elimination_offset): New declaration.
15485 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15486 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15487 macro definition, add case for FRAME_POINTER_REGNUM when
15488 FRAME_GROWS_DOWNWARD.
15489 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15490 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15491 xtensa_initial_elimination_offset.
15493 2017-05-08 Nathan Sidwell <nathan@acm.org>
15495 * doc/invoke.texi: Alphabetize -fdump options.
15497 2017-05-08 Martin Sebor <msebor@redhat.com>
15499 PR translation/80280
15500 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15502 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15504 * target.def (compute_frame_layout): New optional target hook.
15505 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15506 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15507 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15509 * reload1.c (verify_initial_elim_offsets): Likewise.
15510 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15511 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15512 (arm_get_frame_offsets): Split up into this ...
15513 (arm_compute_frame_layout): ... and this function.
15515 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15517 * config/aarch64/constraints.md (Usa): New constraint.
15518 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15520 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15522 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15523 with_multilib_list after it has been checked.
15525 2017-05-08 Richard Biener <rguenther@suse.de>
15527 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15528 (bitmap_set_subtract_values): Likewise.
15530 2017-05-08 Richard Biener <rguenther@suse.de>
15532 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15533 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15535 (gimple_stmt_nonzero_warnv_p): Rename to ...
15536 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15538 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15539 (extract_range_basic): Adjust, do not disable propagation on
15540 strict overflow sensitive simplification.
15541 (vrp_visit_cond_stmt): Likewise.
15543 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15545 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15546 body size unconditionally.
15548 2017-05-07 Jeff Law <law@redhat.com>
15551 2017-05-06 Jeff Law <law@redhat.com>
15552 PR tree-optimization/78496
15553 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15556 PR tree-optimization/78496
15557 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15558 (simplify_stmt_using_ranges): Call it.
15559 (vrp_dom_walker::before_dom_children): Extract equivalences
15560 from an ASSERT_EXPR with an equality comparison against a
15563 2017-05-06 Jeff Law <law@redhat.com>
15565 PR tree-optimization/78496
15566 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15569 PR tree-optimization/78496
15570 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15571 (simplify_stmt_using_ranges): Call it.
15572 (vrp_dom_walker::before_dom_children): Extract equivalences
15573 from an ASSERT_EXPR with an equality comparison against a
15576 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15578 * lra-constraints.c (lra_copy_reg_equiv): New function.
15579 (split_reg): Use it to copy equivalence information from the
15580 original register to the spill register.
15582 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15584 PR rtl-optimization/75964
15585 * simplify-rtx.c (simplify_const_relational_operation): Remove
15586 invalid handling of comparisons of integer ABS.
15588 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15590 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15591 initialize to zero.
15592 (init_regs): Remove declaration.
15593 (function_arg_advance_32): Initialize error_p as boolean variable.
15595 2017-05-05 Nathan Sidwell <nathan@acm.org>
15597 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15598 lines. Use for (;;).
15600 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15602 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15603 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15604 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15605 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15606 VF=2 that require versioning.
15608 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15610 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15613 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15615 * diagnostic.h (diagnostic_override_option_index): Convert from
15616 macro to inline function.
15618 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15620 * diagnostic.c (last_module_changed_p): New function.
15621 (set_last_module): New function.
15622 (diagnostic_report_current_module): Convert macro usage to
15623 the above functions.
15624 * diagnostic.h (diagnostic_context::last_module): Strengthen
15625 from const line_map * to const line_map_ordinary *.
15626 (diagnostic_last_module_changed): Delete macro.
15627 (diagnostic_set_last_module): Delete macro.
15629 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15631 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15632 with diagnostic_report_diagnostic.
15633 (diagnostic_n_impl_richloc): Likewise.
15634 * diagnostic.h (report_diagnostic): Delete macro.
15635 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15636 with diagnostic_report_diagnostic.
15637 * substring-locations.c (format_warning_va): Likewise.
15639 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15641 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15642 save/restor of format_spec. Move option-printing code to...
15643 (print_option_information): ...this new function, and
15644 reimplement by simply printing to the pretty_printer,
15645 rather than appending to the format string.
15647 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15649 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15650 handling logic into...
15651 (update_effective_level_from_pragmas): ...this new function.
15653 2017-05-04 Andrew Waterman <andrew@sifive.com>
15655 * config/riscv/riscv.opt (mstrict-align): New option.
15656 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15657 (SLOW_UNALIGNED_ACCESS): Define.
15658 (riscv_slow_unaligned_access): Declare.
15659 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15661 (riscv_slow_unaligned_access): New variable.
15662 (rocket_tune_info): Set slow_unaligned_access to true.
15663 (optimize_size_tune_info): Set slow_unaligned_access to false.
15664 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15665 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15666 (riscv_option_override): Set riscv_slow_unaligned_access.
15667 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15669 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15671 * config/riscv/riscv.md: Unify indentation.
15673 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15678 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15680 (extendsi<mode>2): Add support for doing sign extension via
15681 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15682 don't have ISA 3.0 instructions.
15683 (extendsi<mode>2 splitter): Likewise.
15684 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15685 generate the normal insns since SImode can now go in vector
15686 registers. Disallow the special UNSPECs needed for previous
15687 machines to hide SImode being used. Add new insns
15688 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15689 (fix_trunc<mode>si2_stfiwx): Likewise.
15690 (fix_trunc<mode>si2_internal): Likewise.
15691 (fixuns_trunc<mode>si2): Likewise.
15692 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15693 (fctiw<u>z_<mode>_smallint): Likewise.
15694 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15695 of floating point to 32-bit integer from doing a direct move to
15696 the GPR registers to do a store.
15697 (fctiwz_<mode>): Break long line.
15699 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15701 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15702 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15703 (addr_list, addr_offset_valid_p): New.
15704 (split_address_groups): Check offset validity with above function.
15705 (gt-tree-ssa-loop-ivopts.h): Include header file.
15707 2017-05-05 Nathan Sidwell <nathan@acm.org>
15709 * config.gcc (arm*-*-*): Add missing 'fi'.
15711 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15713 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15714 included in -fopt-info does not matter.
15715 * doc/optinfo.texi (-fopt-info): Fix description of default
15716 behavour. Explicitly say order of options included in -fopt-info
15719 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15721 * config.gcc: Allow combinations of aprofile and rmprofile values for
15722 --with-multilib-list.
15723 * config/arm/t-multilib: New file.
15724 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15725 variables. Remove setting of ISA and floating-point ABI in
15726 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15727 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15728 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15729 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15731 * config/arm/t-rmprofile: Likewise except for the matches changes.
15732 * doc/install.texi (--with-multilib-list): Document the combination of
15733 aprofile and rmprofile values and warn about pitfalls in doing that.
15735 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15737 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15738 (movdi_aarch64): Likewise.
15740 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15742 PR tree-optimization/80632
15743 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15745 (build_arrays): Initialize it for virtual phis.
15746 (fix_phi_nodes): Use it for virtual phis.
15748 PR tree-optimization/80558
15749 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15750 [x, y] op z into [x op, y op z] for op & or | if conditions
15753 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15754 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15757 * config/arm/arm.md (use_literal_pool): Remove.
15758 (64-bit immediate split): No longer takes cost into consideration
15759 if arm_disable_literal_pool is enabled.
15760 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15761 used when arm_disable_literal_pool is enabled.
15762 (arm_max_const_double_inline_cost): Remove use of
15763 arm_disable_literal_pool.
15764 (push_minipool_fix): Add assert.
15765 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15766 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15767 (no_literal_pool_sf_immediate): New.
15769 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15771 PR tree-optimization/80613
15772 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15773 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15775 2017-05-05 Richard Biener <rguenther@suse.de>
15777 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15779 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15781 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15782 of this flag from insn conditions due to removal from r247495.
15784 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15786 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15788 (arm_early_store_addr_dep_ptr): Likewise.
15789 * config/arm/aarch-common-protos.h
15790 (arm_early_load_addr_dep_ptr): Add prototype.
15791 (arm_early_store_addr_dep_ptr): Likewise.
15792 * config/arm/cortex-a53.md: Add new bypasses.
15794 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15796 * tree.c (next_type_uid): Change type to unsigned.
15797 (type_hash_canon): Decrement back next_type_uid if
15798 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15799 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15802 2017-05-04 Martin Sebor <msebor@redhat.com>
15804 * builtins.c: Fix a trivial typo in a comment.
15806 PR middle-end/79234
15807 * builtins.c (check_sizes): Adjust to handle reading past the end.
15808 Avoid printing excessive upper bound of ranges. Use %E to print
15809 tree nodes instead of converting them to %wu.
15810 (expand_builtin_memchr): New function.
15811 (compute_dest_size): Rename...
15812 (compute_objsize): ...to this.
15813 (expand_builtin_memcpy): Adjust.
15814 (expand_builtin_mempcpy): Adjust.
15815 (expand_builtin_strcat): Adjust.
15816 (expand_builtin_strcpy): Adjust.
15817 (check_strncat_sizes): Adjust.
15818 (expand_builtin_strncat): Adjust.
15819 (expand_builtin_strncpy): Adjust and simplify.
15820 (expand_builtin_memset): Adjust.
15821 (expand_builtin_bzero): Adjust.
15822 (expand_builtin_memcmp): Adjust.
15823 (expand_builtin): Handle memcmp.
15824 (maybe_emit_chk_warning): Check strncat just once.
15826 2017-05-04 Martin Sebor <msebor@redhat.com>
15828 PR preprocessor/79214
15829 PR middle-end/79222
15830 PR middle-end/79223
15831 * builtins.c (check_sizes): Add inlining context and issue
15832 warnings even when -Wno-system-headers is set.
15833 (check_strncat_sizes): Same.
15834 (expand_builtin_strncat): Same.
15835 (expand_builtin_memmove): New function.
15836 (expand_builtin_stpncpy): Same.
15837 (expand_builtin): Handle memmove and stpncpy.
15839 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15841 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15842 which is not used any more.
15844 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15846 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15848 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15850 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15851 (cortexa53_tunings): Likewise.
15852 (cortexa57_tunings): Likewise.
15853 (cortexa72_tunings): Likewise.
15854 (cortexa73_tunings): Likewise.
15856 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15858 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15859 Set loop alignment to 8.
15861 2017-05-04 Martin Sebor <msebor@redhat.com>
15863 PR translation/80280
15864 * builtins.c (expand_builtin_object_size): Add missing quoting to
15865 %D and like directives.
15866 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
15867 (hsa_type_for_tree_type): Same.
15868 (verify_function_arguments): Same.
15869 * symtab.c (symbol_table::change_decl_assembler_name): Same.
15870 * varasm.c (get_section): Same.
15873 2017-05-04 Martin Sebor <msebor@redhat.com>
15875 PR translation/80280
15876 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
15878 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15880 * config/aarch64/aarch64.c (generic_addrcost_table):
15881 Change HI/TI mode setting.
15883 2017-05-04 Martin Jambor <mjambor@suse.cz>
15885 PR tree-optimization/80622
15886 * tree-sra.c (comes_initialized_p): New function.
15887 (build_accesses_from_assign): Only set write lazily when
15888 comes_initialized_p is false.
15889 (analyze_access_subtree): Use comes_initialized_p.
15890 (propagate_subaccesses_across_link): Assert !comes_initialized_p
15891 instead of testing for PARM_DECL.
15893 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15895 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
15896 constraint on operand 0 to allow more general addressing modes.
15897 Adjust output template.
15898 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15900 * config/aarch64/aarch64-protos.h
15901 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15902 * config/aarch64/constraints.md (Dp): New address constraint.
15903 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15906 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15908 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15909 update use of estimate_ipcp_clone_size_and_time.
15910 (estimate_local_effects): Update use of
15911 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15912 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15913 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15914 Return nonspecialized time.
15916 2017-05-04 Richard Biener <rguenther@suse.de>
15918 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15919 for the last VUSE which def dominates the PHI. Directly call
15921 (get_continuation_for_phi_1): Remove.
15923 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15925 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15926 to explain the use of truncating division. Cap the number of
15927 iterations to the maximum given by nb_iterations_upper_bound,
15930 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15932 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15933 * configure: Regenerate.
15934 * config.in: Regenerate.
15935 * config/i386/driver-mingw32.c: new file.
15936 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15937 * config.host: Link driver-mingw32.o on MinGW host.
15938 * doc/install.texi: Document new --enable-mingw-wildcard configure
15941 2017-05-04 Marek Polacek <polacek@redhat.com>
15943 PR tree-optimization/80612
15944 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15946 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15947 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15949 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15950 (movt splitter): Likewise.
15951 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15952 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15953 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15954 block for Thumb-1 with MOVT.
15955 (thumb2_legitimate_address_p): Move code block ...
15956 (can_avoid_literal_pool_for_label_p): ... into this new function.
15957 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15959 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15960 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15961 "M-profile targets with the MOVT instruction".
15963 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15965 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15966 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15967 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15969 2017-05-04 Martin Liska <mliska@suse.cz>
15971 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15972 variable cond_code.
15974 2017-05-04 Richard Biener <rguenther@suse.de>
15976 * tree.c (array_at_struct_end_p): Handle arrays at struct
15977 end with flexarrays more conservatively. Refactor and treat
15978 arrays of arrays or aggregates more strict. Fix
15979 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15980 * tree.c (array_at_struct_end_p): Adjust prototype.
15981 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15982 * gimple-fold.c (get_range_strlen): Likewise.
15983 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15985 2017-05-04 Richard Biener <rguenther@suse.de>
15987 PR tree-optimization/31130
15988 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15990 (supports_overflow_infinity): Likewise.
15991 (is_negative_overflow_infinity): Likewise.
15992 (is_positive_overflow_infinity): Likewise.
15993 (is_overflow_infinity): Likewise.
15994 (stmt_overflow_infinity): Likewise.
15995 (overflow_infinity_range_p): Likewise.
15996 (usable_range_p): Remove as always returning true.
15997 (make_overflow_infinity): Remove.
15998 (negative_overflow_infinity): Likewise.
15999 (positive_overflow_infinity): Likewise.
16000 (avoid_overflow_infinity): Likewise.
16001 (set_value_range): Adjust accordingly.
16002 (set_value_range_to_nonnegative): Likewise, remove now unused
16003 overflow_infinity arg.
16004 (vrp_operand_equal_p): Adjust.
16005 (update_value_range): Likewise.
16006 (range_int_cst_singleton_p): Likewise.
16007 (operand_less_p): Likewise.
16008 (compare_values_warnv): Likewise.
16009 (extract_range_for_var_from_comparison_expr): Likewise.
16010 (vrp_int_const_binop): Likewise.
16011 (zero_nonzero_bits_from_vr): Likewise.
16012 (extract_range_from_multiplicative_op_1): Likewise.
16013 (extract_range_from_binary_expr_1): Likewise.
16014 (extract_range_from_unary_expr): Likewise.
16015 (extract_range_from_comparison): Likewise.
16016 (extract_range_basic): Likewise.
16017 (adjust_range_with_scev): Likewise.
16018 (compare_ranges): Likewise.
16019 (compare_range_with_value): Likewise.
16020 (dump_value_range): Likewise.
16021 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16023 (simplify_cond_using_ranges): Adjust.
16025 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16027 * brig-builtins.def: Added a builtin for class_f64.
16028 * builtin-types.def: Added a builtin type needed by class_f64.
16030 2017-05-03 Jason Merrill <jason@redhat.com>
16032 * timevar.def: Add TV_CONSTEXPR.
16034 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16036 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16038 2017-05-03 Martin Jambor <mjambor@suse.cz>
16040 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16041 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16042 * ipa-cp.c (ipcp_propagate_stage): Do not call
16043 ipa_update_after_lto_read.
16044 * ipa-inline.c (ipa_inline): Likewise.
16046 2017-05-03 Martin Jambor <mjambor@suse.cz>
16048 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16049 tag. Added a default constructor and a destructor.
16050 (ipa_edge_args_sum_t): New class;
16051 (ipa_edge_args_sum): Declare.
16052 (ipa_edge_args_vector): Remove declaration.
16053 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16054 (ipa_free_edge_args_substructures): Remove declaration.
16055 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16056 (ipa_edge_args_info_available_for_edge_p): Likewise.
16057 * ipa-prop.c (ipa_edge_args_vector): Removed.
16058 (edge_removal_hook_holder): Likewise.
16059 (edge_duplication_hook_holder): Likewise.
16060 (ipa_edge_args_sum): New variable.
16061 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16062 ipa_edge_args_vector.
16063 (ipa_free_edge_args_substructures): Likewise.
16064 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16065 ipa_edge_args_vector.
16066 (ipa_edge_removal_hook): Turned into method
16067 ipa_edge_args_sum_t::remove.
16068 (ipa_edge_duplication_hook): Turned into method
16069 ipa_edge_args_sum_t::duplicate.
16070 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16071 registering edge hooks.
16072 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16073 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16074 ipa_edge_args_sum instead of ipa_edge_args_vector.
16075 * ipa-profile.c (ipa_profile): Likewise.
16077 2017-05-03 Martin Jambor <mjambor@suse.cz>
16079 * symbol-summary.h (function_summary): New method exists.
16080 (function_summary::symtab_removal): Deallocate through release.
16081 (call_summary): New class.
16082 (gt_ggc_mx): New overload.
16083 (gt_pch_nx): Likewise.
16084 (gt_pch_nx): Likewise.
16086 2017-05-03 Jeff Law <law@redhat.com>
16088 PR tree-optimization/78496
16089 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16090 from simplify_cond_using_ranges. Split off code to walk
16091 backwards through casts into ...
16092 (simplify_cond_using_ranges_2): New function.
16093 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16094 (execute_vrp): After identifying jump threads, call
16095 simplify_cond_using_ranges_2.
16097 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16100 * ipa-inline.h (inline_summary): Add ctor.
16101 (create_ggc): Do not use ggc_cleared_alloc.
16103 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16104 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16106 * gcc.c (handle_braces): Support escaping in switch matching
16108 * doc/invoke.texi (Spec Files): Document it.
16109 Remove superfluous @code markup in items.
16111 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16113 * diagnostic-show-locus.c (struct column_range): New struct.
16114 (get_affected_columns): New function.
16115 (get_printed_columns): New function.
16116 (struct correction): New struct.
16117 (correction::ensure_capacity): New function.
16118 (correction::ensure_terminated): New function.
16119 (struct line_corrections): New struct.
16120 (line_corrections::~line_corrections): New dtor.
16121 (line_corrections::add_hint): New function.
16122 (layout::print_trailing_fixits): Reimplement in terms of the new
16124 (selftest::test_overlapped_fixit_printing): New function.
16125 (selftest::diagnostic_show_locus_c_tests): Call it.
16127 2017-05-03 Nathan Sidwell <nathan@acm.org>
16129 Canonicalize canonical type hashing
16130 * tree.h (type_hash_canon_hash): Declare.
16131 * tree.c (type_hash_list, attribute_hash_list): Move into
16132 type_hash_canon_hash.
16133 (build_type_attribute_qual_variant): Break out hash code calc into
16134 type_hash_canon_hash.
16135 (type_hash_canon_hash): New. Generic type hash computation.
16136 (build_range_type_1, build_array_type_1, build_function_type,
16137 build_method_type_directly, build_offset_type, build_complex_type,
16138 make_vector_type): Call it.
16140 2017-05-03 Richard Biener <rguenther@suse.de>
16142 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16143 When all DRs have unknown misaligned do not always peel
16144 when there is a store but apply the same costing model as if
16145 there were only loads.
16147 2017-05-03 Richard Biener <rguenther@suse.de>
16150 PR tree-optimization/80492
16151 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16152 compare_base_decls returning dont-know properly.
16154 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16156 * config/arm/iterators.md (CCSI): New mode iterator.
16157 (arch): New mode attribute.
16158 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16159 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16160 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16161 code iterator for success result mode.
16162 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16163 the corresponding new insn generators.
16165 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16168 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16169 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16171 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16173 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16174 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16175 (DDR_A): Wrap DDR argument in brackets.
16176 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16177 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16178 (DDR_REVERSED_P): Likewise.
16180 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16182 PR tree-optimization/79472
16183 * tree-switch-conversion.c (struct switch_conv_info): Add
16184 contiguous_range and default_case_nonstandard fields.
16185 (collect_switch_conv_info): Compute contiguous_range and
16186 default_case_nonstandard fields, don't clear final_bb if
16187 contiguous_range and only the default case doesn't have the required
16189 (check_all_empty_except_final): Set default_case_nonstandard instead
16190 of failing if contiguous_range and the default case doesn't have empty
16192 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16193 and only the default case doesn't have the required constants. Skip
16195 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16196 if default_case_nonstandard.
16197 (build_constructors): Build constant 1 just once. Assert that default
16198 values aren't inserted in between cases if contiguous_range. Skip
16200 (build_arrays): Skip virtual phis.
16201 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16202 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16203 Handle virtual phis.
16204 (gen_inbound_check): Handle default_case_nonstandard case.
16205 (process_switch): Adjust check_final_bb caller. Call
16206 gather_default_values with the first non-default case instead of
16207 default case if default_case_nonstandard.
16209 2017-05-02 Nathan Sidwell <nathan@acm.org>
16211 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16212 check. Fix formatting.
16214 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16216 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16217 errors when comparing specialized and unspecialized times.
16219 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16221 * diagnostic-show-locus.c
16222 (layout::should_print_annotation_line_p): Make private.
16223 (layout::print_annotation_line): Make private.
16224 (layout::annotation_line_showed_range_p): Make private.
16225 (layout::show_ruler): Make private.
16226 (layout::print_source_line): Make private. Pass in line and
16227 line_width, rather than calling location_get_source_line. Drop
16229 (layout::print_leading_fixits): New method.
16230 (layout::print_any_fixits): Rename to...
16231 (layout::print_trailing_fixits): ...this, and make private.
16232 Don't print newline fixits.
16233 (diagnostic_show_locus): Move logic for printing one row into...
16234 (layout::print_line): ...this new function. Move the
16235 location_get_source_line call and error-handling from
16236 print_source_line to here. Call print_leading_fixits, and rename
16237 print_any_fixits to print_trailing_fixits.
16238 (selftest::test_fixit_insert_containing_newline): Update now that
16239 newlines are partially supported.
16240 (selftest::test_fixit_insert_containing_newline_2): New test.
16241 (selftest::test_fixit_replace_containing_newline): Update comments.
16242 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16243 * edit-context.c (class added_line): New class.
16244 (class edited_line): Describe newline handling in comment.
16245 (edited_line::actually_edited_p): New method.
16246 (edited_line::print_content): Delete redundant decl.
16247 (edited_line::m_predecessors): New field.
16248 (edited_file::print_content): Call edited_line::print_content.
16249 (edited_file::print_diff): Update to support newlines.
16250 (edited_file::print_diff_hunk): Likewise.
16251 (edited_file::print_run_of_changed_lines): New function.
16252 (edited_file::print_diff_line): Convert to...
16253 (print_diff_line): ...this.
16254 (edited_file::get_effective_line_count): New function.
16255 (edited_line::edited_line): Initialize new field m_predecessors.
16256 (edited_line::~edited_line): Clean up m_predecessors.
16257 (edited_line::apply_fixit): Handle newlines.
16258 (edited_line::get_effective_line_count): New function.
16259 (edited_line::print_content): New function.
16260 (edited_line::print_diff_lines): New function.
16261 (selftest::test_applying_fixits_insert_containing_newline): New
16263 (selftest::test_applying_fixits_replace_containing_newline): New
16265 (selftest::insert_line): New function.
16266 (selftest::test_applying_fixits_multiple_lines): Add example of
16268 (selftest::edit_context_c_tests): Call the new tests.
16270 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16272 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16273 parameter cand. Update dump information.
16274 (get_computation_cost): Update uses.
16276 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16278 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16279 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16280 (get_computation_at, rewrite_use_address): Update use of
16281 get_computation_aff.
16283 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16285 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16286 (get_computation): Delete.
16287 (get_computation_cost): Implement like get_computation_cost_at.
16288 Use get_computation_at.
16289 (get_computation_cost_at): Delete.
16290 (rewrite_use_nonlinear_expr): Use get_computation_at.
16291 (rewrite_use_compare, remove_unused_ivs): Ditto.
16293 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16295 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16297 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16299 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16300 (ivopts_global_cost_for_size): Rename parameter and update uses.
16301 (iv_ca_recount_cost): Update uses.
16302 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16303 candidates seperately in n_invs and n_cands.
16304 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16306 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16308 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16309 (find_inv_vars_cb): New.
16310 (find_depends): Renamed to ...
16311 (find_inv_vars): ... this.
16312 (add_candidate_1, force_var_cost): Call find_inv_vars.
16313 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16315 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16317 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16318 inv_vars. Add inv_exprs.
16319 (struct iv_cand): Rename depends_on to inv_vars.
16320 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16321 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16322 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16323 (dump_cand): Dump inv_vars.
16324 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16325 (record_invariant, find_depends, add_candidate_1): Ditto.
16326 (set_group_iv_cost, force_var_cost): Ditto.
16327 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16328 (get_computation_cost_at, get_computation_cost): Ditto.
16329 (determine_group_iv_cost_generic): Ditto.
16330 (determine_group_iv_cost_address): Ditto.
16331 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16332 (determine_group_iv_costs): Ditto.
16333 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16334 (iv_ca_set_remove_invariants): Renamed to ...
16335 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16336 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16337 (iv_ca_set_add_invariants): Renamed to ...
16338 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16339 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16340 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16341 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16342 (create_new_ivs): Remove useless dump.
16344 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16346 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16348 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16349 (iv_ca_set_no_cp, create_new_iv): Ditto.
16351 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16353 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16355 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16357 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16358 function tree_check2.
16360 2017-05-02 Martin Liska <mliska@suse.cz>
16362 * doc/gcov.texi: Add missing preposition.
16363 * gcov.c (function_info::function_info): Properly fill up
16364 all member variables.
16366 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16368 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16370 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16372 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16374 2017-05-02 Martin Liska <mliska@suse.cz>
16377 * lto-streamer-in.c (lto_read_tree_1): Remove
16378 LTO_STREAMER_DEBUG.
16379 * lto-streamer.c (struct tree_hash_entry): Likewise.
16380 (struct tree_entry_hasher): Likewise.
16381 (tree_entry_hasher::hash): Likewise.
16382 (tree_entry_hasher::equal): Likewise.
16383 (lto_streamer_init): Likewise.
16384 (lto_orig_address_map): Likewise.
16385 (lto_orig_address_get): Likewise.
16386 (lto_orig_address_remove): Likewise.
16387 * lto-streamer.h: Likewise.
16388 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16389 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16391 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16393 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16394 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16395 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16396 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16397 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16398 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16399 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16400 (mm_maskz_sub_ss): New intrinsics.
16401 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16402 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16403 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16404 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16405 (__builtin_ia32_subss_mask_round): New builtins.
16406 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16407 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16408 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16410 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16411 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16413 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16416 2017-05-02 Martin Jambor <mjambor@suse.cz>
16418 PR tree-optimization/78687
16419 * tree-sra.c (access): New field parent.
16420 (process_subtree_disqualification): New function.
16421 (disqualify_candidate): Call it.
16422 (build_accesses_from_assign): Reset write flag if creating an
16424 (build_access_subtree): Fill in parent field and also prpagate
16425 down grp_write flag.
16426 (create_artificial_child_access): New parameter set_grp_write, set
16427 grp_write to its value.
16428 (propagate_subaccesses_across_link): Also propagate grp_write flag
16430 (propagate_all_subaccesses): Push the closest parent back to work
16431 queue if add_access_to_work_queue returned true.
16433 2017-05-02 Richard Biener <rguenther@suse.de>
16435 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16436 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16437 -fstrict-overflow documentation.
16438 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16439 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16440 flag_strict_overflow.
16441 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16442 * lto-opts.c (lto_write_options): Do not stream it.
16443 * lto-wrapper.c (merge_and_complain): Do not handle it.
16444 * opts.c (default_options_table): Do not set -fstrict-overflow.
16445 (finish_options): Likewise do not clear it when sanitizing.
16446 * simplify-rtx.c (simplify_const_relational_operation): Do not
16447 test flag_strict_overflow.
16449 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16451 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16452 using enabled attribute.
16453 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16454 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16455 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16456 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16457 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16458 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16459 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16460 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16461 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16462 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16464 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16466 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16468 2017-05-02 Richard Biener <rguenther@suse.de>
16470 PR tree-optimization/80591
16472 2017-04-10 Richard Biener <rguenther@suse.de>
16474 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16477 2017-05-02 Richard Biener <rguenther@suse.de>
16479 PR tree-optimization/80549
16480 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16481 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16482 headers if they do not have a preheader.
16484 2017-05-02 Martin Liska <mliska@suse.cz>
16487 * common.opt: Fix typo.
16488 * doc/invoke.texi: Likewise.
16490 2017-05-01 Jan Beulich <jbeulich@suse.com>
16492 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16493 swapping, add (x,x,m,x,n) alternative.
16495 2017-05-01 Nathan Sidwell <nathan@acm.org>
16497 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16498 unnecessary unadjusted_alignment check.
16500 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16503 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16505 * gimplify.c (gimplify_cilk_detach): New function.
16506 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16507 * tree-core.h: Document EXPR_CILK_SPAWN.
16508 * tree.h (EXPR_CILK_SPAWN): Define.
16510 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16512 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16513 to use new fixit_hint representation, using the "replace" logic.
16514 (get_line_span_for_fixit_hint): Likewise.
16515 (layout::print_any_fixits): Likewise.
16516 (selftest::test_one_liner_many_fixits): Rename to...
16517 (selftest::test_one_liner_many_fixits_1): ...this, and update
16518 comment and expected output to reflect that the multiple fix-it
16519 hints are now consolidated into one insertion.
16520 (selftest::test_one_liner_many_fixits_2): New test.
16521 (selftest::test_diagnostic_show_locus_one_liner): Update for
16523 (selftest::test_fixit_consolidation): Update for fix-it API
16525 * diagnostic.c (print_parseable_fixits): Likewise.
16526 * edit-context.c (edited_line::m_line_events): Convert from
16527 auto_vec <line_event *> to auto_vec <line_event>.
16528 (class line_event): Convert from abstract base class to a concrete
16529 class, taking over the role of replace_event.
16530 (class insert_event): Delete.
16531 (class replace_event): Rename to class line_event. Convert to
16533 (edit_context::add_fixits): Reimplement.
16534 (edit_context::apply_insert): Delete.
16535 (edit_context::apply_replace): Rename to...
16536 (edit_context::apply_fixit): ...this. Convert to half-open range.
16537 (edited_file::apply_insert): Delete.
16538 (edited_file::apply_replace): Rename to...
16539 (edited_file::apply_fixit): ...this.
16540 (edited_line::~edited_line): Drop deletion of events.
16541 (edited_line::apply_insert): Delete.
16542 (edited_line::apply_replace): Rename to...
16543 (edited_line::apply_fixit): ...this. Convert to half-open range.
16544 Update for change to type of m_line_events.
16545 * edit-context.h (edit_context::apply_insert): Delete.
16546 (edit_context::apply_replace): Rename to...
16547 (edit_context::apply_fixit): ...this.
16549 2017-05-01 Martin Sebor <msebor@redhat.com>
16551 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16554 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16557 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16558 __get_cpuid_max returns 0.
16559 (__get_cpuid_count): Ditto.
16561 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16563 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16564 replacement expression is another instance of one of its arguments.
16566 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16569 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16570 check for stack push/pop autoinc.
16571 * config/i386/i386.c (ix86_agi_dependent): Return false
16572 if the only reason why modified_in_p returned true is that
16573 addr is SP based and set_insn is a push or pop.
16575 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16577 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16580 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16583 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16584 (account_size_time): Use two predicates - exec_pred and
16586 (evaluate_conditions_for_known_args): Compute both clause and
16588 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16589 (inline_summary_t::duplicate): Update.
16590 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16592 (compute_inline_parameters): Likewise.
16593 (estimate_edge_size_and_time): Update caluclation of time.
16594 (estimate_node_size_and_time): Compute both time and nonspecialized
16596 (estimate_ipcp_clone_size_and_time): Update.
16597 (inline_merge_summary): Update.
16598 (do_estimate_edge_time): Update.
16599 (do_estimate_edge_size): Update.
16600 (do_estimate_edge_hints): Update.
16601 (inline_read_section, inline_write_summary): Stream both new predicates.
16602 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16604 (compute_inlined_call_time): Cleanup.
16605 (big_speedup_p): Update.
16606 (edge_badness): Update.
16607 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16608 (size_time_entry): Replace predicate by exec_predicate and
16609 nonconst_predicate.
16610 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16611 (estimate_edge_time): Return also nonspec_time.
16612 (reset_edge_growth_cache): Update.
16614 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16616 PR rtl-optimization/80491
16617 * ifcvt.c (noce_process_if_block): When looking for x setter
16618 with missing else_bb, don't check only the insn right before
16619 cond_earliest, but look for the last insn that x is modified in
16620 within the same bb.
16622 PR rtl-optimization/80491
16623 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16625 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16627 PR tree-optimization/80487
16628 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16630 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16632 PR tree-optimization/79697
16633 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16634 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16635 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16637 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16638 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16640 2017-04-28 Martin Sebor <msebor@redhat.com>
16642 PR tree-optimization/80523
16643 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16644 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16646 (maybe_warn, format_directive, parse_directive): Use new functions.
16647 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16649 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16651 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16653 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16655 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16656 target_header_dir): Set correctly.
16657 * configure: Regenerated.
16658 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16659 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16660 instead of SYSTEM_HEADER_DIR.
16662 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16664 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16665 (estimate_local_effects): Likewise.
16666 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16667 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16668 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16669 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16670 do_estimate_edge_time, estimate_edge_time): Likewise.
16671 * ipa-inline-analysis.c (estimate_node_size_and_time,
16672 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16673 (estimate_time_after_inlining): Remove.
16675 2017-04-28 Martin Liska <mliska@suse.cz>
16677 * doc/gcov.texi: Enhance documentation of gcov.
16679 2017-04-28 Martin Liska <mliska@suse.cz>
16681 * doc/gcov.texi: Sort options in alphabetic order.
16682 * doc/gcov-dump.texi: Likewise.
16683 * doc/gcov-tool.texi: Likewise.
16684 * gcov.c (print_usage): Likewise.
16685 * gcov-dump.c (print_usage): Likewise.
16686 * gcov-tool.c (print_merge_usage_message): Likewise.
16687 (print_rewrite_usage_message): Likewise.
16688 (print_overlap_usage_message): Likewise.
16690 2017-04-28 Martin Liska <mliska@suse.cz>
16692 PR gcov-profile/53915
16693 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16695 2017-04-28 Martin Liska <mliska@suse.cz>
16697 PR gcov-profile/79891
16698 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16699 is marked by compiler as living on a line.
16700 (get_cycles_count): Remove usage of the union.
16701 (output_intermediate_file): Likewise.
16702 (find_source): Fix GNU coding style.
16703 (accumulate_line_counts): Remove old non-all block mode.
16704 (output_lines): Remove usage of the union.
16705 * profile.c (output_location): Include all BBs, even if
16706 belonging to a same line (and file) as a previous BB.
16708 2017-04-28 Martin Liska <mliska@suse.cz>
16710 * gcov.c (process_args): Handle new argument 'w'.
16711 (read_graph_file): Assign ID to BBs.
16712 (output_branch_count): Display BB # if verbose flag is set.
16713 (output_lines): Likewise for arcs.
16714 (print_usage): Add '--verbose' option help.
16715 * doc/gcov.texi: Document --verbose (-w) option.
16717 2017-04-28 Martin Liska <mliska@suse.cz>
16719 * gcov.c (struct block_location_info): New struct.
16720 (process_file): Fill up the new structure.
16721 (read_graph_file): Replace usage of encoding by the newly added
16723 (add_line_counts): Likewise.
16724 (accumulate_line_counts): Remove usage of the union.
16725 (function_info::function_info): New function.
16726 (function_info::~function_info): Likewise.
16727 (process_file): Call delete instead of release_function.
16728 (release_function): Release the function.
16729 (release_structures): Call delete instead of release_function.
16730 (solve_flow_graph): Replace usage of num_blocks.
16731 (find_exception_blocks): Likewise.
16732 (output_lines): Fix GNU coding style.
16734 2017-04-28 Martin Liska <mliska@suse.cz>
16737 * coverage.c (coverage_remove_note_file): New function.
16738 * coverage.h: Declare the function.
16739 * toplev.c (finalize): Clean if an error has been seen.
16741 2017-04-28 Martin Liska <mliska@suse.cz>
16743 PR gcov-profile/80031
16744 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16745 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16746 * gcov.c (read_graph_file): Read just number of blocks.
16747 * profile.c (branch_prob): Do not stream 0 flags per a basic
16750 2017-04-28 Martin Liska <mliska@suse.cz>
16752 * gcov-dump.c (tag_*): Add new argument to declarations.
16753 (dump_gcov_file): Likewise.
16754 (tag_blocks): Add and use new argument depth.
16755 (tag_arcs): Likewise.
16756 (tag_lines): Likewise.
16757 (tag_counters): Likewise.
16758 (tag_summary): Likewise.
16759 (dump_working_sets): Use depth to do a proper indentation.
16761 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16764 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16765 * symtab.c (symtab_node::debug_symtab): Move definition here.
16767 2017-04-28 Richard Biener <rguenther@suse.de>
16769 * lto-streamer.h (LTO_major_version): Bump to 7.
16771 2017-04-28 Richard Biener <rguenther@suse.de>
16773 * tree-vrp.c (assert_info): New struct.
16774 (add_assert_info): New helper.
16775 (register_edge_assert_for_2): Refactor to add asserts to a vector
16777 (register_edge_assert_for_1): Likewise.
16778 (register_edge_assert_for): Likewise.
16779 (finish_register_edge_assert_for): New helper actually registering
16780 asserts where live on edge.
16781 (find_conditional_asserts): Adjust.
16782 (find_switch_asserts): Likewise.
16783 (evrp_dom_walker::try_find_new_range): Generalize.
16784 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16786 2017-04-27 Marek Polacek <polacek@redhat.com>
16789 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16790 arg10 and arg11 to itype.
16792 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16794 * doc/extend.texi (Object Size Checking): Improve grammar.
16796 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16799 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16800 that the logic for permitting reciprocal estimates matches that
16803 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16806 * tree.c (type_cache_hasher::equal): Only compare
16807 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16808 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16809 non-aggregate element types.
16810 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16811 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16813 2017-04-27 Richard Biener <rguenther@suse.de>
16815 PR middle-end/80533
16816 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16817 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16818 keeping a reference to a trailing array.
16820 2017-04-27 Richard Biener <rguenther@suse.de>
16822 PR middle-end/80539
16823 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16824 being in loop-closed SSA form conservatively.
16825 (chrec_fold_multiply_poly_poly): Likewise.
16827 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16829 PR middle-end/79665
16830 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16831 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16833 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16836 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16837 (aarch64_function_arg_alignment): Return unsigned int again, but still
16838 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16839 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16840 Don't emit -Wpsabi note.
16841 (aarch64_function_arg_boundary): Likewise.
16842 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16845 2017-04-26 Nathan Sidwell <nathan@acm.org>
16847 * tree.h (crc32_unsigned_n): Declare.
16848 (crc32_unsigned, crc32_unsigned): Make inline.
16849 * tree.c (crc32_unsigned_bits): Replace with ...
16850 (crc32_unsigned_n): ... this.
16851 (crc32_unsigned, crc32_byte): Remove.
16852 (crc32_string): Remove unnecessary braces.
16854 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16856 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16857 * ipa-inline-analysis.c (MAX_TIME): Remove.
16858 (account_size_time): Use sreal for time.
16859 (dump_inline_summary): Update.
16860 (estimate_function_body_sizes): Update.
16861 (estimate_edge_size_and_time): Update.
16862 (estimate_calls_size_and_time): Update.
16863 (estimate_node_size_and_time): Update.
16864 (inline_merge_summary): Update.
16865 (inline_update_overall_summary): Update.
16866 (estimate_time_after_inlining): Update.
16867 (inline_read_section): Update.
16868 (inline_write_summary): Update.
16869 * ipa-inline.c (compute_uninlined_call_time): Update.
16870 (compute_inlined_call_time): Update.
16871 (recursive_inlining): Update.
16872 (inline_small_functions): Update.
16873 (dump_overall_stats): Update.
16874 * ipa-inline.h: Include sreal.h.
16875 (size_time_entry): Turn time to sreal.
16876 (inline_summary): Turn self_time nad time to sreal.
16878 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16880 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
16882 (sreal::stream_out, sreal::stream_in): New.
16883 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
16885 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16887 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
16890 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
16893 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
16895 (dimode_scalar_chain::compute_convert_gain): Ditto.
16896 (dimode_scalar_chain::make_vector_copies): Ditto.
16897 (dimode_scalar_chain::convert_reg): Ditto.
16898 (dimode_scalar_chain::convert_insn): Ditto.
16899 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16900 (VI248_AVX512BW_1): New mode iterator.
16901 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16902 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16905 2017-04-25 Martin Sebor <msebor@redhat.com>
16907 PR tree-optimization/80497
16908 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16909 constants are representable in HOST_WIDE_INT.
16910 (parse_directive): Ditto.
16912 2017-04-25 Martin Sebor <msebor@redhat.com>
16915 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16916 (new_zero_array): Adjust signature.
16917 (dom_info::dom_init): Used unsigned rather that size_t.
16918 (dom_info::dom_info): Same.
16920 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16921 Jakub Jelinek <jakub@redhat.com>
16924 * config/arm/arm.c: Include gimple.h.
16925 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16926 returns negative, increment ncrn only if it returned positive.
16927 (arm_needs_doubleword_align): Return int instead of bool,
16928 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16929 members, but if there is any such non-FIELD_DECL
16930 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16931 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16932 returns negative, increment nregs only if it returned positive.
16933 (arm_setup_incoming_varargs): Likewise.
16934 (arm_function_arg_boundary): Emit -Wpsabi note if
16935 arm_needs_doubleword_align returns negative, return
16936 DOUBLEWORD_ALIGNMENT only if it returned positive.
16938 2017-04-25 Marek Polacek <polacek@redhat.com>
16941 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16942 first argument to type.
16944 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16948 type checks to test for compatibility instead of equality.
16950 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16951 Jakub Jelinek <jakub@redhat.com>
16954 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16956 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16957 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16958 the alignment computation, but return their maximum in warn_alignment.
16959 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16960 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16962 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16963 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16966 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16968 * config/arc/simdext.md (dmpyh): Fix typo.
16970 2017-04-25 Richard Biener <rguenther@suse.de>
16972 PR tree-optimization/80492
16973 * alias.c (compare_base_decls): Handle registers with asm
16974 specification conservatively.
16975 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16976 compare_base_decls returning dont-know properly.
16978 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16980 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16981 (legitimate_offset_address_p): New function.
16982 (arc_legitimate_address_p): Use above function.
16984 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16986 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16988 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16990 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16991 ACCH registers whenever they are available.
16993 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16995 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16996 double regs fix when not used.
16998 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17000 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17002 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17003 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17005 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17007 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17008 when emitting short ADD instructions.
17010 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17012 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17014 (cmpsi_cc_c_insn): Likewise.
17015 (cbranchsi4_scratch): Compute proper instruction length using
17016 compact_hreg_operand.
17017 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17019 2017-04-25 Richard Biener <rguenther@suse.de>
17021 PR middle-end/80509
17022 * passes.c (pass_manager::pass_manager): Initialize
17023 m_name_to_pass_map.
17025 2017-04-25 Richard Biener <rguenther@suse.de>
17027 PR tree-optimization/79201
17028 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17030 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17033 * config/s390/vector.md: Split MEM->GPR vector moves for
17034 non-s_operand addresses.
17036 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17039 * config/s390/predicates.md (reload_const_wide_int_operand): New
17041 * config/s390/s390.md ("movti"): Remove d/P alternative.
17042 ("movti_bigconst"): New pattern definition.
17044 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17047 * s390-protos.h (s390_expand_cs_hqi): Removed.
17048 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17049 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17050 modes as well as CCZ1mode and CCZmode.
17051 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17052 signature of s390_emit_compare_and_swap.
17053 (s390_expand_cs_hqi): Likewise, make static.
17054 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17055 compare-and-swap, in some cases.
17056 (s390_expand_cs): Wrapper function.
17057 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17059 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17060 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17061 patterns for small and large integers. Forbid symref memory operands.
17062 Move expander to s390.c. Require cc register.
17063 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17064 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17065 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17066 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17067 symref memory operands. Remove CC mode and call s390_match_ccmode
17069 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17071 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17073 * config/s390/s390.md (define_peephole2): New peephole to help
17074 combining the load-and-test pattern with volatile memory.
17076 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17078 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17079 with CCZmode for TARGET_Z196.
17081 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17083 PR rtl-optimization/80501
17084 * combine.c (make_compound_operation_int): Set subreg_code to SET
17085 even for AND with mask of the sign bit of mode.
17087 PR rtl-optimization/80500
17088 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17089 sum's initial value.
17091 2017-04-25 Julian Brown <julian@codesourcery.com>
17092 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17094 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17096 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17098 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17100 2017-04-25 Julian Brown <julian@codesourcery.com>
17101 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17103 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17104 (thunderx2t99_sha): New Reservation.
17106 2017-04-25 Julian Brown <julian@codesourcery.com>
17107 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17109 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17110 type for 1-element load.
17112 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17114 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17116 2017-04-24 Martin Jambor <mjambor@suse.cz>
17118 PR tree-optimization/80293
17119 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17120 char arrays not totally scalarizable if it is false.
17121 (analyze_all_variable_accesses): Pass correct value in the new
17122 parameter. Add a statistics counter.
17124 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17126 PR middle-end/79931
17127 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17129 2017-04-24 Richard Biener <rguenther@suse.de>
17131 PR tree-optimization/80494
17132 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17133 out for complex types.
17135 2017-04-24 Richard Biener <rguenther@suse.de>
17137 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17138 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17139 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17140 (DFS): Adjust and never fail.
17141 (sccvn_dom_walker::fail): Remove.
17142 (sccvn_dom_walker::before_dom_children): Adjust.
17143 (run_scc_vn): Likewise and never fail.
17144 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17145 (pass_fre::execute): Likewise.
17147 2017-04-24 Richard Biener <rguenther@suse.de>
17149 PR tree-optimization/79725
17150 * tree-ssa-sink.c (statement_sink_location): Return whether
17151 failure reason was zero uses. Move that check later.
17152 (sink_code_in_bb): Deal with zero uses by removing the stmt
17155 2017-04-24 Richard Biener <rguenther@suse.de>
17158 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17159 pointer-based references.
17161 2017-04-24 Richard Biener <rguenther@suse.de>
17164 * pass_manager.h (pass_manager::operator new): Remove.
17165 (pass_manager::operator delete): Likewise.
17166 * passes.c (pass_manager::operator new): Remove.
17167 (pass_manager::operator delete): Likewise.
17168 (pass_manager::pass_manager): Zero individual pass members.
17170 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17173 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17174 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17175 Check "XEXP (src, 1)" operand here.
17176 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17177 Check "XEXP (src, 1)" operand here.
17178 (dimode_scalar_chain::make_vector_copies): Detect count register
17179 of a shift instruction. Zero extend count register from QImode
17180 to DImode to satisfy vector shift pattern count operand predicate.
17181 Substitute vector shift count operand with a DImode copy.
17182 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17185 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17187 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17188 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17189 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17190 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17191 (UNSPEC_NOREX_MEM): Remove definition.
17193 2017-04-21 Richard Biener <rguenther@suse.de>
17195 PR tree-optimization/79547
17196 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17197 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17198 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17199 without any constraints.
17201 2017-04-21 Richard Biener <rguenther@suse.de>
17203 PR tree-optimization/78847
17204 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17206 2017-04-21 Richard Biener <rguenther@suse.de>
17208 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17209 (build_distinct_type_copy): Likewise.
17210 (build_variant_type_copy): Likewise.
17211 * tree.c (build_qualified_type): Pass down mem-stat info.
17212 (build_distinct_type_copy): Likewise.
17213 (build_variant_type_copy): Likewise.
17215 2017-04-21 Richard Biener <rguenther@suse.de>
17217 PR tree-optimization/80237
17218 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17220 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17221 for a simplified result.
17223 2016-04-21 Richard Biener <rguenther@suse.de>
17225 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17226 sth as strict as a simple_iv but a chrec without symbols and an
17227 operand defined in the loop we are peeling (and not some subloop).
17228 (propagate_constants_for_unrolling): Propagate all constants.
17230 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17233 * config/i386/i386.c (print_reg): Remove assert for disalowed
17234 regno values, call output_operand_lossage instead.
17236 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17239 * config/i386/constraints.md (Yc): New register constraint.
17240 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17241 Use Yc constraint for alternative 2 of operand 0. Remove
17242 preferred_for_speed attribute.
17244 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17246 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17247 lastprivate clauses in SIMT case.
17249 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17251 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17253 2017-04-20 Richard Biener <rguenther@suse.de>
17255 PR tree-optimization/57796
17256 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17257 as N scalar stores.
17258 (vect_model_load_cost): Cost gathers as N scalar loads.
17260 2017-04-20 Richard Biener <rguenther@suse.de>
17262 * ggc-page.c (ggc_allocated_p): Rename to ...
17263 (safe_lookup_page_table_entry): ... this and return the lookup
17265 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17267 2017-04-20 Richard Biener <rguenther@suse.de>
17269 PR tree-optimization/80453
17270 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17271 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17272 from the conditions.
17273 (vn_phi_eq): Pass them down.
17274 (vn_phi_lookup): Record them.
17275 (vn_phi_insert): Likewise.
17277 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17279 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17280 uninitialized variable warning to avoid buffer overrun.
17282 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17285 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17286 is suppressed for '{ 0 }' in C.
17288 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17290 * BASE-VER: Set to 8.0.0.
17292 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17294 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17295 priority .init_array and .fini_array section with SECTION_NOTYPE
17298 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17300 PR middle-end/80423
17301 * tree.h (build_array_type): Add typeless_storage default argument.
17302 * tree.c (type_cache_hasher::equal): Also compare
17303 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17304 (build_array_type): Add typeless_storage argument, set
17305 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17307 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17308 (build_array_type): Likewise. Add typeless_storage argument.
17310 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17311 Jakub Jelinek <jakub@redhat.com>
17313 PR tree-optimization/80426
17314 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17315 operation on symbolic operands, also compute the overflow for the
17316 invariant part when the operation degenerates into a negation.
17318 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17321 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17322 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17325 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17327 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17330 * config/avr/avr.c (tree.h): Include it.
17331 (cgraph.h): Include it.
17332 (avr_encode_section_info): Don't warn for uninitialized progmem
17333 variable if it's just an alias.
17335 2017-04-19 Richard Biener <rguenther@suse.de>
17338 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17339 when needed by AutoPGO.
17341 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17344 * doc/lto.texi: Remove an extra 'that'.
17346 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17348 PR rtl-optimization/80429
17349 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17350 are only used in debug insns.
17352 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17353 Vladimir Makarov <vmakarov@redhat.com>
17355 * config/sparc/predicates.md (input_operand): Add comment. Return
17356 true for any memory operand when LRA is in progress.
17357 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17359 2017-04-18 Jeff Law <law@redhat.com>
17362 * mips.md ({return,simple_return}_internal): Do not overwrite
17365 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17367 PR tree-optimization/80443
17368 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17369 instead of adding 1, subtract -1 and similarly instead of subtracting
17372 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17374 PR rtl-optimization/80357
17375 * haifa-sched.c (tmp_bitmap): New variable.
17376 (model_recompute): Handle duplicate use records.
17377 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17378 (free_global_sched_pressure_data): Free it.
17380 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17383 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17384 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17385 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17386 instead of SYSTEM_HEADER_DIR.
17388 2017-04-18 Jeff Law <law@redhat.com>
17390 PR middle-end/80422
17391 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17392 predecessors after walking up the insn chain.
17394 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17397 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17398 sizetype type into debug info.
17400 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17403 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17404 unneeded test for TARGET_UPPER_REGS_SF.
17405 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17407 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17410 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17411 instead of gsi_after_labels.
17413 2017-04-18 Jeff Law <law@redhat.com>
17415 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17419 2017-04-13 Jeff Law <law@redhat.com>
17420 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17421 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17423 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17426 * config/avr/avr.c (intl.h): Include it.
17427 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17429 2017-04-18 Martin Liska <mliska@suse.cz>
17431 PR gcov-profile/78783
17432 * gcov-tool.c (gcov_output_files): Validate that destination
17433 file is either removed by the tool or by a user.
17435 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17436 Guy Benyei <guybe@mellanox.com>
17438 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17439 block, and do not negate it, the stored id is already negative.
17441 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17443 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17445 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17448 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17449 masks of options that should be turned off if the VSX vector
17450 options are turned off.
17451 (OTHER_P8_VECTOR_MASKS): Likewise.
17452 (OTHER_VSX_VECTOR_MASKS): Likewise.
17453 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17454 rs6000_disable_incompatible_switches to validate no type switches
17456 (rs6000_incompatible_switch): New function to disallow turning on
17457 other vector options if -mno-vsx, -mno-power8-vector, or
17458 -mno-power9-vector are specified.
17460 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17462 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17464 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17466 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17467 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17468 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17469 (ARG_POINTER_CFA_OFFSET): Likewise.
17471 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17473 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17474 conditions to take advantage of various optimizations.
17476 2017-04-13 Jeff Law <law@redhat.com>
17478 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17479 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17480 (zero_extendsidi2_dext): Likewise.
17482 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17485 * fold-const.c (fold_ternary_loc): Revert
17486 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17489 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17491 PR rtl-optimization/80343
17492 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17493 new scratch pseudo.
17495 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17498 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17499 to ubsan_encode_value.
17501 2017-04-13 Jeff Law <law@redhat.com>
17503 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17504 appearing in DEBUG_INSNs.
17506 2017-04-13 Martin Liska <mliska@suse.cz>
17508 PR gcov-profile/80413
17509 * gcov-io.c (gcov_write_string): Copy to buffer just when
17510 allocated size is greater than zero.
17512 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17515 * dwarf2out.c (decls_for_scope): Ignore declarations of
17516 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17518 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17521 * ipa-visibility.c (non_local_p): Fix typos.
17522 (localize_node): When localizing symbol in same comdat group,
17523 dissolve the group only when we know external symbols are going
17525 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17527 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17529 PR tree-optimization/79390
17530 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17531 order does not result in usable sequence, retry with reversed operand
17537 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17538 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17539 op0 instead of fold_convert_loc (loc, type, arg0).
17541 2017-04-12 Jeff Law <law@redhat.com>
17543 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17544 has a delay slot in the generated code.
17546 * config/cris/cris.md (cris_preferred_reload_class): Return
17547 GENNONACR_REGS rather than GENERAL_REGS.
17549 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17552 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17553 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17554 signedness of the result type.
17556 2017-04-12 Richard Biener <rguenther@suse.de>
17557 Jeff Law <law@redhat.com>
17559 PR tree-optimization/80359
17560 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17561 trim stores to TARGET_MEM_REFs.
17563 2017-04-12 Richard Biener <rguenther@suse.de>
17565 PR tree-optimization/79390
17566 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17567 threading case even more.
17569 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17572 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17573 for quad_address_p for TImode, instead of just not indexed_address.
17575 2017-04-12 Richard Biener <rguenther@suse.de>
17576 Bernd Edlinger <bernd.edlinger@hotmail.de>
17578 PR middle-end/79671
17579 * alias.c (component_uses_parent_alias_set_from): Handle
17580 TYPE_TYPELESS_STORAGE.
17581 (get_alias_set): Likewise.
17582 * tree-core.h (tree_type_common): Add typeless_storage flag.
17583 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17584 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17585 for types containing members with TYPE_TYPELESS_STORAGE.
17586 (place_field): Likewise.
17587 (layout_type): Likewise for ARRAY_TYPE.
17588 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17589 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17590 TYPE_TYPELESS_STORAGE.
17591 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17593 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17596 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17597 first argument to type.
17599 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17603 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17604 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17605 (rs6000_expand_binop_builtin): Likewise.
17606 (rs6000_expand_ternop_builtin): Likewise; also add missing
17607 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17608 vshasigma built-ins.
17609 * doc/extend.texi: Document that vec_xxpermdi's third argument
17610 must be a constant.
17612 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17614 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17615 Use shift_const cost parameter when calculating gain of STV shifts.
17617 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17619 PR rtl-optimization/70478
17620 * lra-constraints.c (process_alt_operands): Check memory for
17621 disfavoring memory insn operand.
17623 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17625 PR middle-end/80100
17626 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17627 left shift in unsigned HOST_WIDE_INT type.
17629 PR rtl-optimization/80385
17630 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17631 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17634 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17635 if they have any depend clauses.
17637 2017-04-11 Martin Liska <mliska@suse.cz>
17640 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17641 * ipa-split.c (split_function): Create a local comdat symbol
17642 if caller is in a comdat group.
17644 2017-04-11 Martin Liska <mliska@suse.cz>
17647 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17650 2017-04-11 Martin Sebor <msebor@redhat.com>
17652 PR middle-end/80364
17653 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17654 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17656 (directive::set_width, directive::set_precision, format_character):
17658 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17661 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17664 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17665 conflict, set target->arch_name instead of target->cpu_name.
17667 2017-04-11 Richard Biener <rguenther@suse.de>
17669 PR tree-optimization/80374
17670 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17671 build_zero_cst, remove fold_convertible_p check again.
17673 2017-04-11 Martin Liska <mliska@suse.cz>
17676 * ubsan.c (instrument_object_size): Do not instrument register
17679 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17682 * config/i386/i386-builtin-types.def
17683 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17684 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17685 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17686 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17687 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17688 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17689 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17690 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17691 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17692 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17693 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17694 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17695 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17696 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17697 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17698 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17699 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17700 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17701 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17702 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17703 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17704 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17705 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17706 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17707 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17708 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17709 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17710 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17711 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17712 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17713 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17714 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17715 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17716 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17717 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17718 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17719 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17720 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17721 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17722 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17723 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17724 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17725 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17726 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17727 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17728 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17730 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17731 flag to second_arg_count, handle 4 argument function type _COUNT
17732 aliases, handle second_arg_count on second argument rather than last.
17734 2017-04-10 Jeff Law <law@redhat.com>
17736 PR tree-optimization/80374
17737 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17738 record anything if we can not convert integer_zero_node to the
17741 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17744 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17745 Enhance special handling given to the TARGET_P9_MINMAX option in
17746 relation to certain other options.
17748 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17750 PR tree-optimization/80153
17751 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17752 remove POINTER_PLUS_EXPR's base part directly, rather than through
17755 2017-04-10 Richard Biener <rguenther@suse.de>
17756 Bin Cheng <bin.cheng@arm.com>
17758 PR tree-optimization/80153
17759 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17760 the first element of pointer type aff_tree. Build result expr in
17762 (add_elt_to_tree): Convert to type unconditionally. Remove other
17763 fold_convert calls.
17764 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17765 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17767 2017-04-10 Richard Biener <rguenther@suse.de>
17769 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17772 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17774 PR rtl-optimization/70478
17775 * lra-constraints.c (curr_small_class_check): New.
17776 (update_and_check_small_class_inputs): New.
17777 (process_alt_operands): Update curr_small_class_check. Disfavor
17778 alternative insn memory operands. Check available regs for small
17781 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17784 * config/mips/mips.opt (-mvirt): Update description.
17785 * doc/invoke.texi (-mvirt): Likewise.
17787 2017-04-10 Richard Biener <rguenther@suse.de>
17789 PR middle-end/80362
17790 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17791 looking for NEGATE_EXPR in -A / -B to A / B folding.
17793 2017-04-10 Martin Liska <mliska@suse.cz>
17795 PR gcov-profile/80224
17796 * gcov.c (print_usage): Fix usage string.
17797 (get_gcov_intermediate_filename): Remove.
17798 (output_gcov_file): Use both for normal and intermediate format.
17799 (generate_results): Do not initialize special file for
17800 intermediate format.
17802 2017-04-10 Richard Biener <rguenther@suse.de>
17804 PR tree-optimization/80304
17805 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17808 2017-04-10 Nathan Sidwell <nathan@acm.org>
17811 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17812 (rs6000_init_builtins): Use it.
17814 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17816 * config/arm/arm.md (<mrc>): Add mode to SET source.
17817 (<mrrc>): Likewise.
17819 2017-04-10 Richard Biener <rguenther@suse.de>
17821 PR middle-end/80344
17822 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17824 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17827 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17828 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17829 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17830 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17831 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17832 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17833 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17834 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17835 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17836 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17837 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17838 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17839 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17840 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17841 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17842 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17843 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17844 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17845 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17846 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17847 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17848 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17849 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17851 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17853 PR rtl-optimization/70478
17854 * lra-constraints.c: Reverse the last patch.
17856 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17858 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17859 Add comment for WCHAR_T.
17861 2017-04-08 Martin Liska <mliska@suse.cz>
17864 2017-04-07 Martin Liska <mliska@suse.cz>
17867 * ipa-split.c (split_function): Add function part to a same comdat
17870 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17873 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
17875 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
17877 * rs6000/rs6000.c (vec_load_pendulum): Rename...
17878 (vec_pairing): ...to this.
17879 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
17880 (rs6000_sched_init): Adjust for name change.
17881 (struct rs6000_sched_context): Likewise.
17882 (rs6000_init_sched_context): Likewise.
17883 (rs6000_set_sched_context): Likewise.
17885 2017-04-07 Jakub Jelinek <jakub@redhat.com>
17891 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
17893 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
17894 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
17895 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
17897 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17899 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17901 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17903 PR rtl-optimization/70703
17904 * ira-color.c (update_conflict_hard_regno_costs): Use
17905 int64_t instead of HOST_WIDE_INT.
17907 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17909 PR rtl-optimization/70478
17910 * lra-constraints.c (process_alt_operands): Disfavor alternative
17911 insn memory operands.
17913 2017-04-07 Jeff Law <law@redhat.com>
17915 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17916 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17918 2017-04-07 Martin Liska <mliska@suse.cz>
17921 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17922 Show error message instead of an ICE.
17924 2017-04-07 Martin Liska <mliska@suse.cz>
17927 * ipa-split.c (split_function): Add function part to a same comdat
17930 2017-04-07 Richard Biener <rguenther@suse.de>
17932 PR middle-end/80341
17933 * tree.c (get_unwidened): Also handle ! for_type case for
17935 * convert.c (do_narrow): Split out from ...
17936 (convert_to_integer_1): ... here. Do not pass final truncation
17937 type to get_unwidened for TRUNC_DIV_EXPR.
17939 2017-04-07 Richard Biener <rguenther@suse.de>
17941 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17943 (aff_combination_const): Adjust.
17944 (aff_combination_scale): Likewise.
17945 (aff_combination_add_elt): Likewise.
17946 (aff_combination_add_cst): Likewise.
17947 (aff_combination_convert): Likewise.
17948 (add_elt_to_tree): Likewise. Remove unused argument.
17949 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17951 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17953 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17955 * config/arm/arm.c (arm_default_short_enums): Use
17956 ARM_DEFAULT_SHORT_ENUMS.
17957 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17959 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17962 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17963 members with redundant out-of-class redeclaration.
17965 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17968 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17969 * config/i386/i386.md (*zero_extendsidi2):
17970 Add (?*x,*x) and (?*v,*v) alternatives.
17972 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17975 * config/i386/i386.c (ix86_expand_builtin)
17976 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17977 mode from insn data. Convert operands to insn operand mode.
17978 Copy operands that don't satisfy insn predicate to a register.
17980 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17982 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17985 2017-04-06 Richard Biener <rguenther@suse.de>
17987 PR tree-optimization/80334
17988 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17989 preserve alignment of accesses.
17991 2017-04-06 Richard Biener <rguenther@suse.de>
17993 PR tree-optimization/80262
17994 * tree-sra.c (build_ref_for_offset): Preserve address-space
17996 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17997 Drop useless address-space information on MEM_REF offsets.
17999 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18001 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18003 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18005 PR rtl-optimization/70703
18006 * ira-color.c (update_conflict_hard_regno_costs): Use
18007 HOST_WIDE_INT instead of long.
18009 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18012 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18013 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18015 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18016 for x86_64 target. Handle -m3dnowa option.
18018 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18020 PR rtl-optimization/70703
18021 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18022 (update_conflict_hard_regno_costs): Use long instead of unsigned
18023 arithmetic for cost calculation.
18025 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18026 Bernd Edlinger <bernd.edlinger@hotmail.de>
18029 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18032 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18035 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18036 ptr_mode with Pmode throughout.
18037 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18038 into probe_stack_range and use DImode.
18040 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18043 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18044 call_eh_return is true.
18046 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18048 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18049 Initialize last_match_fntype_index.
18051 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18054 * tree-nvr.c: Include internal-fn.h.
18055 (pass_return_slot::execute): Ignore internal calls without
18058 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18059 Richard Biener <rguenther@suse.de>
18062 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18063 captures used multiple times, except for the last use.
18064 * generic-match-head.c: Include gimplify.h.
18066 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18068 PR tree-optimization/79390
18069 * target.h (struct noce_if_info): Declare.
18070 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18071 * target.def (noce_conversion_profitable_p): New target hook.
18072 * ifcvt.h (struct noce_if_info): New type, moved from ...
18073 * ifcvt.c (struct noce_if_info): ... here.
18074 (noce_conversion_profitable_p): Renamed to ...
18075 (default_noce_conversion_profitable_p): ... this. No longer
18077 (noce_try_store_flag_constants, noce_try_addcc,
18078 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18079 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18080 instead of noce_conversion_profitable_p.
18081 * config/i386/i386.c: Include ifcvt.h.
18082 (ix86_option_override_internal): Don't override
18083 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18084 (ix86_noce_conversion_profitable_p): New function.
18085 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18086 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18087 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18088 * doc/tm.texi: Regenerated.
18090 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18092 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18095 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18098 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18099 instructions for small multiply cores.
18101 2017-04-04 Jeff Law <law@redhat.com>
18103 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18105 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18106 that are not set by the loop over the elements.
18108 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18111 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18112 int mode, convert_modes it to mode as unsigned, otherwise use
18113 lowpart_subreg to mode rather than SImode.
18114 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18115 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18116 Use DImode instead of SImode for the shift count operand.
18117 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18120 2017-04-04 Richard Biener <rguenther@suse.de>
18122 PR middle-end/80281
18123 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18124 arithmetic done for the negate or the plus. Simplify.
18125 (A - (-B) -> A + B): Likewise.
18126 * fold-const.c (split_tree): Make sure to not negate pointers.
18128 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18130 PR rtl-optimization/60818
18131 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18132 a compare of comparisons with the thing compared if this results
18133 in a different machine mode.
18135 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18137 * alias.c (base_alias_check): Fix typo in comment.
18138 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18139 * cgraphunit.c (symbol_table::compile): Likewise.
18140 * collect2.c (maybe_run_lto_and_relink): Likewise.
18141 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18142 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18143 * config/avr/avr.c (avr_map_op_t): Likewise.
18144 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18145 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18146 * config/epiphany/epiphany.md (movcc): Likewise.
18147 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18148 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18150 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18151 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18152 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18153 * config/sh/sh.c (sh_rtx_costs): Likewise.
18154 * fold-const.c (fold_truth_andor): Likewise.
18155 * genautomata.c (collapse_flag): Likewise.
18156 * gengtype.h (struct type::u::s): Likewise.
18157 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18158 * input.c (FORMAT_AMOUNT): Likewise.
18159 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18160 (known_aggs_to_agg_replacement_list): Likewise.
18161 * ipa-inline-analysis.c: Likewise.
18162 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18163 * ipa-polymorphic-call.c
18164 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18165 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18166 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18168 * modulo-sched.c (apply_reg_moves): Likewise.
18169 * omp-expand.c (build_omp_regions_1): Likewise.
18170 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18171 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18172 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18173 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18174 * value-prof.c: Likewise.
18175 * var-tracking.c (val_reset): Likewise.
18177 2017-04-03 Richard Biener <rguenther@suse.de>
18179 PR tree-optimization/80275
18180 * fold-const.c (split_address_to_core_and_offset): Handle
18183 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18185 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18186 descriptors is at least equal to that of functions.
18188 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18190 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18192 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18195 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18196 (mov<IMOD4:mode>): New expander.
18197 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18199 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18201 PR rtl-optimization/79405
18202 * fwprop.c (propagations_left): New variable.
18203 (forward_propagate_into): Decrement it.
18204 (fwprop_init): Initialize it.
18205 (fw_prop): If the variable has reached zero, stop propagating.
18206 (fwprop_addr): Ditto.
18208 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18211 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18212 a FUNCTION_DECL, pass it as decl instead of origin to
18215 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18217 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18220 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18223 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18224 TARGET_VSX_SMALL_INTEGER.
18226 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18228 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18229 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18231 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18233 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18234 extraction from odd-numbered MSA register.
18236 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18238 PR middle-end/80173
18239 * expmed.c (store_bit_field_1): Don't attempt to create
18240 a word subreg out of hard registers wider than word if they
18241 have HARD_REGNO_NREGS of 1 for their mode.
18243 PR middle-end/80163
18244 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18245 conversions to integer types wider than word and pointer.
18248 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18249 (rtx_equal_for_cselib_p): Pass 0 to it.
18250 * cselib.c (cselib_hasher::equal): Likewise.
18251 (rtx_equal_for_cselib_1): Add depth argument. If depth
18252 is 128, don't look up VALUE locs and punt. Increment
18253 depth in recursive calls when walking VALUE locs.
18255 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18257 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18258 (make_gcov_file_name): Use the canonical path name for generating
18260 (read_line): Fix handling of files with ascii null bytes.
18262 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18264 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18265 to initialise a vector register instead
18266 of using a const_int.
18268 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18270 PR translation/80189
18271 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18272 diagnostic messages.
18274 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18277 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18278 (dfp_diex_<mode>): Update mode of operand 1.
18279 * doc/extend.texi (dxex, dxexq): Document change to return type.
18280 (diex, diexq): Document change to argument type.
18282 2017-03-30 Martin Jambor <mjambor@suse.cz>
18285 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18286 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18287 it reflects the signature changes performed at the callee side.
18288 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18289 to cgraph_build_function_type_skip_args.
18290 (build_function_decl_skip_args): Adjust call to the above function.
18292 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18295 * config/i386/sse.md
18296 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18297 register as dest whenever it is a MEM not rtx_equal_p to the
18298 corresponding dup operand, and when forcing into reg move the
18299 reg into the memory afterwards.
18300 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18301 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18302 for the force_reg mode.
18303 (avx512vl_vextractf128<mode>): Use register as dest either
18304 always when a MEM, or when it is a MEM not rtx_equal_p to the
18305 corresponding dup operand, or even not when it is a CONST_VECTOR
18306 depending on the mode and lo vs. hi.
18307 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18309 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18310 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18311 Likewise. Require that operands[2] is even.
18312 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18313 Remove extraneous parens. Require that operands[2] is a multiple
18315 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18316 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18317 disallow memory then.
18319 2017-03-30 Richard Biener <rguenther@suse.de>
18321 PR tree-optimization/77498
18322 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18323 to non-constants over backedges.
18325 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18327 PR rtl-optimization/80233
18328 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18329 as last_combined_insn. Do not test for BARRIER_P separately.
18331 2017-03-29 Andreas Schwab <schwab@suse.de>
18334 * calls.c (prepare_call_address): Convert funexp to Pmode before
18335 copying to temp reg.
18337 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18339 PR tree-optimization/80158
18340 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18341 Handle possible future case of more than one alternate
18343 (replace_rhs_if_not_dup): Likewise.
18344 (replace_one_candidate): Likewise.
18346 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18348 PR rtl-optimization/80193
18349 * ira.c (ira): Do not check allocation for LRA.
18351 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18353 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18354 (nvptx_output_simt_exit): Declare.
18355 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18356 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18357 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18358 (nvptx_declare_function_name): ...here. Emit declaration of local
18359 memory space buffer for omp_simt_enter insn.
18360 (nvptx_output_unisimt_switch): New.
18361 (nvptx_output_softstack_switch): New.
18362 (nvptx_output_simt_enter): New.
18363 (nvptx_output_simt_exit): New.
18364 * config/nvptx/nvptx.h (struct machine_function): New fields
18365 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18366 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18367 (UNSPECV_SIMT_EXIT): Ditto.
18368 (omp_simt_enter_insn): New insn.
18369 (omp_simt_enter): New expansion.
18370 (omp_simt_exit): New insn.
18371 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18373 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18374 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18375 (expand_GOMP_SIMT_EXIT): New.
18376 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18377 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18378 (GOMP_SIMT_EXIT): Ditto.
18379 * target-insns.def (omp_simt_enter): New insn.
18380 (omp_simt_exit): Ditto.
18381 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18383 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18384 (lower_rec_input_clauses): Likewise.
18385 (lower_lastprivate_clauses): Handle SIMT privatization.
18387 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18388 (ompdevlow_adjust_simt_enter): New.
18389 (find_simtpriv_var_op): New.
18390 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18391 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18393 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18394 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18395 (copy_decl_for_dup_finish): Ditto.
18397 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18399 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18402 * config/i386/i386.c (ix86_option_override_internal): Always
18403 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18405 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18407 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18409 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18411 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18412 mark new edge's irreducible flag accordign to it.
18413 (vect_do_peeling): Check loop preheader edge's irreducible flag
18414 and pass it to function slpeel_add_loop_guard.
18416 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18418 PR tree-optimization/80218
18419 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18420 Update block frequencies and counts.
18422 2017-03-28 Richard Biener <rguenther@suse.de>
18424 PR tree-optimization/78644
18425 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18426 of a simplification result we may not use it at all.
18428 2017-03-28 Richard Biener <rguenther@suse.de>
18431 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18432 without arguments, generate default definition of a SSA name.
18434 2017-03-28 Richard Biener <rguenther@suse.de>
18436 PR middle-end/80222
18437 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18438 TYPE_REF_CAN_ALIAS_ALL references.
18439 * fold-const.c (fold_indirect_ref_1): Likewise.
18441 2017-03-28 Martin Liska <mliska@suse.cz>
18444 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18445 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18447 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18448 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18450 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18451 (EXTRA_SPECS): Define.
18452 (SUBTARGET_EXTRA_SPECS): Likewise.
18453 (SUBTARGET_CPP_SPEC): Likewise.
18454 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18455 SUBTARGET_EXTRA_SPECS.
18456 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18458 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18460 * config/arc/simdext.md (vst64_insn): Update pattern.
18461 (vld32wh_insn): Likewise.
18462 (vld32wl_insn): Likewise.
18463 (vld64_insn): Likewise.
18464 (vld32_insn): Likewise.
18466 2017-03-28 Marek Polacek <polacek@redhat.com>
18469 * fold-const.c (fold_comparison): Use protected_set_expr_location
18470 instead of SET_EXPR_LOCATION.
18472 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18474 * tree.c (add_expr): Avoid name lookup warning.
18476 2017-03-27 Jeff Law <law@redhat.com>
18478 PR tree-optimization/80216
18479 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18480 function name. Limit recursion depth.
18481 (record_temporary_equivalences): Corresponding changes.
18483 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18485 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18488 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18491 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18493 * cfgcleanup.c (reg_note_cfa_p): New array.
18494 (insns_have_identical_cfa_notes): New function.
18495 (old_insns_match_p): Don't cross-jump in between /f
18496 and non-/f instructions. If both i1 and i2 are frame related,
18497 verify all CFA notes, their order and content.
18499 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18502 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18503 HImode and SImode with zero extend to DImode to one insn.
18504 (bswap<mode>2_extenddi): Likewise.
18505 (bswapsi2_extenddi): Likewise.
18506 (bswaphi2_extendsi): Likewise.
18507 (bswaphi2): Combine bswap HImode and SImode into one insn.
18508 Separate memory insns from swapping register.
18509 (bswapsi2): Likewise.
18510 (bswap<mode>2): Likewise.
18511 (bswaphi2_internal): Delete, no longer used.
18512 (bswapsi2_internal): Likewise.
18513 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18514 store, and gpr<-gpr swap insns.
18515 (bswap<mode>2_store): Likewise.
18516 (bswaphi2_reg): Register only splitter, combine with the splitter.
18517 (bswaphi2 splitter): Likewise.
18518 (bswapsi2_reg): Likewise.
18519 (bswapsi2 splitter): Likewise.
18520 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18521 the insns into load, store, and register/register insns.
18522 (bswapdi2_ldbrx): Likewise.
18523 (bswapdi2_load): Likewise.
18524 (bswapdi2_store): Likewise.
18525 (bswapdi2_reg): Likewise.
18527 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18529 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18530 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18532 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18535 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18537 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18538 special handling for target option conflicts between dform
18539 options (-mpower9-dform, -mpower9-dform-vector,
18540 -mpower9-dform-scalar) and -mno-direct-move.
18542 2017-03-27 Richard Biener <rguenther@suse.de>
18544 PR tree-optimization/80181
18545 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18547 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18549 * config/arc/predicates.md (move_double_src_operand): Replace the
18550 call to move_double_src_operand with a call to address_operand.
18552 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18554 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18555 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18556 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18558 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18560 * config/arc/predicates.md (long_immediate_loadstore_operand):
18561 Consider scaled addresses cases.
18563 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18565 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18566 restored when in interrupt.
18567 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18568 doesn't have delay slot.
18570 2017-03-27 Richard Biener <rguenther@suse.de>
18573 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18574 inlined thunk clones.
18576 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18579 * asan.c (instrument_derefs): Copy over last operand from
18580 original COMPONENT_REF to the new COMPONENT_REF with
18581 DECL_BIT_FIELD_REPRESENTATIVE.
18582 * ubsan.c (instrument_object_size): Likewise.
18584 2017-03-27 Richard Biener <rguenther@suse.de>
18586 PR tree-optimization/80170
18587 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18588 sure DR/SCEV didnt fold in constants we do not see when looking
18589 at the reference base alignment.
18591 2017-03-27 Richard Biener <rguenther@suse.de>
18593 PR middle-end/80171
18594 * gimple-fold.c (fold_ctor_reference): Properly guard against
18595 NULL return value from canonicalize_constructor_val.
18597 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18600 * config/i386/i386.c (ix86_expand_builtin)
18601 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18602 flags reg setting and flags reg using instructions.
18603 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18604 clobbering instructions to zero extend op2.
18606 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18608 * doc/install.texi (Configuration) <--with-aix-soname>:
18609 Update link to AIX ld.
18611 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18613 PR rtl-optimization/80160
18614 PR rtl-optimization/80159
18615 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18616 reg_alternate_class into account.
18618 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18621 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18622 to consider in curr_insn_transform.
18624 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18626 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18627 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18628 and emit_mode_inner.
18630 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18632 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18633 argument to the overloaded builtin variants. Use the new flag to
18634 deprecate certain builtin variants.
18635 * config/s390/s390-builtin-types.def: Add new builtin types.
18636 * config/s390/s390-builtins.h: Support new flags field for
18637 overloaded builtins.
18638 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18639 (s390_macro_to_expand): Enable vector float data type.
18640 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18641 builtins by incrementing the __VEC__ version number.
18642 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18644 (s390_resolve_overloaded_builtin): Emit error messages depending
18645 on the builtin flags.
18646 * config/s390/s390.c (s390_expand_builtin): Support additional
18647 flags argument. Change error message to match the messages
18648 emitted in s390-c.c.
18649 * config/s390/s390.md: New UNSPEC_* constants.
18650 (op_type): Add new instruction types.
18651 * config/s390/vecintrin.h: Add new builtins and test data class
18653 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18654 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18655 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18656 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18657 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18658 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18660 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18661 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18662 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18663 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18665 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18666 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18667 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18668 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18670 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18671 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18672 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18673 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18674 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18675 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18676 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18678 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18679 ("vec_scatter_element<V_HW_4:mode>_DI")
18680 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18681 ("vec_fpint<mode>", "vflls")
18682 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18683 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18684 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18685 ("*vec_cmphe<mode>_cc"): ... these.
18687 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18688 mode constant instead of magic value.
18690 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18692 * config/s390/s390.c (s390_expand_vec_compare): Support other
18693 vector floating point modes than just V2DF.
18694 (s390_expand_vcond): Likewise.
18695 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18696 (s390_cannot_change_mode_class): Prevent mode changes between TF
18697 and V1TF in vector registers.
18698 * config/s390/s390.md (DF, SF): New mode attributes.
18699 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18700 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18701 SFmode support for VRs.
18702 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18704 (VFT, VF_HW): New mode iterators.
18705 (vw, sdx): New mode attributes.
18706 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18707 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18708 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18709 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18710 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18711 also the new vector floating point modes. Renaming to ...
18713 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18714 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18715 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18716 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18717 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18718 ("vec_unordered<mode>"): ... these.
18720 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18721 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18722 ("*vec_extendv2df"): New insn definitions.
18724 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18726 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18727 ("mulditi3_2", "*muldi3_sign"): New patterns.
18728 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18729 rename the pattern definition.
18731 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18733 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18735 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18737 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18739 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18740 instruction if possible.
18741 * config/s390/vector.md (vec_halfnumelts): New mode
18743 ("*vec_vllezlf<mode>"): New pattern.
18745 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18747 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18748 ("popcountv4si2", "popcountv2di2"): Rename to ...
18749 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18750 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18752 ("popcount<mode>2_vxe"): New pattern.
18754 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18756 * common/config/s390/s390-common.c (processor_flags_table): Add
18758 * config.gcc: Add arch12.
18759 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18760 Default to arch12 for unknown CPU model numbers.
18761 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18762 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18763 PROCESSOR_max sanity check.
18764 * config/s390/s390-opts.h (enum processor_type): Add
18766 * config/s390/s390.c (processor_table): Add arch12.
18767 (s390_expand_builtin): Add check for B_VXE flag.
18768 (s390_issue_rate): Add PROCESSOR_ARCH12.
18769 (s390_get_sched_attrmask): Likewise.
18770 (s390_get_unit_mask): Likewise.
18771 (s390_sched_score): Enable z13 scheduling for arch12.
18772 (s390_sched_reorder): Likewise.
18773 (s390_sched_variable_issue): Likewise.
18774 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18776 (s390_tune_attr): Use z13 scheduling also for arch12.
18777 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18778 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18779 (TARGET_VXE_P): New macros.
18780 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18781 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18782 * config/s390/s390.opt: Add arch12 as processor_type.
18784 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18786 * config/s390/s390.md
18787 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18788 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18789 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18791 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18792 Rename expanders to ...
18794 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18795 ("fixuns_truncdddi2_emu"): ... these.
18797 ("fixuns_trunc<mode>si2_emu"): New expander.
18799 ("*fixuns_truncdfdi2_z13"): Rename to ...
18800 ("*fixuns_truncdfdi2_vx"): ... this.
18802 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18804 * config/s390/2964.md: Remove the single element vector compare
18805 instructions which are no longer used.
18806 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18808 (s390_canonicalize_comparison): Remove handling of DFmode
18810 (s390_expand_vec_compare_scalar): Remove function.
18811 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18812 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18814 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18816 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18818 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18820 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18821 will anyway by matched by mov<mode>_64dfp.
18823 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18825 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18826 vlef/vstef. Add missing operand to vleif.
18828 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18830 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18831 pair for all vector types with 64 bit elements.
18832 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18833 * config/s390/vector.md (V_HW_64): ... here.
18834 (V_128_NOSINGLE): New mode iterator.
18835 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18836 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18837 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18838 ("*vec_load_pairv2di"): Change to ...
18839 ("*vec_load_pair<mode>"): ... this one.
18841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18843 * config/s390/constraints.md: Add comments.
18844 (jKK): Reject element sizes > 8 bytes.
18845 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18847 * config/s390/s390.md: Add the s_operand checks formerly in
18848 s390_split_ok_p to various splitters where they are still
18850 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18851 for 128 bit vectors. Plus two splitters.
18853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18855 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18858 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18861 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18862 error if the boundary argument is not constant.
18864 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18866 PR rtl-optimization/80112
18867 * loop-doloop.c (doloop_condition_get): Don't check condition
18868 if cmp isn't SET with IF_THEN_ELSE src.
18870 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18872 PR tree-optimization/80158
18873 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
18874 replacing a candidate statement, also replace it for the
18875 candidate's alternate interpretation.
18876 (replace_rhs_if_not_dup): Likewise.
18877 (replace_one_candidate): Likewise.
18879 2017-03-24 Richard Biener <rguenther@suse.de>
18881 PR tree-optimization/80167
18882 * graphite-isl-ast-to-gimple.c
18883 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
18885 (translate_isl_ast_to_gimple::get_rename): Likewise.
18887 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
18889 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
18890 handling of certain combinations of target options, including the
18891 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
18892 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
18894 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18897 * config/arm/arm.md (*load_multiple): Add reload_completed to
18898 matching condition.
18900 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18901 Richard Biener <rguenth@suse.de>
18903 PR tree-optimization/79908
18904 PR tree-optimization/80136
18905 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18906 been cast away, gimplify_and_add suffices.
18908 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18910 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18912 2017-03-23 Richard Biener <rguenther@suse.de>
18914 PR tree-optimization/80032
18915 * gimplify.c (gimple_push_cleanup): Forced unconditional
18916 cleanups still have to go to the conditional_cleanups
18919 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18921 PR tree-optimization/80072
18922 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18924 (next_operand_entry_id): Change type to unsigned int.
18925 (sort_by_operand_rank): Make sure to return the right return value
18926 even if unsigned fields are bigger than INT_MAX.
18927 (struct oecount): Change cnt and id type to unsigned int.
18928 (oecount_hasher::equal): Formatting fix.
18929 (oecount_cmp): Make sure to return the right return value
18930 even if unsigned fields are bigger than INT_MAX.
18931 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18934 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18935 TREE_READONLY on result if writing it more than once.
18938 * doc/invoke.texi (-fsanitize=thread): Document that with
18939 -fnon-call-exceptions atomics are not able to throw
18943 * tsan.c: Include tree-eh.h.
18944 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18945 maybe_clean_or_replace_eh_stmt where needed.
18946 (instrument_memory_accesses): Add cfg_changed argument.
18947 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18948 if it returned true.
18949 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18951 PR rtl-optimization/63191
18952 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18953 wrapper function, moved the whole old content into ...
18954 (ix86_delegitimize_address_1): ... this. New inline function.
18955 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18956 true as last argument instead of ix86_delegitimize_address.
18958 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18960 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18961 cortexa57_branch_cost.
18963 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18965 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18967 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18970 * doc/md.texi (Constraints): Document wA constraint.
18971 * config/rs6000/constraints.md (wA): New.
18972 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18973 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18974 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18975 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18977 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18980 * gimplify.c (is_oacc_declared): New function.
18981 (oacc_default_clause): Use it to set default flags for acc declared
18982 variables inside parallel regions.
18983 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18984 declared variables.
18985 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18986 declare attribute to any decl as necessary.
18988 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18991 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18992 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18993 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18994 (arm_arch_lpae): This.
18995 * config/arm/arm.c (arm_arch7ve): Rename into ...
18996 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18997 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19000 2017-03-22 Martin Liska <mliska@suse.cz>
19003 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19004 error message instead of an ICE.
19006 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19008 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19010 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19012 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19014 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19017 2017-03-21 Martin Sebor <msebor@redhat.com>
19019 * doc/extend.texi: Use "cannot" instead of "can't."
19020 * doc/hostconfig.texi: Same.
19021 * doc/install.texi: Same.
19022 * doc/invoke.texi: Same.
19023 * doc/loop.texi: Same.
19024 * doc/md.texi: Same.
19025 * doc/objc.texi: Same.
19026 * doc/rtl.texi: Same.
19027 * doc/tm.texi: Same.
19028 * doc/tm.texi.in: Same.
19029 * doc/trouble.texi: Same.
19031 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19034 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19035 (collect_checksum_attributes): Set it.
19036 (die_checksum_ordered): Use it.
19038 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19040 PR tree-optimization/79908
19041 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19042 change: For a VA_ARG whose LHS has been cast away, use
19043 force_gimple_operand to construct the side effects.
19045 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19047 PR translation/80001
19048 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19049 more amenable to translation.
19050 (oacc_loop_auto_partitions): Likewise.
19052 2017-03-21 Marek Polacek <polacek@redhat.com>
19053 Martin Sebor <msebor@redhat.com>
19055 PR tree-optimization/80109
19056 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19057 on INTEGRAL_TYPE_P.
19059 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19060 Segher Boessenkool <segher@kernel.crashing.org>
19063 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19064 check reg_used_between_p between insn and one of succ or succ2
19065 depending on if succ is artificial insn not inserted into insn
19068 2017-03-21 Martin Liska <mliska@suse.cz>
19070 PR gcov-profile/80081
19071 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19072 * doc/gcc.texi: Include gcov-dump stuff.
19073 * doc/gcov-dump.texi: New file.
19075 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19077 PR rtl-optimization/79150
19078 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19079 conditional jump, if the jump is the last insn of the loop.
19081 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19082 Richard Biener <rguenth@suse.de>
19084 PR tree-optimization/79908
19085 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19086 been cast away, use force_gimple_operand to construct the side
19089 2017-03-21 Martin Liska <mliska@suse.cz>
19091 PR libfortran/79956
19092 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19095 2017-03-21 Brad Spengler <spender@grsecurity.net>
19098 * plugin.c (htab_hash_plugin): New function.
19099 (add_new_plugin): Use it and adjust.
19100 (parse_plugin_arg_opt): Adjust.
19101 (init_one_plugin): Likewise.
19103 2017-03-21 Richard Biener <rguenther@suse.de>
19105 PR tree-optimization/80032
19106 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19107 if set force the cleanup to happen unconditionally.
19108 (gimplify_target_expr): Push inserted clobbers with force_uncond
19109 to avoid them being removed by control-dependent DCE.
19111 2017-03-21 Richard Biener <rguenther@suse.de>
19113 PR tree-optimization/80122
19114 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19115 va_arg_pack_len when the inlined call stmt requires pack
19117 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19119 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19122 * tsan.c (instrument_builtin_call): If the memory model argument
19123 is not a constant, assume it is valid.
19126 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19129 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19131 PR rtl-optimization/79910
19132 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19133 if its dest is used by an insn before I2 (other than the combined
19134 insns themselves, which are properly handled already).
19136 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19139 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19141 * combine.c (record_used_regs): New static function.
19142 (try_combine): Handle situations where there is an additional
19143 instruction between I2 and I3 which needs to have a LOG_LINK
19147 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19149 * combine.c (try_combine): Delete redundant i1 test. Call
19150 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19152 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19155 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19156 alternatives 13/14.
19158 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19160 PR tree-optimization/80054
19161 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19162 the optimization if a PHI or any of its arguments is not dominated
19163 by the candidate's basis. Use gphi* rather than gimple* as
19165 (replace_profitable_candidates): Clean up a gimple* variable that
19166 should be a gphi* variable.
19168 2017-03-20 Martin Sebor <msebor@redhat.com>
19171 * doc/extend.texi (attribute constructor): Document present limitation.
19173 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19176 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19177 __POWER9_VECTOR__ #ifdef control, change template definition to
19178 use Power9-specific built-in function.
19179 (vec_any_eq): Likewise.
19180 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19181 to control outcomes from this test.
19182 (vector_ae_<mode>p): For VEC_F modes, likewise.
19184 2017-03-20 Ian Lance Taylor <iant@google.com>
19186 * config/i386/i386.c (ix86_function_regparm): Save an extra
19187 register for -fsplit-stack with DECL_STATIC_CHAIN.
19189 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19192 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19193 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19195 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19197 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19199 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19202 2017-03-20 Marek Polacek <polacek@redhat.com>
19205 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19207 2017-03-20 Richard Biener <rguenther@suse.de>
19209 PR tree-optimization/80113
19210 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19211 allocate extra SSA name for PHI def.
19212 (add_close_phis_to_outer_loops): Likewise.
19213 (add_close_phis_to_merge_points): Likewise.
19214 (copy_loop_close_phi_args): Likewise.
19215 (copy_cond_phi_nodes): Likewise.
19217 2017-03-20 Martin Liska <mliska@suse.cz>
19219 PR middle-end/79753
19220 * tree-chkp.c (chkp_build_returned_bound): Do not build
19221 returned bounds for a LHS that's not a BOUNDED_P type.
19223 2017-03-20 Martin Liska <mliska@suse.cz>
19227 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19228 COMPLEX_CST and VECTOR_CST.
19230 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19233 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19234 target operand. A new splitter adds the clobber statement in case
19235 the target operand is dead anyway.
19237 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19239 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19240 to age-old versions of binutils and glibc.
19242 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19244 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19246 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19248 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19250 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19252 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19253 requirement for binutils 2.13.
19255 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19257 * combine.c (try_combine): Delete redundant i1 test. Call
19258 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19260 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19262 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19263 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19265 <riscv64-*-elf>: Re-arrange section
19266 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19267 <riscv32-*-linux>: Likewise.
19268 <riscv64-*-elf>: Likewise
19269 <riscv64-*-linux>: Likewise.
19271 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19274 * aarch64.opt(verbose-cost-dump): Fix typo.
19276 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19279 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19280 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19282 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19284 * reload.c (find_reloads): When reloading a nonoffsettable address,
19285 use RELOAD_OTHER for it and its address reloads.
19287 PR rtl-optimization/79910
19288 * combine.c (record_used_regs): New static function.
19289 (try_combine): Handle situations where there is an additional
19290 instruction between I2 and I3 which needs to have a LOG_LINK
19293 2017-03-17 Jeff Law <law@redhat.com>
19295 PR tree-optimization/71437
19296 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19297 conditional in the hash table first.
19298 (vrp_dom_walker::before_dom_children): Extract condition from
19299 ASSERT_EXPR. Record condition, its inverion and any implied
19300 conditions as well.
19302 2017-03-17 Marek Polacek <polacek@redhat.com>
19303 Markus Trippelsdorf <markus@trippelsdorf.de>
19305 PR tree-optimization/80079
19306 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19309 2017-03-17 Richard Biener <rguenther@suse.de>
19311 PR middle-end/80075
19312 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19313 Properly verify the LHS before the RHS possibly claims to be
19315 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19318 2017-03-17 Martin Jambor <mjambor@suse.cz>
19320 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19321 (List of -O2 options): Likewise.
19322 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19325 2017-03-17 Tom de Vries <tom@codesourcery.com>
19327 * gcov-dump.c (print_usage): Print bug_report_url.
19329 2017-03-17 Richard Biener <rguenther@suse.de>
19331 PR middle-end/80050
19332 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19333 (parser::peek): Likewise.
19335 2017-03-17 Richard Biener <rguenther@suse.de>
19337 PR tree-optimization/80048
19338 * sese.c (free_sese_info): Properly release rename_map and
19339 copied_bb_map elements.
19341 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19343 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19344 Add linked-list forward and backlinks. Insert on
19345 construction, remove on destruction.
19346 (class pass_store_merging): Add m_stores_head field.
19347 (pass_store_merging::terminate_and_process_all_chains):
19348 Iterate over m_stores_head list.
19349 (pass_store_merging::terminate_all_aliasing_chains):
19351 (pass_store_merging::execute): Check for debug stmts first.
19352 Push new chains onto the m_stores_head stack.
19354 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19357 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19358 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19359 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19361 2017-03-16 Jeff Law <law@redhat.com>
19363 PR tree-optimization/71437
19364 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19365 member function. Implementation moved into after_dom_children
19366 member function and into the threader's thread_outgoing_edges
19368 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19369 some code into new thread_outgoing_edges.
19370 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19371 definition. Simplify marker handling (do it here). Assume we always
19372 have the available expression and the const/copies tables.
19373 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19375 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19376 * tree-vrp.c (equiv_stack): No longer file scoped.
19377 (vrp_dom_walker): New class.
19378 (vrp_dom_walker::before_dom_children): New member function.
19379 (vrp_dom_walker::after_dom_children): Likewise.
19380 (identify_jump_threads): Setup domwalker. Use it rather than
19381 walking edges in a random order by hand. Simplify setup/finalization.
19382 (finalize_jump_threads): Remove.
19383 (vrp_finalize): Do not call identify_jump_threads here.
19384 (execute_vrp): Do it here instead and call thread_through_all_blocks
19387 PR tree-optimization/71437
19388 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19390 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19392 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19393 (dom_opt_dom_walker::thread_across_edge): Remove
19394 handle_dominating_asserts argument. All callers changed.
19395 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19396 changes. Remove calls to lhs_of_dominating_assert. Other
19397 uses of handle_dominating_asserts turn into unconditional code
19398 (simplify_control_stmt_condition_1): Likewise.
19399 (simplify_control_stmt_condition): Likewise.
19400 (thread_through_normal_block, thread_across_edge): Likewise.
19401 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19402 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19403 object if it is not an SSA_NAME.
19404 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19405 before calling into the VRP specific simplifiers.
19406 (identify_jump_threads): Remove handle_dominating_asserts
19409 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19412 * tree-diagnostic.c (default_tree_printer): No longer static.
19413 * tree-diagnostic.h (default_tree_printer): New prototype.
19415 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19417 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19418 Change ins into fmov.
19420 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19422 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19423 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19424 Use h_con constraint for operand 1.
19425 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19426 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19428 2017-03-15 Jeff Law <law@redhat.com>
19430 PR tree-optimization/71437
19431 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19432 (record_temporary_equivalences): Use it.
19434 PR tree-optimization/71437
19435 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19436 tree-ssa-scopedtables.
19437 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19438 (record_conditions, record_cond, vuse_eq): Likewise.
19439 (record_edge_info): Adjust to API tweak of record_conditions.
19440 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19441 (record_temporary_equivalences, optimize_stmt): Likewise.
19442 (eliminate_redundant_computations): Likewise.
19443 (record_equivalences_from_stmt): Likewise.
19444 * tree-ssa-scopedtables.c: Include options.h and params.h.
19445 (vuse_eq): New function, moved from tree-ssa-dom.c
19446 (build_and_record_new_cond): Likewise.
19447 (record_conditions): Likewise. Accept vector of conditions rather
19448 than edge_equivalence structure for first argument.
19449 for the first argument.
19450 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19451 from tree-ssa-dom.c.
19452 (avail_exprs_stack::record_cond): Likewise.
19453 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19454 from tree-ssa-dom.c.
19455 (avail_exprs_stack): Add new member functions lookup_avail_expr
19457 (record_conditions): Declare.
19459 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19462 * lra-constraints.c (process_alt_operands): Increase reject for
19463 reloading an input/output operand.
19465 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19468 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19469 insns to convert from signed/unsigned char/short to IEEE 128-bit
19471 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19473 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19476 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19477 subreg of inner mode for values already in registers.
19479 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19481 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19482 iteration reg is used after the loop.
19484 2017-03-14 Martin Sebor <msebor@redhat.com>
19486 PR tree-optimization/79800
19487 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19488 precision in negative-positive range.
19489 (format_floating): Call non-const overload with adjusted precision.
19491 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19494 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19497 2017-03-14 Martin Sebor <msebor@redhat.com>
19499 PR middle-end/80020
19500 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19501 * builtins.def (aligned_alloc): Use it.
19504 * Makefile.in (GTFILES): Add calls.c.
19505 * calls.c: Include "gt-calls.h".
19507 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19509 PR rtl-optimization/79728
19510 * regs.h (struct target_regs): New field
19511 x_contains_allocatable_regs_of_mode.
19512 (contains_allocatable_regs_of_mode): New macro.
19513 * reginfo.c (init_reg_sets_1): Initialize it, and change
19514 contains_reg_of_mode so it includes global regs as well.
19515 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19516 rather than contains_regs_of_mode.
19518 2017-03-14 Martin Liska <mliska@suse.cz>
19520 * doc/invoke.texi: Document options that can't be combined with
19521 -fcheck-pointer-bounds.
19523 2017-03-14 Martin Liska <mliska@suse.cz>
19525 PR middle-end/79831
19526 * doc/invoke.texi (-Wchkp): Document the option.
19528 2017-03-14 Martin Liska <mliska@suse.cz>
19530 * Makefile.in: Install gcov-dump.
19532 2017-03-14 Martin Liska <mliska@suse.cz>
19534 * multiple_target.c (expand_target_clones): Bail out for
19535 an invalid attribute.
19537 2017-03-14 Richard Biener <rguenther@suse.de>
19539 * alias.c (struct alias_set_entry): Pack properly.
19540 * cfgloop.h (struct loop): Likewise.
19541 * cse.c (struct set): Likewise.
19542 * ipa-utils.c (struct searchc_env): Likewise.
19543 * loop-invariant.c (struct invariant): Likewise.
19544 * lra-remat.c (struct cand): Likewise.
19545 * recog.c (struct change_t): Likewise.
19546 * rtl.h (struct address_info): Likewise.
19547 * symbol-summary.h (function_summary): Likewise.
19548 * tree-loop-distribution.c (struct partition): Likewise.
19549 * tree-object-size.c (struct object_size_info): Likewise.
19550 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19551 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19552 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19553 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19554 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19555 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19556 (struct _stmt_vec_info): Likewise.
19558 2017-03-14 Martin Liska <mliska@suse.cz>
19561 * multiple_target.c (create_dispatcher_calls): Check that
19562 a target can create a function dispatcher.
19564 2017-03-14 Martin Liska <mliska@suse.cz>
19567 * multiple_target.c (expand_target_clones): Drop local.local
19568 flag for default implementation.
19570 2017-03-14 Richard Biener <rguenther@suse.de>
19572 PR tree-optimization/80030
19573 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19575 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19577 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19578 gcc_fallthrough() instead of __attribute__((fallthrough));
19580 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19582 * doc/gcc.texi: Remove "up" link to (DIR).
19583 * doc/gccint.texi: Ditto.
19585 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19587 * doc/install.texi (Specific) <avr>: Remove reference to
19590 2017-03-13 Jeff Law <law@redhat.com>
19592 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19593 attribute rather than comments.
19595 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19596 match_scratch operand is highest.
19598 2017-03-13 Martin Liska <mliska@suse.cz>
19600 PR middle-end/78339
19601 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19602 is a CHKP clone, use original declaration.
19604 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19606 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19607 (arc_conditional_register_usage): Use a different allocation order
19608 when optimizing for size.
19609 * common/config/arc/arc-common.c (arc_option_optimization_table):
19610 Section anchors default on when optimizing for size.
19612 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19614 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19616 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19618 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19619 * config/arc/arc.md (cpu_facility): Add cd variant.
19620 (*movqi_insn): Add code density variant.
19621 (*movhi_insn): Likewise.
19622 (*movqi_insn): Likewise.
19623 (*addsi3_mixed): Likewise.
19624 (subsi3_insn): Likewise.
19626 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19628 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19630 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19632 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19633 expressions with MINUS and UNARY ops.
19635 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19638 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19640 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19641 between vec_select and vector argument.
19642 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19643 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19644 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19645 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19646 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19647 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19649 2017-03-13 Richard Biener <rguenther@suse.de>
19652 * params.def (vect-max-peeling-for-alignment): Fix typo.
19654 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19656 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19657 issue that only occurred with binutils below 2.18.
19659 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19661 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19662 refer to binutils 2.11/2.12 minimum.
19664 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19666 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19667 ftp.kernel.org and simplify binutils requirement.
19669 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19671 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19673 (Optimize Options): Ditto. Also remove redundancy.
19675 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19677 PR translation/79848
19678 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19680 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19681 to G_ to avoid double translation.
19683 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19685 PR translation/79923
19686 * auto-profile.c (get_combined_location): Convert leading
19687 character of diagnostics to lower case and remove trailing period.
19688 (read_profile): Likewise for various diagnostics.
19689 * config/arm/arm.c (arm_option_override): Remove trailing period
19690 from various diagnostics.
19691 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19692 (msp430_expand_delay_cycles): Likewise.
19694 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19697 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19698 full command-line argument, rather than just "str".
19699 (aarch64_validate_march): Likewise.
19700 (aarch64_validate_mtune): Likewise.
19702 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19704 PR rtl-optimization/78911
19705 * lra-assigns.c (must_not_spill_p): New function.
19706 (spill_for): Use it.
19708 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19710 PR tree-optimization/79981
19711 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19712 ATOMIC_COMPARE_EXCHANGE ifn result.
19713 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19714 IFN_ATOMIC_COMPARE_EXCHANGE.
19716 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19719 * opts.c (parse_sanitizer_options): Add missing question mark to
19720 "did you mean" message.
19722 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19724 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19726 (VMULEUH_UNS): Likewise.
19727 (VMULOUB_UNS): Likewise.
19728 (VMULOUH_UNS): Likewise.
19729 * config/rs6000/rs6000.c (builtin_function_type): Remove
19730 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19732 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19735 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19736 x with result of extra_parsing_for_operand_code_0.
19737 (function_reader::extra_parsing_for_operand_code_0): Convert
19738 return type from void to rtx, returning x. When reading
19739 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19740 larger size containing struct block_symbol.
19742 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19744 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19745 -mfloat128-hardware without -m64.
19747 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19750 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19751 entries to the case statement that marks unsigned arguments to
19752 overloaded functions.
19754 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19756 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19757 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19759 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19762 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19763 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19765 2017-03-10 Martin Liska <mliska@suse.cz>
19769 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19770 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19773 2017-03-10 Olivier Hainque <hainque@adacore.com>
19775 * tree-switch-conversion (array_value_type): Start by resetting
19776 candidate type to it's main variant.
19778 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19780 PR rtl-optimization/79909
19781 * combine.c (try_combine): Use simplify_replace_rtx on individual
19782 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19783 of the whole CALL_INSN_FUNCTION_USAGE.
19785 PR tree-optimization/79972
19786 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19787 get_range_info on SSA_NAMEs. Formatting fixes.
19789 2017-03-10 Richard Biener <rguenther@suse.de>
19790 Jakub Jelinek <jakub@redhat.com>
19792 PR tree-optimization/77975
19793 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19794 edge to be constant.
19795 (get_val_for): For constant x return it. Formatting fix.
19796 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19797 would use the same bases as the current one.
19799 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19801 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19802 instead of vec_select for V1TImode.
19803 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19805 (VSX_LE_128): Add V1TI to this mode iterator.
19806 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19807 (*vsx_le_perm_store_<mode>): Likewise.
19808 (pre-reload splitter for VSX stores): Likewise.
19809 (post-reload splitter for VSX stores): Likewise.
19810 (*vsx_xxpermdi2_le_<mode>): Likewise.
19811 (*vsx_lxvd2x2_le_<mode>): Likewise.
19812 (*vsx_stxvd2x2_le_<mode>): Likewise.
19814 2017-03-09 Michael Eager <eager@eagercon.com>
19816 Correct failures with --enable-checking=yes,rtl.
19818 * config/microblaze/microblaze.c (microblaze_expand_shift):
19819 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19820 test for const0_rtx.
19821 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19822 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19824 2017-03-09 Richard Biener <rguenther@suse.de>
19826 PR tree-optimization/79977
19827 * graphite-scop-detection.c (scop_detection::merge_sese):
19828 Handle the case of extra exits to blocks dominating the entry.
19830 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19832 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19835 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19837 PR rtl-optimization/79949
19838 * lra-constraints.c (process_alt_operands): Check memory when
19839 trying to predict a cycle. Print about the overall increase.
19841 2017-03-09 Richard Biener <rguenther@suse.de>
19843 PR middle-end/79971
19844 * gimple-expr.c (useless_type_conversion_p): Preserve
19845 TYPE_SATURATING for fixed-point types.
19847 2017-03-09 Richard Biener <rguenther@suse.de>
19850 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19851 alignment of BLKmode params.
19853 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19856 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19857 (VALL_NO_V2Q): Likewise.
19859 * config/aarch64/aarch64-simd.md
19860 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19862 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19863 VALL_NO_V2Q mode iterator.
19864 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
19866 2017-03-09 Martin Liska <mliska@suse.cz>
19868 PR tree-optimization/79631
19869 * tree-chkp-opt.c (chkp_is_constant_addr): Call
19870 tree_int_cst_sign_bit just for INTEGER constants.
19872 2017-03-09 Martin Liska <mliska@suse.cz>
19876 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
19879 2017-03-09 Marek Polacek <polacek@redhat.com>
19882 * tree.c (inchash::add_expr): Handle TREE_VEC.
19884 2017-03-09 Martin Liska <mliska@suse.cz>
19887 (chkp_narrow_size_and_offset): New function.
19888 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
19889 (void chkp_parse_bit_field_ref): New function.
19890 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
19891 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
19893 2017-03-09 Martin Liska <mliska@suse.cz>
19896 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
19897 (chkp_find_bounds_1): Remove gcc_unreachable.
19899 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19902 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19903 BUILT_IN_SYNC*, determine the access type from the size suffix and
19904 always build a MEM_REF with that type. Handle forgotten
19905 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19908 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19909 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19910 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19911 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19912 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19913 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19914 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19915 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19916 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19917 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19918 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19919 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19920 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19921 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19922 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19923 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19924 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19925 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19926 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19927 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19928 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19929 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19930 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19931 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19932 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19933 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19934 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19935 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19936 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19937 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19938 definitions outside of __OPTIMIZE__ guarded section.
19941 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19942 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19943 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19944 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19947 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19949 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19950 ("vfenez<mode>"): Add missing constraints.
19952 2017-03-08 Martin Sebor <msebor@redhat.com>
19955 * config/nds32/nds32.c (nds32_option_override):
19956 Fix misspelled diagnostic.
19958 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19961 * gimplify.c (gimplify_omp_for): Replace index var in outer
19962 taskloop statement with an artificial variable and add
19963 OMP_CLAUSE_PRIVATE clause for it.
19965 2017-03-08 Richard Biener <rguenther@suse.de>
19967 PR tree-optimization/79955
19968 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19969 for accesses that are completely outside of the variable.
19971 2017-03-08 Andrew Haley <aph@redhat.com>
19973 PR tree-optimization/79943
19974 * tree-ssa-loop-split.c (compute_new_first_bound): When
19975 calculating the new upper bound, (END-BEG) should be added, not
19978 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19980 * config/avr/avr.md (setmemhi): Make sure match_dup
19981 operand number comes before match_scratch.
19983 2017-03-08 Richard Biener <rguenther@suse.de>
19985 PR tree-optimization/79920
19986 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19987 with ncopies == 1 to ...
19988 (vect_transform_slp_perm_load): ... here. Properly compute
19989 all element loads by iterating VF times over the group. Do
19990 not handle ncopies (computed in a broken way) in
19991 vect_create_mask_and_perm.
19993 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19996 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19997 is a uniform vector, use uniform_vector_p return value instead of
19998 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20000 2017-03-07 Marek Polacek <polacek@redhat.com>
20002 PR middle-end/79809
20003 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20004 (alloca_call_type): Likewise.
20006 2017-03-07 Martin Liska <mliska@suse.cz>
20008 * gcov.c (process_args): Put comment to correct location.
20010 2017-03-07 Martin Liska <mliska@suse.cz>
20012 PR middle-end/68270
20013 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20014 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20015 (chkp_narrow_bounds_for_field): Likewise.
20016 (chkp_parse_array_and_component_ref): Pass one more argument to
20019 2017-03-07 Richard Biener <rguenther@suse.de>
20021 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20024 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20026 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20027 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20029 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20032 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20033 to end of description.
20034 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20036 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20038 PR rtl-optimization/79901
20039 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20041 (*avx512f_<code><mode>3<mask_name>): ... this.
20042 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20043 iterator instead of VI8_AVX2_AVX512BW.
20045 PR rtl-optimization/79901
20046 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20047 min/max expander, expand it using expand_vec_cond_expr.
20050 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20053 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20056 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20057 to void * for PCH reasons.
20058 * dwarf2out.c (output_loc_operands, output_die): Cast
20059 v.val_vec.array to unsigned char *.
20061 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20064 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20067 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20069 PR rtl-optimization/79571
20070 * lra-constraints.c (process_alt_operands): Calculate static
20071 reject and subtract it from overall when only addresses will be
20074 2017-03-06 Julia Koval <julia.koval@intel.com>
20077 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20078 incoming stack boundary to 128 for 64-bit targets.
20080 2017-03-06 Richard Biener <rguenther@suse.de>
20082 PR tree-optimization/79894
20083 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20084 to NULL after folding it.
20086 2017-03-06 Richard Biener <rguenther@suse.de>
20088 PR tree-optimization/79824
20089 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20090 check disabling peeling for gaps.
20092 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20094 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20095 attributes): Document gettimeofday.
20097 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20099 * config/s390/s390.c (s390_option_override_internal): Set
20100 PARAM_MIN_VECT_LOOP_BOUND
20102 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20104 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20105 * config/s390/s390.md: Likewise.
20107 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20110 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20111 (<avx2_avx512>_perm<mode>): Rename to ...
20112 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20114 (<avx512>_perm<mode>_mask): Rename to ...
20115 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20117 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20118 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20119 instead of VI8F_256_512.
20120 (avx512f_perm<mode>): New define_expand.
20121 (avx512f_perm<mode>_mask): Likewise.
20122 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20123 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20125 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20127 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20128 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20130 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20132 2017-03-06 Martin Liska <mliska@suse.cz>
20135 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20136 when having a SSA NAME w/o VAR_DECL assigned to it.
20138 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20140 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20141 msa_dpsub_<su>_d): Fix MODE for vec_select.
20143 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20145 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20147 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20149 2017-03-06 Richard Biener <rguenther@suse.de>
20151 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20152 * plugin.c (register_plugin_info): Likewise.
20153 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20155 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20157 * config/i386/sse.md (sse_storehps, sse_storelps,
20158 avx_<castmode><avxsizesuffix>_<castmode>,
20159 avx512f_<castmode><avxsizesuffix>_<castmode>,
20160 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20161 in condition that at least one operand is not a MEM.
20163 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20165 PR middle-end/79805
20166 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20167 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20169 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20170 gimple_call_nothrow_p flag based on whether original builtin can throw.
20171 If it can, emit following stmts on the fallthrough edge.
20172 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20173 don't create new bb if inserting just debug stmts on the edge, try to
20174 insert them on the fallthru bb or just reset debug stmts.
20176 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20179 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20180 restore recog_data (including the operand rtxes inside it) around
20181 the call to get_insn_template.
20183 2017-03-03 Martin Sebor <msebor@redhat.com>
20185 PR tree-optimization/79699
20186 * context.c (context::~context): Free MPFR caches to avoid
20187 a memory leak on program exit.
20189 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20191 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20192 Use wide_int::ulow () instead of .elt (0).
20194 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20196 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20197 (*pushxf): Limit oF constraint to 32bit targets and add oC
20198 constraint for 64bit targets.
20199 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20200 (*pushdf): Change rmF constraint to rmC.
20202 2017-03-03 Martin Liska <mliska@suse.cz>
20204 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20205 Remove unused variable.
20207 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20210 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20211 is a memory operand, increase num_memory.
20212 (ix86_expand_args_builtin): Likewise.
20214 2017-03-03 Jan Hubicka <jh@suse.cz>
20217 * ipa-devirt.c (maybe_record_node): Properly handle
20218 __cxa_pure_virtual visibility.
20220 2017-03-03 Martin Liska <mliska@suse.cz>
20222 PR tree-optimization/79803
20223 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20225 (pass_loop_prefetch::execute): Disabled optimization if an
20226 assumption about L1 cache size is not met.
20228 2017-03-03 Martin Liska <mliska@suse.cz>
20230 PR rtl-optimization/79574
20231 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20232 (hash_scan_set): Likewise.
20233 (dump_hash_table): Likewise.
20234 (hoist_code): Likewise.
20236 2017-03-03 Richard Biener <rguenther@suse.de>
20238 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20239 in place of uhigh ().
20240 (fixed_convert_from_real): Likewise.
20242 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20245 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20247 2017-03-03 Richard Biener <rguenther@suse.de>
20249 PR middle-end/79818
20250 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20251 TYPE_OVERFLOW_UNDEFINED check.
20253 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20255 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20257 (vector_ae_<mode>_p): Likewise.
20258 (vector_nez_<mode>_p): Likewise.
20259 (vector_ne_v2di_p): Likewise.
20260 (vector_ae_v2di_p): Likewise.
20261 (vector_ne_<mode>_p): Likewise.
20262 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20264 (vsx_tsqrt<mode>2_fe): Likewise.
20266 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20269 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20271 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20273 PR rtl-optimization/79780
20274 * cprop.c (one_cprop_pass): When second and further conditional trap
20275 in a single basic block is turned into an unconditional trap, turn it
20276 into a deleted note to avoid RTL verification failures.
20278 2017-03-02 Richard Biener <rguenther@suse.de>
20280 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20282 2017-03-02 Richard Biener <rguenther@suse.de>
20284 PR tree-optimization/79345
20286 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20287 param and abort the walk, returning -1 if it is hit.
20288 (walk_aliased_vdefs): Take a limit param and pass it on.
20289 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20290 defaulting to 0 and return a signed int.
20291 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20292 (check_defs): New helper.
20293 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20294 about uninitialized memory.
20295 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20296 bogus uninitialized warning.
20297 (fixed_convert_from_real): Likewise.
20299 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20301 PR tree-optimization/66768
20302 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20303 iv_use if base object can't be determined.
20305 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20307 PR tree-optimization/79345
20308 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20309 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20310 (get_pattern_stats): Initialize it.
20311 * genemit.c (gen_expand): Verify match_scratch numbers come after
20312 match_operand/match_dup numbers.
20313 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20314 match_scratch numbers.
20315 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20317 * config/s390/s390.md (trunctdsd2): Likewise.
20319 2017-03-02 Richard Biener <rguenther@suse.de>
20321 * wide-int.h (wide_int_storage::operator=): Implement in terms
20324 2017-03-02 Richard Biener <rguenther@suse.de>
20326 PR tree-optimization/79777
20327 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20328 the to insert expression to sth existing.
20330 2017-03-01 Martin Sebor <msebor@redhat.com>
20332 PR middle-end/79692
20333 * gimple-ssa-sprintf.c
20334 (directive::known_width_and_precision): New function.
20335 (format_integer): Use it.
20336 (get_mpfr_format_length): Consider the full range of precision
20337 when computing %g output with the # flag. Set the likely byte
20338 count to 3 rather than 1 when precision is indeterminate.
20339 (format_floating): Correct the lower bound of precision.
20341 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20343 * doc/invoke.texi: Document default code model for 64-bit Linux.
20345 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20348 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20349 udiv rather than div since input pattern is unsigned.
20351 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20353 * config/i386/i386.c (print_reg): Warn for values of
20354 unsupported size in integer register.
20356 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20359 * config/rs6000/predicates.md (current_file_function_operand): Do
20360 not allow self calls to be local if the function is replaceable.
20362 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20365 * config/rs6000/altivec.h (vec_ctz and others): Change the
20366 preprocessor macro that controls conditional compilation from
20367 _ARCH_PWR9 to __POWER9_VECTOR__.
20368 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20369 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20370 control (instead of _ARCH_PWR9 control) so that template
20371 definition uses power9-specific function.
20372 (vec_any_eq): Likewise.
20373 (vec_all_ne): Change macro definition to use a power9-specific
20374 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20375 _ARCH_PWR9 control).
20376 (vec_any_eq) Likewise.
20377 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20378 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20379 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20380 support for xvcmpnedp instruction.
20381 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20382 macro expansion so that Power9 implementation of vec_all_ne does
20383 not use the AltiVec predicate framework.
20384 (VCMPNEH_P): Likewise.
20385 (VCMPNEW_P): Likewise.
20386 (VCMPNED_P): Likewise.
20387 (VCMPNEFP_P): Likewise.
20388 (VCMPNEDP_P): Likewise.
20389 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20390 implementation of vec_any_eq to not use AltiVec predicate
20392 (VCMPAEH_P): Likewise.
20393 (VCMPAEW_P): Likewise.
20394 (VCMPAED_P): Likewise.
20395 (VCMPAEFP_P): Likewise.
20396 (VCMPAEDP_P): Likewise.
20397 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20398 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20399 not use the AltiVec predicate framework.
20400 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20401 of vec_any_eq to not use AltiVec predicate framework.
20402 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20403 support for predefined __POWER9_VECTOR__ macro to indicate that
20404 Power9 instruction selection is enabled.
20405 (altivec_overloaded_builtins): Remove extraneous
20406 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20407 function argument types RS6000_BTI_bool_V16QI and
20408 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20409 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20410 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20411 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20412 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20413 Power9 for implementations of vec_cmpne. Change the signature for
20414 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20415 (representing vec_all_ne) to remove the previously described first
20416 argument of type RS6000_BTI_INTSI, as this was an artifact of
20417 reliance on the AltiVec predicate framework, which is no longer
20418 used in the implementation of these functions. Add
20419 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20420 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20421 since, unlike the AltiVec predicate framework implementation, we
20422 do not share function descriptors between vec_alle and vec_anyeq.
20423 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20424 set of modes that receive special treatment even when
20425 TARGET_P9_VECTOR is true. The special treatment emits code that
20426 does not depend on Power9 instructions.
20427 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20428 define_expand to not rely on AltiVec predicate framework.
20429 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20431 (vector_ne_v2di_p): Change this define_expand to not rely on
20432 AltiVec predicate framework.
20433 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20435 (vector_ne_<mode>_p): Change this define_expand to not rely on
20436 AltiVec predicate framework.
20437 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20439 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20440 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20441 define_insn pattern.
20442 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20443 define_insn pattern because the xvcmpne<VSs>. instruction is not
20445 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20446 instruction is not supported.
20448 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20450 * config/nvptx/nvptx.c: Include intl.h.
20452 2017-03-01 Martin Jambor <mjambor@suse.cz>
20455 * ipa-prop.h (ipa_bits): Removed field known.
20456 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20457 to pointers. Adjusted their comments to warn about their sharing.
20458 (ipcp_transformation_summary): Change bits to a vector of pointers.
20459 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20460 (ipa_get_ipa_bits_for_value): Declare.
20461 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20462 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20463 (ipa_bits_hash_table): Likewise.
20464 (ipa_vr_ggc_hash_traits): Likewise.
20465 (ipa_vr_hash_table): Likewise.
20466 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20467 being pointers and vr_known being removed.
20468 (ipa_set_jf_unknown): Likewise.
20469 (ipa_get_ipa_bits_for_value): New function.
20470 (ipa_set_jfunc_bits): Likewise.
20471 (ipa_get_value_range): New overloaded functions.
20472 (ipa_set_jfunc_vr): Likewise.
20473 (ipa_compute_jump_functions_for_edge): Use the above functions to
20474 construct bits and vr parts of jump functions.
20475 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20476 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20478 (ipcp_grow_transformations_if_necessary): Also allocate
20479 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20481 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20482 them. Fix too long lines.
20483 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20484 vr_known being removed.
20485 (ipa_read_jump_function): Use new setter functions to construct bits
20486 and vr parts of jump functions or set them to NULL.
20487 (write_ipcp_transformation_info): Adjust for bits being pointers.
20488 (read_ipcp_transformation_info): Likewise.
20489 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20491 Include gt-ipa-prop.h.
20492 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20494 (ipcp_store_bits_results): Likewise.
20495 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20496 Do not write to existing jump functions but use a temporary instead.
20498 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20501 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20502 attempt to use its first operand as BIT_FIELD_REF base.
20504 2017-03-01 Richard Biener <rguenther@suse.de>
20506 PR middle-end/79721
20507 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20508 interpolating formula in wrapping arithmetic.
20509 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20511 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20513 PR tree-optimization/79734
20514 * tree-vect-generic.c (expand_vector_condition): Optimize
20515 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20516 Handle VEC_COND_EXPR where comparison has different inner width from
20517 type's inner width.
20519 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20521 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20522 markup, and similar issues. Remove @opindex entries for things
20523 that aren't options. Add missing -mmpy-option entries.
20525 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20527 PR tree-optimization/79737
20528 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20529 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20530 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20531 instead of byte_size. Formatting fix.
20532 (shift_bytes_in_array_right): Formatting fix.
20534 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20537 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20538 condition on optimize for the leaf function test.
20540 2017-02-28 Martin Liska <mliska@suse.cz>
20543 * read-rtl-function.c (function_reader::handle_unknown_directive):
20544 Bail out when one uses -flto.
20546 2017-02-28 Martin Liska <mliska@suse.cz>
20548 * common.opt: Replace space with tabular for options of <number>
20550 * config/i386/i386.opt: Show <number> value for
20551 -mlarge-data-threshold.
20552 * opts.c (print_filtered_help): Do not display number in hexadecimal
20555 2017-02-28 Martin Liska <mliska@suse.cz>
20557 * common.opt: Fix --help=option -Q for options which are of
20560 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20562 * config/i386/i386.c (print_reg): Error out for values
20563 of 8-bit size in invalid integer register.
20565 2017-02-28 Martin Sebor <msebor@redhat.com>
20567 PR tree-optimization/79691
20568 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20570 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20573 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20574 gcc_unreachable with output_operand_lossage.
20576 2017-02-28 Richard Biener <rguenther@suse.de>
20578 PR tree-optimization/79740
20579 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20581 (visit_nary_op): Insert the nary into the hashtable if we
20582 pattern-matched sth.
20583 * tree-ssa-pre.c (eliminate_insert): Robustify.
20585 2017-02-28 Richard Biener <rguenther@suse.de>
20587 PR middle-end/79731
20588 * fold-const.c (decode_field_reference): Reject out-of-bound
20591 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20593 * config/i386/i386.c: Include intl.h.
20594 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20595 instead of just cond ? "..." : "...".
20596 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20597 * coverage.c (read_counts_file): Likewise.
20598 * omp-offload.c: Include intl.h.
20599 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20600 of just cond ? "..." : "...".
20601 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20602 of just cond ? "..." : "...".
20604 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20607 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20609 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20610 'tune for' CPU name.
20611 * config/arm/arm-cpu-data.h: Regenerated.
20613 2017-02-28 Richard Biener <rguenther@suse.de>
20615 PR tree-optimization/79732
20616 * tree-inline.c (expand_call_inline): Do not shadow var.
20618 2017-02-28 Richard Biener <rguenther@suse.de>
20620 PR tree-optimization/79723
20621 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20622 address-space properly.
20624 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20626 * doc/optinfo.texi (Optimization groups): Fix option used for
20628 * doc/invoke.texi (-fopt-info): Document "omp".
20629 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20630 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20631 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20632 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20633 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20635 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20637 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20638 OPTGROUP_OMP with "omp".
20640 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20643 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20644 for arithmetic shift of unsigned V2DI.
20646 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20648 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20649 arc/linux.h headers.
20650 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20651 (LINK_SPEC): Likewise.
20652 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20653 (EXTRA_SPECS): Likewise.
20654 (STARTFILE_SPEC): Likewise.
20655 (ENDFILE_SPEC): Likewise.
20656 (LIB_SPEC): Likewise.
20657 (TARGET_SDATA_DEFAULT): Likewise.
20658 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20659 (MULTILIB_DEFAULTS): Likewise.
20660 (DWARF2_UNWIND_INFO): Likewise.
20661 * config/arc/big.h: New file.
20662 * config/arc/elf.h: Likewise.
20663 * config/arc/linux.h: Likewise.
20664 * config/arc/t-uClibc: Remove.
20666 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20668 PR tree-optimization/77536
20669 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20670 (tree_transform_and_unroll_loop): Use above function to compute the
20671 estimated niter of unrolled loop and use it when scaling profile.
20672 Also use count info rather than frequency if it's non-zero.
20673 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20674 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20675 (vect_transform_loop): Call above function.
20677 2017-02-27 Richard Biener <rguenther@suse.de>
20679 PR tree-optimization/45397
20680 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20681 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20682 (visit_nary_op): Add pattern matching for CSEing sign-changed
20683 or truncated operations with wider ones.
20685 2017-02-27 Richard Biener <rguenther@suse.de>
20687 PR tree-optimization/79690
20688 * tree-vect-stmts.c (vectorizable_store): Use vector type
20689 built from the DR with address-space.
20691 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20693 * doc/invoke.texi (Optimize Options): Refine the description
20694 of asan-use-after-return.
20696 2017-02-25 Alan Modra <amodra@gmail.com>
20698 PR rtl-optimization/79584
20699 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20700 base, not ad->base_term, the reg within base. Remove assertion
20701 that ad->base == ad->base_term. Replace gen_int_mode using
20702 bogus mode with const0_rtx.
20704 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20706 PR middle-end/79396
20707 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20708 FMA_EXPR like tcc_binary or tcc_unary.
20710 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20713 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20715 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20716 (output_loc_operands): Handle DW_OP_call_ref and
20717 DW_OP_GNU_variable_value.
20718 (struct variable_value_struct): New type.
20719 (struct variable_value_hasher): Likewise.
20720 (variable_value_hash): New variable.
20721 (string_types): Remove.
20722 (copy_loc_descr): New function.
20723 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20724 (prepend_loc_descr_to_each): New function.
20725 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20726 instead of add_loc_descr_to_each if the first argument is single
20727 location list and the second has multiple.
20728 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20729 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20730 when looking for variable value which doesn't have other location info.
20731 (loc_list_from_tree): Formatting fix.
20732 (gen_array_type_die): Simplify DW_AT_string_length handling.
20733 (adjust_string_types): Remove.
20734 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20735 string_types. Call resolve_variable_values.
20736 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20737 (resolve_addr_in_expr): Likewise. Add A argument.
20738 (copy_deref_exprloc): Remove deref argument. Adjust for the
20739 original expression being DW_OP_GNU_variable_value with optionally
20740 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20741 optionally after it.
20742 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20743 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20744 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20745 (variable_value_hasher::hash, variable_value_hasher::equal): New
20747 (resolve_variable_value_in_expr, resolve_variable_value,
20748 resolve_variable_values, note_variable_value_in_expr,
20749 note_variable_value): New functions.
20750 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20753 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20756 * opts.h (handle_generated_option): Add GENERATED_P argument.
20757 * opts-common.c (handle_option): Adjust function comment.
20758 (handle_generated_option): Add GENERATED_P argument, pass it to
20760 (control_warning_option): Pass false to handle_generated_option
20762 * opts.c (maybe_default_option): Pass true to handle_generated_option
20764 * optc-gen.awk: Likewise.
20766 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20768 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20769 a REG, look at the REG it is a SUBREG of.
20770 (splitter for cmpeqsi_t): Ditto.
20772 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20774 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20775 the special USEs with the pattern of the insn, not the insn itself.
20777 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20780 * doc/invoke.texi: Document -mload-store-pairs.
20782 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20783 Sandra Loosemore <sandra@codesourcery.com>
20785 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20786 argument isn't a CONST_INT.
20787 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20788 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20789 (nios2_validate_compare): Bypass alternate compare logic if *op2
20790 is not a CONST_INT.
20791 (ldstwm_operation_p): Return false if first_base is not a REG or
20792 if first_offset is not a CONST_INT.
20794 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20796 * config/cris/cris.md: Use correct operand in a define_peephole2.
20798 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20800 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20802 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20804 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20805 this_insn if it is an INSN or JUMP_INSN.
20806 (force_offsettable): Look at base, not at addr.
20807 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20808 on things that aren't necessarily CONST_INTs.
20810 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20812 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20813 -mfpmath=sse is the default also for x86-32 targets with SSE2
20814 instruction set when @option{-ffast-math} is enabled
20816 2017-02-24 Jeff Law <law@redhat.com>
20818 PR rtl-optimizatoin/79286
20819 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20822 2017-02-24 Richard Biener <rguenther@suse.de>
20824 PR tree-optimization/79389
20825 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20828 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20830 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20831 function comment to reflect reality.
20832 (loop_exits_before_overflow): Fix typo in function description.
20834 2017-02-24 Richard Biener <rguenther@suse.de>
20836 PR tree-optimization/79389
20837 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20838 properly that a threading opportunity exists. Detect conditional
20839 copy/constant propagation opportunities.
20841 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20843 * config/visium/visium.md (type): Add trap.
20844 (b): New mode attribute.
20845 (*btst): Rename into...
20846 (*btst<mode>): ...this and adjust.
20847 (*cbranchsi4_btst_insn): Rename into...
20848 (*cbranch<mode>4_btst_insn): ...this and adjust.
20849 (trap): New define_insn.
20851 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20853 PR tree-optimization/79389
20854 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20855 (noce_reversed_cond_code): New function.
20856 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20857 reversed_comparison_code. Formatting fix.
20858 (noce_try_store_flag): Test rev_cond != NULL in addition to
20859 reversed_comparison_code.
20860 (noce_try_store_flag_constants): Likewise.
20861 (noce_try_store_flag_mask): Likewise.
20862 (noce_try_addcc): Use rev_cond if non-NULL instead of
20863 reversed_comparison_code.
20864 (noce_try_cmove_arith): Likewise. Formatting fixes.
20865 (noce_try_minmax, noce_try_abs): Clear rev_cond.
20866 (noce_find_if_block): Initialize rev_cond.
20867 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
20868 instead of false as last argument never attempt to reverse it
20871 2017-02-23 Bin Cheng <bin.cheng@arm.com>
20873 PR tree-optimization/79663
20874 * tree-predcom.c (combine_chains): Process refs in reverse order
20875 only for ZERO length chains, and add explaining comment.
20877 2017-02-23 Jeff Law <law@redhat.com>
20879 PR tree-optimization/79578
20880 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
20881 in call to operand_equal_p.
20883 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
20886 * config/i386/cpuid.h: Fix another undefined behavior.
20888 2017-02-23 Richard Biener <rguenther@suse.de>
20890 PR tree-optimization/79683
20891 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
20892 vector types for data-refs.
20894 2017-02-23 Martin Liska <mliska@suse.cz>
20896 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
20898 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20900 PR middle-end/79665
20901 * internal-fn.c (get_range_pos_neg): Moved to ...
20902 * tree.c (get_range_pos_neg): ... here. No longer static.
20903 * tree.h (get_range_pos_neg): New prototype.
20904 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20905 are known to be in between 0 and signed maximum inclusive, try to
20906 expand both unsigned and signed divmod and use the cheaper one from
20909 2017-02-22 Jeff Law <law@redhat.com>
20911 PR tree-optimization/79578
20912 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20913 to compare base operands.
20915 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20918 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20919 gpc_reg_operand instead of fpr_reg_operand.
20921 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20923 * config/mips/mips.c (mips_return_in_memory): Force FP
20924 vector types to be returned in memory for o32 ABI.
20926 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20928 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20929 instead of DW_TAG_member for static data member declarations and don't
20930 set no_linkage_name for static inline data members.
20931 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20934 2017-02-22 Martin Liska <mliska@suse.cz>
20936 * doc/invoke.texi: Replace inequality signs with square brackets
20939 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20941 PR tree-optimization/68644
20942 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20944 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20947 * lra-constraints.c (simplify_operand_subreg): Handle
20948 WORD_REGISTER_OPERATIONS targets.
20950 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20953 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20954 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20955 elimination by swapping fld*.
20957 2017-02-22 Richard Biener <rguenther@suse.de>
20959 PR tree-optimization/79673
20960 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20961 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20962 irrelevant address-space qualifiers and avoiding a
20963 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20965 2017-02-22 Richard Biener <rguenther@suse.de>
20967 PR tree-optimization/79666
20968 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20969 to not symbolically negate if that may introduce undefined
20972 2017-02-22 Martin Liska <mliska@suse.cz>
20975 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20976 * data-streamer-out.c (streamer_write_gcov_count_stream):
20978 * value-prof.c (stream_out_histogram_value): Make assert more
20979 precise based on type of counter.
20981 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20984 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20985 Use nonimmediate_operand instead of memory_operand for operand 1.
20986 (float-extend standard_x87sse_constant_load splitter): Ditto.
20988 2017-02-21 Jeff Law <law@redhat.com>
20990 PR tree-optimization/79621
20991 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20992 blocks with edges to themselves.
20994 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20997 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20998 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20999 Use gimple_call_builtin_p.
21002 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21003 on temporarily removed DEBUG_INSNs.
21005 PR tree-optimization/79649
21006 * tree-loop-distribution.c (classify_partition): Give up on
21007 non-generic address space loads/stores.
21009 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21011 * doc/loop.texi (Loop manipulation): Remove nonexistent
21012 tree_ssa_loop_version from the documentation.
21013 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21015 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21018 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21019 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21020 * config/rs6000/rs6000.c: Include except.h.
21021 (rs6000_expand_split_stack_prologue): Call
21022 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21024 2017-02-21 Martin Jambor <mjambor@suse.cz>
21027 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21028 have been analyzed.
21030 2017-02-21 Martin Jambor <mjambor@suse.cz>
21032 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21033 for backward compatibility only.
21034 * doc/invoke.texi (Option Summary): Remove all references to
21035 -fipa-cp-alignment.
21037 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21041 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21043 * lra-constraints.c (curr_insn_transform): Handle
21044 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21046 2017-02-21 Martin Liska <mliska@suse.cz>
21048 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21050 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21052 PR translation/79638
21053 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21055 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21058 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21059 (arm_function_ok_for_sibcall): Return false for an indirect call by
21060 descriptor if all the argument registers are used.
21061 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21062 alignment of the function.
21064 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21066 PR tree-optimization/61441
21067 * simplify-rtx.c (simplify_const_unary_operation): For
21068 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21069 the sNaN unmodified.
21071 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21073 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21074 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21075 instead of SYSTEM_HEADER_DIR.
21077 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21078 Martin Liška <mliska@suse.cz>
21080 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21081 Fix typos and grammar, use active voice, and clarify.
21083 2017-02-20 Marek Polacek <polacek@redhat.com>
21085 PR middle-end/79537
21086 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21089 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21091 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21094 * config/i386/i386.c (ix86_expand_builtin): Handle
21095 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21096 ix86_builtins_isa[fcode].isa as a requirement of those
21097 flags and any other flag in the bitmask.
21098 (ix86_init_mmx_sse_builtins): Use 0 instead of
21099 ~OPTION_MASK_ISA_64BIT as mask.
21100 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21101 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21102 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21103 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21105 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21108 * lra-constraints.c (split_reg): Check requested split mode
21109 is supported by the register.
21111 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21113 * lra-constraints.c (simplify_operand_subreg): Remove early
21116 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21119 * lra-constraints.c (curr_insn_transform): Tighten condition
21120 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21122 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21125 * lra-constraints.c (curr_insn_transform): Handle
21126 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21128 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21131 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21133 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21135 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21138 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21141 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21143 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21146 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21148 * final.c (last_columnnum, override_columnnum): New variables.
21149 (final_start_function): Set last_columnnum, pass it to begin_prologue
21150 hook and pass 0 to dwarf2out_begin_prologue.
21151 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21152 to source_line debug hook.
21153 (notice_source_line): Compute last_columnnum and for debug_column_info
21154 return true on column changes.
21155 * debug.h (struct gcc_debug_hooks): Add column argument to
21156 source_line and begin_prologue hooks.
21157 (debug_nothing_int_charstar_int_bool): Remove prototype.
21158 (debug_nothing_int_int_charstar,
21159 debug_nothing_int_int_charstar_int_bool): New prototypes.
21160 (dwarf2out_begin_prologue): Add column argument.
21161 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21162 begin_prologue hooks.
21163 (debug_nothing_int_charstar_int_bool): Remove.
21164 (debug_nothing_int_int_charstar,
21165 debug_nothing_int_int_charstar_int_bool): New functions.
21166 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21167 through to dwarf2out_source_line.
21168 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21169 (dwarf2out_source_line): Add column argument, emit it if requested.
21170 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21172 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21173 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21174 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21175 through to dwarf2out_begin_prologue.
21176 (vmsdbgout_source_line): Add column argument, pass it through to
21177 dwarf2out_source_line.
21178 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21179 dbxout_source_line caller.
21180 (dbxout_source_line): Add column argument.
21182 * common.opt (gno-column-info, gcolumn-info): New options.
21183 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21184 (check_die): Also test for multiple DW_AT_decl_column attributes.
21185 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21186 DW_AT_decl_column if requested.
21187 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21189 (gen_variable_die): Likewise.
21190 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21191 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21194 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21195 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21196 (ix86_handle_option): Handle OPT_m3dnowa.
21197 * doc/invoke.texi (-m3dnowa): Document.
21198 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21199 -m3dnowa instead of -m3dnow -march=athlon.
21202 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21203 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21205 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21208 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21209 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21210 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21211 generator for vsx_xxpermdi_<mode>_be.
21212 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21213 force big-endian semantics.
21214 (vsx_xxpermdi_<mode>_be): New define_expand with same
21215 implementation as previous version of vsx_xxpermdi_<mode>.
21217 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21219 PR tree-optimization/79327
21220 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21221 variable, its initialization and use.
21223 2017-02-17 Julia Koval <julia.koval@intel.com>
21225 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21226 (OPTION_MASK_ISA_PKU_UNSET): New.
21227 (ix86_handle_option): Handle -mrdpid.
21228 * config/i386/cpuid.h (bit_RDPID): New.
21229 * config/i386/driver-i386.c (host_detect_local_cpu):
21230 Detect RDPID feature.
21231 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21232 * config/i386/i386-c.c (ix86_target_macros_internal):
21234 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21235 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21236 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21237 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21238 * config/i386/i386.md (define_insn "rdpid"): New.
21239 * config/i386/i386.opt Add -mrdpid.
21240 * config/i386/immintrin.h (_rdpid_u32): New.
21242 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21244 PR rtl-optimization/79541
21245 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21246 instead of transforming it into USE.
21248 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21250 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21251 If HONOR_SNANS (SFmode) force the input to a register.
21252 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21253 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21254 an frsp or similar insn.
21256 2017-02-17 Martin Liska <mliska@suse.cz>
21258 PR rtl-optimization/79577
21259 * params.def (selsched-max-sched-times): Increase minimum to 1.
21261 2017-02-17 Martin Liska <mliska@suse.cz>
21263 PR rtl-optimization/79574
21264 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21266 2017-02-17 Martin Liska <mliska@suse.cz>
21268 PR tree-optimization/79529
21269 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21270 ssa_defined_default_def_p to handle cases which are implicitly
21272 * tree-ssa.c (ssa_defined_default_def_p): New function.
21273 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21274 which are implicitly defined.
21275 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21277 2017-02-17 Richard Biener <rguenther@suse.de>
21279 PR middle-end/79576
21280 * params.def (max-ssa-name-query-depth): Limit to 10.
21282 2017-02-17 Richard Biener <rguenther@suse.de>
21284 PR tree-optimization/79552
21285 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21288 2017-02-17 Richard Biener <rguenther@suse.de>
21291 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21293 2017-02-17 Marek Polacek <polacek@redhat.com>
21295 PR middle-end/79536
21296 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21297 (fold_negate_expr): New wrapper.
21299 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21301 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21302 Correct terminology and de-emphasize pre-standard behavior.
21304 2017-02-16 Alan Modra <amodra@gmail.com>
21306 PR rtl-optimization/79286
21307 * ira.c (def_dominates_uses): New function.
21308 (update_equiv_regs): Don't create an equivalence for insns that
21309 may trap where the register def does not dominate the use.
21311 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21313 PR rtl-optimization/78127
21314 * lra.c (lra): Call lra_eliminate before finish the loop after
21317 2017-02-16 Richard Biener <rguenther@suse.de>
21319 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21321 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21322 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21323 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21324 (isl_val_int_from_wi): New function.
21325 (extract_affine_gmp): Rename to ...
21326 (extract_affine_wi): ... this, take a widest_int.
21327 (extract_affine_int): Just wrap extract_affine_wi.
21328 (add_param_constraints): Use isl_val_int_from_wi.
21329 (add_loop_constraints): Likewise, and extract_affine_wi.
21331 2017-02-15 Jeff Law <law@redhat.com>
21333 PR middle-end/79521
21334 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21335 ira_init_register_move_cost_if_necessary.
21337 2017-02-15 Martin Sebor <msebor@redhat.com>
21339 PR middle-end/32003
21340 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21341 removed in a prior commit.
21343 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21345 PR tree-optimization/79347
21346 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21347 counters during peeling.
21349 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21351 * Makefile.in (site.exp): Remove "set ISLVER".
21353 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21356 * real.c (real_from_integer): Call real_convert even for decimal.
21358 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21361 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21363 2017-02-14 Andrew Pinski <apinski@cavium.com>
21365 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21366 cores and change the partno/implementer to be correct.
21367 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21368 the 'B" as the implementer.
21369 * config/aarch64/aarch64-tune.md: Regenerate.
21371 2017-02-14 Carl Love <cel@us.ibm.com>
21373 * config/rs6000/rs6000.c: Add case statement entry to make the
21374 xvcvuxdsp built-in argument unsigned.
21375 * config/rs6000/vsx.md: Fix the source and return operand types so they
21376 match the instruction definitions from the ISA document. Fix typo
21377 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21380 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21383 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21384 member early_clobber_alts.
21385 * lra-lives.c (reg_early_clobber_p): New.
21386 (process_bb_lives): Use it.
21387 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21388 (debug_operand_data): Initialize early_clobber_alts.
21389 (setup_operand_alternative): Set up early_clobber_alts.
21390 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21391 alternatives to new_insn_reg.
21392 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21394 (lra_update_insn_regno_info): Pass the new arg.
21396 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21398 PR middle-end/79505
21399 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21400 (new_oacc_loop_raw): Don't clear already cleared fields.
21403 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21404 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21405 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21407 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21410 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21412 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21415 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21416 the extra instruction to the right place to store 128-bit constant
21419 2017-02-14 Martin Sebor <msebor@redhat.com>
21421 PR middle-end/79448
21422 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21423 warning for strings of unknown length.
21425 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21427 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21429 2017-02-14 Jeff Law <law@redhat.com>
21432 * ira-costs.c (scan_one_insn): Initialize register move costs
21433 for pseudos seen in USE/CLOBBER insns.
21435 PR tree-optimization/79095
21436 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21437 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21438 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21439 if the operands are known to be not equal, then the resulting range
21441 (intersect_ranges): If the new range is ~[0,0] and the old range is
21442 wide, then prefer ~[0,0].
21443 * tree-vrp.c (overflow_comparison_p_1): New function.
21444 (overflow_comparison_p): New function.
21445 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21446 if NAME is used in an overflow test.
21447 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21448 overflow check that can be expressed as an equality test, then adjust
21449 ops to be that equality test.
21451 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21453 * config/s390/s390-builtin-types.def: Remove flags argument.
21454 * config/s390/s390.c (s390_init_builtins): Likewise.
21456 2017-02-14 Martin Liska <mliska@suse.cz>
21458 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21459 vector. Fix trailing white spaces.
21461 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21463 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21466 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21468 PR rtl-optimization/68664
21469 * config/arm/arm.c (arm_sched_can_speculate_insn):
21470 New function. Declare prototype.
21471 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21473 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21475 PR rtl-optimization/68664
21476 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21478 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21480 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21482 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21483 max skip bytes for function, loop and jump.
21485 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21487 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21488 ABS_EXPR for gimple dump.
21490 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21493 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21495 PR tree-optimization/79408
21496 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21497 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21498 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21499 also if rhs1 is INTEGER_CST.
21501 2017-02-14 Richard Biener <rguenther@suse.de>
21503 PR middle-end/79432
21504 * tree-into-ssa.c (insert_phi_nodes): When the function can
21505 have abnormal edges rewrite SSA names with broken use-def
21506 dominance out of SSA and register them for PHI insertion.
21508 2017-02-13 Martin Sebor <msebor@redhat.com>
21510 PR middle-end/79496
21511 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21512 clearing info.nowrite flag when snprintf size argument is a range.
21514 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21516 * cprop.c (cprop_jump): Add missing space in string literal.
21517 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21518 (get_constraint_for_component_ref): Likewise.
21519 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21520 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21521 * lra-constraints.c (process_alt_operands): Likewise.
21522 * ipa-inline.c (inline_small_functions): Likewise.
21523 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21524 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21525 * trans-mem.c (diagnose_tm_1_op): Likewise.
21526 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21527 (grid_parallel_clauses_gridifiable): Likewise.
21529 * config/nvptx/mkoffload.c (process): Add space in between
21532 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21533 "MOD4_SSE_REGS" and "ALL_REGS".
21535 * spellcheck.c (test_data): Add , in between "foo" and "food".
21537 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21540 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21541 boundary crossing check and subsequent code generation agree.
21543 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21545 * config/aarch64/aarch64.c (has_memory_op): Delete.
21546 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21549 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21551 PR rtl-optimization/79388
21552 PR rtl-optimization/79450
21553 * combine.c (distribute_notes): When removing TEM_INSN for which
21554 corresponding dest has last value recorded, invalidate that last
21557 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21559 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21560 of explicit '@'. Add missing assembly comment marker on branch costs
21563 2017-02-13 Nathan Sidwell <nathan@acm.org>
21565 * gengtype-lex.l (<in_struct>): Add '/'.
21567 2017-02-13 Martin Liska <mliska@suse.cz>
21570 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21572 2017-02-13 Richard Biener <rguenther@suse.de>
21574 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21576 * configure: Re-generate.
21577 * config.in: Likewise.
21578 * graphite-dependences.c: Simplify as if
21579 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21580 * graphite-isl-ast-to-gimple.c: Likewise.
21581 * graphite-optimize-isl.c: Likewise.
21582 * graphite-poly.c: Likewise.
21583 * graphite-sese-to-poly.c: Likewise.
21584 * graphite.h: Likewise.
21585 * toplev.c: Include isl/version.h and use isl_version () for
21586 printing the ISL version.
21587 * doc/install.texi: Update ISL requirement.
21589 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21591 * doc/standards.texi (Standards): Update reference to
21594 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21596 * doc/extend.texi (Named Address Spaces): sourceware.org now
21598 * doc/install.texi (Binaries): Ditto.
21601 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21603 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21604 standard terminology "stringize"/"stringizing" throughout.
21605 * doc/cppinternals.texi: Likewise.
21607 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21609 * doc/extend.texi: Fix some spelling mistakes and typos.
21610 * doc/invoke.texi: Likewise.
21612 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21615 * params.def (inline-min-speedup) Change from 10 to 8.
21617 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21619 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21622 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21625 * ipa-inline-analysis.c (get_minimal_bb): New function.
21626 (record_modified): Use it.
21627 (remap_edge_change_prob): Handle also ancestor functions.
21629 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21631 * doc/contrib.texi (Contributors): Remove broken link into
21632 the Mauve CVS repository.
21634 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21636 PR middle-end/79454
21637 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21638 result computation whenever lhs doesn't have vector mode, not
21639 just when it has BLKmode.
21641 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21643 * doc/makefile.texi (profiledbootstrap): Refer to the
21644 installation instructions only in textual form.
21646 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21649 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21651 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21653 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21654 (Specific): Update mingw-w64 reference.
21656 (Specific): Remove broken link to Renesas RX processor.
21658 2017-02-10 Richard Biener <rguenther@suse.de>
21660 * toplev.c (process_options): Do not mention obsolete graphite
21661 options when printing sorry message about missing graphite support.
21662 Mention -floop-nest-optimize.
21664 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21666 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21667 (vtst_p16): Likewise.
21668 (vtstq_p8): Likewise.
21669 (vtstq_p16): Likewise.
21671 (vtstq_p64): Likewise.
21672 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21673 (vset_lane_p64): New.
21674 (vsetq_lane_p64): New.
21676 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21678 PR tree-optimization/79411
21679 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21680 stmt operands are SSA_NAMEs used in abnormal phis.
21681 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21684 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21687 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21690 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21692 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21694 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21696 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21700 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21701 not arbitrary TREE_CONSTANT.
21704 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21705 "omp declare target link" attribute unless is_global_var.
21706 * omp-offload.c (find_link_var_op): Likewise.
21708 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21709 Chung-Lin Tang <cltang@codesourcery.com>
21711 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21713 (gimplify_adjust_omp_clauses): Don't delete TILE.
21714 (gimplify_omp_for): Deal with TILE.
21715 * internal-fn.c (expand_GOACC_TILE): New function.
21716 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21718 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21719 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21721 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21722 avoid DIV for outermost collapse var.
21723 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21724 Remove out of date comments, fix whitespace.
21725 * omp-general.c (omp_extract_for_data): Deal with tiling.
21726 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21727 adjust OLF_DIM_BASE value.
21728 (struct omp_for_data): Add tiling field.
21729 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21730 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21731 for auto loops. Remove default auto determining, moved to
21732 oacc_loop_fixed_partitions.
21733 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21734 stmts, add e_mask field.
21735 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21736 (oacc_thread_numbers): Use oacc_dim_call.
21737 (oacc_xform_tile): New.
21738 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21739 (finish_oacc_loop): Adjust for ifns vector.
21740 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21741 add case for GOACC_TILE fns.
21742 (oacc_loop_xform_loop): Delete.
21743 (oacc_loop_process): Iterate over call list directly, and add
21744 handling for GOACC_TILE fns.
21745 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21747 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21748 vector partitioning to outer loops. Assign 2 partitions to loops
21749 when available. Add TILE handling.
21750 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21751 (execite_oacc_device_lower): Process GOACC_TILE fns,
21752 ignore unknown specs.
21753 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21754 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21755 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21757 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21759 * configure.ac (ACX_BUGURL): Update.
21760 * configure: Regenerate.
21762 2017-02-09 Richard Biener <rguenther@suse.de>
21764 PR tree-optimization/69823
21765 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21766 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21768 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21770 * config/arc/arc-c.def: Add __NPS400__ definition.
21771 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21772 (TARGET_NPS400): Define.
21774 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21776 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21778 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21779 pointer, arch_info.
21780 (arc_cpu_types): Fill the arch_info field with a pointer into the
21781 arc_arch_types table.
21782 (arc_selected_cpu): Declare.
21783 * config/arc/arc.c (arc_selected_cpu): Make global.
21784 (arc_selected_arch): Delete.
21785 (arc_base_cpu): Delete.
21786 (arc_override_options): Remove references to deleted variables,
21787 update access to arch information.
21788 (ARC_OPT): Update access to arch information.
21789 (ARC_OPTX): Likewise.
21790 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21791 (TARGET_ARC600): Update access to arch information.
21792 (TARGET_ARC601): Likewise.
21793 (TARGET_ARC700): Likewise.
21794 (TARGET_EM): Likewise.
21795 (TARGET_HS): Likewise.
21796 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21799 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21802 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21803 condition/operands for integer GE/LE/GEU/LEU operations.
21805 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21807 PR translation/79397
21808 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21811 2017-02-08 Martin Jambor <mjambor@suse.cz>
21814 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21815 whether allocation happened.
21816 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21817 nothing was allocated.
21819 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21821 PR tree-optimization/79408
21822 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21823 constant, but SSA_NAME with a known integer range, use the minimum
21824 of that range instead of op1 to determine if modulo can be replaced
21825 with its first operand.
21827 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21829 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21831 2017-02-08 Richard Biener <rguenther@suse.de>
21833 PR tree-optimization/71824
21834 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21835 Check all loops contained in the merged region.
21837 2017-02-07 Andrew Pinski <apinski@cavium.com>
21839 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21841 2017-02-07 Andrew Pinski <apinski@cavium.com>
21843 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21844 (thunderxt88): Likewise.
21845 (thunderxt81): Disable LSE and change v8.1 to v8.
21846 (thunderxt83): Likewise.
21848 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21849 Richard Biener <rguenther@suse.de>
21851 PR middle-end/79399
21852 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21853 type from int to size_t.
21854 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21856 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21858 PR rtl-optimization/79386
21859 * cprop.c (bypass_conditional_jumps): Initialize
21860 bypass_last_basic_block already before splitting bbs after
21861 unconditional traps...
21862 (bypass_conditional_jumps): ... rather than here.
21865 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
21866 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
21867 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
21868 fix -masm=intel patterns.
21870 2017-02-07 Richard Biener <rguenther@suse.de>
21872 PR tree-optimization/79256
21873 PR middle-end/79278
21874 * builtins.c (get_object_alignment_2): Use min_align_of_type
21875 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
21876 and ADJUST_FIELD_ALIGN.
21878 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
21880 * doc/tm.texi: Regenerate.
21881 * stor-layout.c (layout_decl): Adjust.
21882 (update_alignment_for_field): Likewise.
21883 (place_field): Likewise.
21884 (min_align_of_type): Likewise.
21885 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
21886 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
21887 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
21888 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
21889 * config/frv/frv.c (frv_adjust_field_align): Likewise.
21890 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
21891 * config/i386/i386.c (x86_field_alignment): Likewise.
21892 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
21893 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
21894 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
21895 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
21896 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
21897 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21901 2017-01-30 Richard Biener <rguenther@suse.de>
21903 PR tree-optimization/79256
21904 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21905 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21908 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21910 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21911 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21912 builtins to SImode and emit a zero-extend, if necessary.
21914 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21916 * docs/invoke.texi (RISC-V Options): Alphabetize.
21918 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21920 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21923 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21925 * config/riscv/riscv.c: New file.
21926 * common/config/riscv/riscv-common.c: Likewise.
21927 * config.gcc: Likewise.
21928 * config/riscv/constraints.md: Likewise.
21929 * config/riscv/elf.h: Likewise.
21930 * config/riscv/generic.md: Likewise.
21931 * config/riscv/linux.h: Likewise.
21932 * config/riscv/multilib-generator: Likewise.
21933 * config/riscv/peephole.md: Likewise.
21934 * config/riscv/pic.md: Likewise.
21935 * config/riscv/predicates.md: Likewise.
21936 * config/riscv/riscv-builtins.c: Likewise.
21937 * config/riscv/riscv-c.c: Likewise.
21938 * config/riscv/riscv-ftypes.def: Likewise.
21939 * config/riscv/riscv-modes.def: Likewise.
21940 * config/riscv/riscv-opts.h: Likewise.
21941 * config/riscv/riscv-protos.h: Likewise.
21942 * config/riscv/riscv.h: Likewise.
21943 * config/riscv/riscv.md: Likewise.
21944 * config/riscv/riscv.opt: Likewise.
21945 * config/riscv/sync.md: Likewise.
21946 * config/riscv/t-elf-multilib: Likewise.
21947 * config/riscv/t-linux: Likewise.
21948 * config/riscv/t-linux-multilib: Likewise.
21949 * config/riscv/t-riscv: Likewise.
21950 * configure.ac: Likewise.
21951 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21952 Waterman as RISC-V maintainers.
21953 * doc/install.texi: Add RISC-V entries.
21954 * doc/invoke.texi: Add RISC-V options section.
21955 * doc/md.texi: Add RISC-V constraints section.
21956 * configure: Regenerated.
21958 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21961 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21962 false values to be constant vectors with all 0 or all 1 bits set.
21963 (vcondu<mode><mode>): Likewise.
21964 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21966 (fpmask_comparison_operator): Update comment.
21967 (vecint_comparison_operator): New predicate.
21968 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21969 vector conditionals when the true and false values are constant
21970 vectors with all 0 bits or all 1 bits set.
21972 2017-02-06 Martin Sebor <msebor@redhat.com>
21974 PR tree-optimization/79376
21975 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21977 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21979 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21980 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21981 to simplify split condition.
21983 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21985 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21986 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21989 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21991 PR rtl-optimization/68664
21992 * target.def (can_speculate_insn): New hook.
21993 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21994 * doc/tm.texi: Regenerate.
21995 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21996 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21997 (rs6000_sched_can_speculate_insn): New function.
21999 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22001 PR tree-optimization/79284
22002 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22003 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22004 vectorizable_mask_load_store, vectorizable_operation,
22005 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22006 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22007 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22008 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22009 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22010 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22011 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22012 is_gimple_assign (stmt). Replace another such test with
22013 is_gimple_assign (stmt).
22015 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22018 * config/avr/avr.c (rtl-iter.h): Include it.
22019 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22020 (avr_legitimate_combined_insn): ...and implementation.
22022 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22024 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22025 * config/s390/s390.c (s390_const_operand_ok)
22026 (s390_canonicalize_comparison, s390_extract_part)
22027 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22028 (s390_contiguous_bitmask_p, s390_rtx_costs)
22029 (legitimize_pic_address): Likewise.
22030 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22031 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22032 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22033 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22034 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22036 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22038 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22039 REGNO($0) == REGNO($1).
22041 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22043 * config/s390/linux.h(SIZE_TYPE): Add comment.
22045 2017-02-06 Julian Brown <julian@codesourcery.com>
22046 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22047 Virendra Pathak <virendra.pathak@broadcom.com>
22049 * config/aarch64/aarch64-cores.def: Change the scheduler
22051 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22052 * config/aarch64/thunderx2t99.md: New file.
22054 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22056 * doc/standards.texi (Go Language): Update link to language
22059 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22061 * tree-eh.c (lower_resx): Sanitize profile.
22062 (cleanup_empty_eh_move_lp): Likewise.
22064 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22067 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22069 * cfgloopmanip.h (loop_version): Update prototype.
22070 * modulo-sched.c (sms_schedule): Update call of loop_version.
22071 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22072 * tree-parloops.c (gen_parallel_loop): Likewise.
22073 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22074 * tree-ssa-loop-split.c (split_loop): Likewise.
22075 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22076 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22078 2017-02-05 Martin Liska <mliska@suse.cz>
22081 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22083 (print_operand_address): Initialize a struct to zero.
22085 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22087 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22088 garbage collector only in textual form.
22090 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22092 * doc/extend.texi (x86 specific memory model extensions for
22093 transactional memory): Simplify a phrase.
22095 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22098 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22099 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22100 (atomic_storedi_1): Likewise.
22102 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22104 PR tree-optimization/79338
22105 * tree-parloops.c (gather_scalar_reductions): Don't call
22106 vect_analyze_loop_form for loop->inner before destroying loop's
22109 2017-02-03 Martin Sebor <msebor@redhat.com>
22111 PR tree-optimization/79327
22112 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22113 when precision has resulted in leading zeros.
22114 (format_integer): Adjust the likely counter to assume an unknown
22115 argument that may be zero is non-zero.
22117 2017-02-03 Jason Merrill <jason@redhat.com>
22120 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22121 avoid copying non-taken branch.
22123 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22125 PR tree-optimization/79340
22126 * tree-vect-loop.c (vectorizable_reduction): Release
22127 vec_defs elements after safe_splicing them into other vectors.
22130 PR tree-optimization/79327
22131 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22132 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22134 (format_integer): Use wide_int_to_tree instead of build_int_cst
22135 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22136 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22137 of shortest and longest sequence.
22139 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22141 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22142 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22144 2017-02-03 Walter Lee <walt@tilera.com>
22147 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22148 after initial stackframe link reg save.
22149 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22151 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22154 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22155 wu for stxssp alternative.
22157 2017-02-03 Martin Sebor <msebor@redhat.com>
22159 PR tree-optimization/79352
22160 * gimple-fold.c (get_range_strlen): Add argument.
22161 (get_range_strlen): Change return type to bool.
22162 (get_maxval_strlen): Pass in a dummy argument.
22163 * gimple-fold.h (get_range_strlen): Change return type to bool.
22164 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22165 * tree.h (array_at_struct_end_p): Add argument.
22166 * tree.c (array_at_struct_end_p): Handle it.
22168 2017-02-03 Martin Liska <mliska@suse.cz>
22171 * multiple_target.c (create_dispatcher_calls): Redirect edge
22172 from a caller of a dispatcher.
22173 (expand_target_clones): Make the clones local.
22174 (ipa_target_clone): Do both target clones and resolvers.
22175 (ipa_dispatcher_calls): Remove the pass.
22176 (pass_dispatcher_calls::gate): Likewise.
22177 (make_pass_dispatcher_calls): Likewise.
22178 * passes.def (pass_target_clone): Put as very first IPA early
22181 2017-02-03 Martin Liska <mliska@suse.cz>
22183 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22184 in case of a function with ifunc attribute.
22186 2017-02-03 Martin Liska <mliska@suse.cz>
22188 * cgraph.c (cgraph_node::dump): Dump function version info.
22189 * symtab.c (symtab_node::dump_base): Add missing new line.
22191 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22193 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22194 (ifcombine_ifandif): Use it.
22196 2017-02-03 Martin Liska <mliska@suse.cz>
22198 * doc/invoke.texi: Document default value for
22199 use-after-scope-direct-emission-threshold.
22201 2017-02-03 Martin Liska <mliska@suse.cz>
22203 PR tree-optimization/79339
22204 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22205 (format_floating): Likewise.
22207 2017-02-03 Martin Liska <mliska@suse.cz>
22210 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22212 (ipa_node_params_t::remove): Likewise.
22213 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22214 initialization from removed ipa_node_params_t::insert.
22215 (ipa_node_params::~ipa_node_params): Move from removed
22216 ipa_node_params_t::release.
22217 * symbol-summary.h (symbol_summary::m_released): New member.
22218 Do not release a summary twice. Do not allow to call finalizer
22219 for types of a summary that live in GGC memory.
22221 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22223 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22226 2017-02-02 Martin Sebor <msebor@redhat.com>
22228 PR middle-end/79275
22229 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22230 (format_string): Tighten up the range of output for non-constant
22231 strings and correct the expected range for wide non-constant strings.
22233 2017-02-02 Martin Sebor <msebor@redhat.com>
22235 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22237 PR middle-end/32003
22238 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22240 (-fdump-tree-@var): Add to index and document how to come up
22241 with pass-specific option and dump file names.
22242 (-fdump-passes): Clarify where to look for output.
22244 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22246 PR middle-end/77445
22247 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22248 statistics of the analyzed path; allow threading for speed when
22249 any of BBs along the path are optimized for speed.
22251 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22253 PR middle-end/78468
22254 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22255 settings of the virtual registers.
22258 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22260 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22261 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22264 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22266 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22267 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22269 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22271 * config/s390/s390.md: Add missing comments with the expanded
22273 * config/s390/vector.md: Likewise.
22274 * config/s390/vx-builtins.md: Likewise.
22276 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22279 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22280 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22281 conditions on a single line.
22283 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22285 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22286 __S390_VX__ to __VX__.
22288 2017-02-01 Andrew Pinski <apinski@cavium.com>
22290 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22291 stmt_info to record_stmt_cost.
22292 (vect_get_known_peeling_cost): Pass stmt_info if known to
22294 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22295 cpu_vector_cost field into
22296 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22297 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22298 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22299 splitting of scalar_stmt_cost and vec_stmt_cost.
22300 (thunderx_vector_cost): Likewise.
22301 (cortexa57_vector_cost): LIkewise.
22302 (exynosm1_vector_cost): Likewise.
22303 (xgene1_vector_cost): Likewise.
22304 (thunderx2t99_vector_cost): Improve after the splitting of the two
22306 (aarch64_builtin_vectorization_cost): Update for the splitting of
22307 scalar_stmt_cost and vec_stmt_cost.
22309 2017-02-01 Torvald Riegel <triegel@redhat.com>
22310 Richard Henderson <rth@redhat.com>
22312 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22313 conditional on existance of a fast atomic load.
22314 * optabs-query.c (can_atomic_load_p): New function.
22315 * optabs-query.h (can_atomic_load_p): Declare it.
22316 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22317 no fast atomic load is available for the particular size of access.
22318 (expand_atomic_compare_and_swap): Likewise.
22319 (expand_atomic_load): Likewise.
22320 (expand_atomic_store): Likewise.
22321 (expand_atomic_fetch_op): Likewise.
22322 * testsuite/lib/target-supports.exp
22323 (check_effective_target_sync_int_128): Remove x86 because it provides
22324 no fast atomic load.
22325 (check_effective_target_sync_int_128_runtime): Likewise.
22327 2017-02-01 Richard Biener <rguenther@suse.de>
22329 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22330 (graphite_transform_loops): Provide opt-info for optimized nests.
22331 * tree-parloop.c (parallelize_loops): Provide opt-info for
22332 parallelized loops.
22334 2017-02-01 Richard Biener <rguenther@suse.de>
22336 PR middle-end/79315
22337 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22338 was not set before.
22340 2017-02-01 Richard Biener <rguenther@suse.de>
22342 PR tree-optimization/71824
22343 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22344 Verify the loops are valid in the merged SESE region.
22345 (scop_detection::can_represent_loop_1): Check analyzing the
22346 evolution of the number of iterations in the region succeeds.
22348 2017-01-31 Ian Lance Taylor <iant@golang.org>
22350 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22351 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22353 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22355 PR preprocessor/79210
22356 * input.c (get_substring_ranges_for_loc): Replace line_width
22357 assertion with error-handling.
22359 2017-01-31 Richard Biener <rguenther@suse.de>
22361 PR tree-optimization/77318
22362 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22363 (create_pw_aff_from_tree): Take loop parameter.
22364 (add_condition_to_pbb): Pass loop of the condition to
22365 create_pw_aff_from_tree.
22367 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22369 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22370 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22372 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22376 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22378 (convert_int_to_float128): Likewise.
22379 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22380 (convert_int_to_float128): Likewise.
22381 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22382 (UNSPEC_IEEE128_CONVERT): Likewise.
22383 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22384 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22385 Use local variables for IBM extended format.
22386 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22387 (fix_trunc<mode>si2_fprs): Likewise.
22388 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22389 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22390 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22391 to know that we can now have integers of all sizes in vector
22393 (fix<uns>_<mode>di2_hw): Likewise.
22394 (float<uns>_<mode>si2_hw): Likewise.
22395 (fix_<mode>si2_hw): Likewise.
22396 (fixuns_<mode>si2_hw): Likewise.
22397 (float<uns>_<mode>di2_hw): Likewise.
22398 (float_<mode>di2_hw): Likewise.
22399 (float_<mode>si2_hw): Likewise.
22400 (floatuns_<mode>di2_hw): Likewise.
22401 (floatuns_<mode>si2_hw): Likewise.
22402 (xscvqp<su>wz_<mode>): Delete, no longer used.
22403 (xscvqp<su>dz_<mode>): Likewise.
22404 (xscv<su>dqp_<mode>): Likewise.
22405 (ieee128_mfvsrd_64bit): Likewise.
22406 (ieee128_mfvsrd_32bit): Likewise.
22407 (ieee128_mfvsrwz): Likewise.
22408 (ieee128_mtvsrw): Likewise.
22409 (ieee128_mtvsrd_64bit): Likewise.
22410 (ieee128_mtvsrd_32bit): Likewise.
22412 2017-01-31 Martin Liska <mliska@suse.cz>
22415 * ipa-prop.c (ipa_free_all_node_params): Call release method
22416 instead of ~sumbol_summary to not to trigger double times
22419 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22421 PR tree-optimization/71691
22422 * bitmap.h (class auto_bitmap): New.
22423 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22424 is_maybe_undefined instead of ssa_undefined_value_p.
22426 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22428 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22429 __S390_ARCH_LEVEL__ to __ARCH__.
22431 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22433 PR tree-optimization/79267
22434 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22435 if should_remove_lhs_p is true.
22437 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22440 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22441 (add_alignment_attribute): New.
22442 (base_type_die): Add alignment attribute.
22443 (subrange_type_die): Likewise.
22444 (modified_type_die): Likewise.
22445 (gen_array_type_die): Likewise.
22446 (gen_descr_array_type_die: Likewise.
22447 (gen_enumeration_type_die): Likewise.
22448 (gen_subprogram_die): Likewise.
22449 (gen_variable_die): Likewise.
22450 (gen_field_die): Likewise.
22451 (gen_ptr_to_mbr_type_die): Likewise.
22452 (gen_struct_or_union_type_die): Likewise.
22453 (gen_subroutine_type_die): Likewise.
22454 (gen_typedef_die): Likewise.
22455 (base_type_cmp): Compare alignment attribute.
22457 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22460 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22461 (setb_unsigned) New pattern for setb with CCUNS.
22462 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22463 subfc./subfe sequence to avoid overflow problems. Generate a
22464 shorter sequence with cmpld/setb for power9.
22465 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22466 for generating subfc. instruction.
22467 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22468 now uses this instruction.
22470 2017-01-30 Ian Lance Taylor <iant@google.com>
22473 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22474 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22476 2017-01-30 Martin Sebor <msebor@redhat.com>
22478 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22479 Move constant to the right of a relational operator.
22480 (get_mpfr_format_length, format_character, format_string): Ditto.
22481 (should_warn_p, maybe_warn): Same.
22483 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22485 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22488 * asan.c (get_translation_unit_decl): Remove function.
22489 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22491 2017-01-30 Martin Liska <mliska@suse.cz>
22493 PR gcov-profile/79259
22494 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22495 -fprofile-generate.
22497 2017-01-30 Martin Liska <mliska@suse.cz>
22500 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22501 Initialize variables with NULL value.
22503 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22506 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22508 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22510 2017-01-30 Richard Biener <rguenther@suse.de>
22512 PR tree-optimization/79276
22513 * tree-vrp.c (process_assert_insertions): Properly adjust common
22514 when removing a duplicate.
22516 * gcc.dg/torture/pr79276.c: New testcase.
22518 2017-01-30 Richard Biener <rguenther@suse.de>
22520 PR tree-optimization/79256
22521 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22522 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22524 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22526 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22529 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22530 ("*r<noxa>sbg_<mode>_sll_bitmask")
22531 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22532 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22533 Use contiguous_bitmask_nowrap_operand.
22535 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22538 * config/rs6000/altivec.h (vec_xl): Revise #define.
22539 (vec_xst): Likewise.
22541 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22543 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22545 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22547 PR rtl-optimization/79194
22548 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22549 traps before call to bypass_conditional_jumps.
22551 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22553 PR tree-optimization/71374
22554 * lra-constraints.c (check_conflict_input_operands): New.
22555 (match_reload): Use it.
22557 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22560 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22561 account to calculate conflict_set.
22563 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22565 PR rtl-optimization/78559
22566 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22567 other_insn in combine.
22569 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22571 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22572 uint16_type_node for BT_UINT16.
22574 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22576 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22577 "RTL Tests" to menu.
22578 (GIMPLE Tests): New node.
22579 (RTL Tests): New node.
22581 2017-01-27 Richard Biener <rguenther@suse.de>
22583 PR tree-optimization/79245
22584 * tree-loop-distribution.c (distribute_loop): Apply cost
22585 modeling also to detected patterns.
22587 2017-01-27 Richard Biener <rguenther@suse.de>
22589 PR tree-optimization/71433
22590 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22591 (compare_assert_loc): New function.
22592 (process_assert_insertions): Sort and optimize assert locations
22593 to remove duplicates and push down identical assertions on
22594 edges to their destination block.
22596 2017-01-27 Richard Biener <rguenther@suse.de>
22598 PR tree-optimization/79244
22599 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22600 out SSA names even if abnormal.
22602 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22604 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22605 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22606 instead of MPFR_RNDN.
22608 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22611 * arm.c (arm_option_override): Don't call build_target_option_node
22612 until after doing all option overrides.
22613 (arm_valid_target_attribute_tree): Likewise.
22615 2017-01-27 Martin Liska <mliska@suse.cz>
22617 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22618 for {cd}tors and C++ {cd}tors.
22620 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22622 * config/s390/s390.md ("*setmem_long_and")
22623 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22625 2017-01-26 Martin Sebor <msebor@redhat.com>
22627 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22630 2017-01-26 Martin Sebor <msebor@redhat.com>
22632 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22633 HAVE_DFmode before using XFmode or DFmode.
22634 (parse_directive): Avoid using the z length modifier to avoid
22635 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22637 PR middle-end/78703
22638 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22639 to accept adjustment as an array.
22640 (get_int_range): New function.
22641 (struct directive): Make width and prec arrays.
22642 (directive::set_width, directive::set_precision): Call get_int_range.
22643 (format_integer, format_floating): Handle width and precision ranges.
22644 (format_string, parse_directive): Same.
22646 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22649 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22650 comdat_type_p set, just clone them, but keep the children in the
22654 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22655 which have direct callers with -fvar-tracking-assignments enabled
22657 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22658 inside of type units.
22660 2017-01-26 Martin Sebor <msebor@redhat.com>
22662 PR middle-end/78703
22663 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22665 (struct format_result): Replace number_chars, number_chars_min,
22666 and number_chars_max with a single member of struct result_range.
22668 (format_result::operator+=): Adjust.
22669 (struct fmtresult): Remove bounded. Handle likely and unlikely
22671 (fmtresult::adjust_for_width_or_precision): New function.
22672 (fmtresult:type_max_digits): New function.
22673 (bytes_remaining): Handle likely and unlikely counters.
22674 (min_bytes_remaining): Remove.
22675 (format_percent): Simplify.
22676 (format_integer, format_floating): Set likely and unlikely counters.
22677 (get_string_length, format_character, format_string): Same.
22678 (format_plain, should_warn_p): New function.
22679 (maybe_warn): Call should_warn_p. Update diagnostic messages
22680 and handle those for all directives, including plain strings.
22681 (format_directive): Handle likely and unlikely counters.
22682 Remove unnecessary quoting from diagnostics. Add an informational
22684 (add_bytes): Remove.
22685 (pass_sprintf_length::compute_format_length): Simplify.
22686 (try_substitute_return_value): Handle likely and unlikely counters.
22688 2017-01-26 Carl Love <cel@us.ibm.com>
22690 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22691 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22693 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22696 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22697 endianess for subregs into account.
22698 * lra-constraints.c (lra_constraints): Do risky transformations
22699 always on the first iteration.
22700 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22701 last_call_used_reg_set.
22702 (process_bb_lives): Define and use last_call_used_reg_set.
22703 * lra.c (lra): Always continue after lra_constraints on the first
22706 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22708 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22710 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22712 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22714 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22715 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22716 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22717 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22718 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22719 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22720 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22721 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22722 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22724 2017-01-26 Marek Polacek <polacek@redhat.com>
22727 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22728 for the third operand.
22730 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22732 PR middle-end/79236
22733 * omp-low.c (struct omp_context): Add simt_stmt field.
22734 (scan_omp_for): Return omp_context *.
22735 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22736 context to the _simt_ SIMD stmt.
22737 (lower_omp_for): For combined SIMD with sibling _simt_
22738 SIMD, make sure to use the same decls in _looptemp_
22739 clauses as in the sibling.
22741 2017-01-26 David Sherwood <david.sherwood@arm.com>
22743 PR middle-end/79212
22744 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22747 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22750 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22751 emit fld b; fld a; if possible.
22753 * brig-builtins.def: Update copyright years.
22754 * config/arm/arm_acle_builtins.def: Update copyright years.
22756 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22759 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22760 constraint instead of o for the stxsd instruction.
22762 2017-01-25 Carl Love <cel@us.ibm.com>
22764 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22765 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22767 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22769 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22771 2017-01-25 Richard Biener <rguenther@suse.de>
22773 PR tree-optimization/69264
22774 * target.def (vector_alignment_reachable): Improve documentation.
22775 * doc/tm.texi: Regenerate.
22776 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22778 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22779 earlier changes with respect to TYPE_USER_ALIGN.
22780 (vector_alignment_reachable_p): Likewise. Improve dumping.
22782 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22785 * config/arm/arm.md (xordi3): Force constant operand into a register
22788 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22790 * doc/invoke.texi (-fstore-merging): Correct default optimization
22791 levels at which it is enabled.
22792 (-O): Move -fstore-merging from list to...
22795 2017-01-25 Richard Biener <rguenther@suse.de>
22798 * omp-expand.c: Include debug.h.
22799 (expand_omp_taskreg): Make sure to generate early debug before
22800 outlining anything from a function.
22801 (expand_omp_target): Likewise.
22802 (grid_expand_target_grid_body): Likewise.
22804 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22807 * asan.c (get_translation_unit_decl): New function.
22808 (asan_add_global): Extract modules file name from globals
22809 TRANSLATION_UNIT_DECL name.
22811 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22814 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22815 for long calls with APCS frame and VFP.
22817 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22819 * cfg.c (original_copy_tables_initialized_p): New function.
22820 * cfg.h (original_copy_tables_initialized_p): New decl.
22821 * cfgrtl.c (relink_block_chain): Guard the call to
22822 free_original_copy_tables with a call to
22823 original_copy_tables_initialized_p.
22824 * cgraph.h (symtab_node::native_rtl_p): New decl.
22825 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22826 (symtab_node::needed_p): Don't assert for early assembly output
22827 for __RTL functions.
22828 (cgraph_node::finalize_function): Set "force_output" for __RTL
22830 (cgraph_node::analyze): Bail out early for __RTL functions.
22831 (analyze_functions): Update assertion to support __RTL functions.
22832 (cgraph_node::expand): Bail out early for __RTL functions.
22833 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22835 * function.h (struct function): Update comment for field
22837 * gimple-expr.c: Include "tree-pass.h".
22838 (gimple_has_body_p): Return false for __RTL functions.
22839 * Makefile.in (OBJS): Add run-rtl-passes.o.
22840 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22842 (gcc::pass_manager::get_clean_slate): New accessor.
22843 * passes.c: Include "insn-addr.h".
22844 (should_skip_pass_p): Add logging. Update logic for running
22845 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22846 property-provider override so it is only done for gimple passes.
22848 (skip_pass): New function.
22849 (execute_one_pass): Call skip_pass when skipping passes.
22850 * read-md.c (md_reader::read_char): Support filtering
22851 the input to a subset of line numbers.
22852 (md_reader::md_reader): Initialize fields
22853 m_first_line and m_last_line.
22854 (md_reader::read_file_fragment): New function.
22855 * read-md.h (md_reader::read_file_fragment): New decl.
22856 (md_reader::m_first_line): New field.
22857 (md_reader::m_last_line): New field.
22858 * read-rtl-function.c (function_reader::create_function): Only
22859 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22860 curr_properties. Set DECL_INITIAL to a dummy block.
22861 (read_rtl_function_body_from_file_range): New function.
22862 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22864 * run-rtl-passes.c: New file.
22865 * run-rtl-passes.h: New file.
22867 2017-01-24 Jeff Law <law@redhat.com>
22869 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
22872 2017-01-24 Bin Cheng <bin.cheng@arm.com>
22874 PR tree-optimization/79159
22875 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
22876 (record_nonwrapping_iv): Improve boundary using above function if no
22877 value range information.
22879 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
22880 Martin Jambor <mjambor@suse.cz>
22882 * brig-builtins.def: New file.
22883 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
22884 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
22885 (DEF_HSAIL_SAT_BUILTIN): Likewise.
22886 (DEF_HSAIL_INTR_BUILTIN): Likewise.
22887 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
22888 * builtin-types.def (BT_INT8): New.
22889 (BT_INT16): Likewise.
22890 (BT_UINT8): Likewise.
22891 (BT_UINT16): Likewise.
22892 (BT_FN_ULONG): Likewise.
22893 (BT_FN_UINT_INT): Likewise.
22894 (BT_FN_UINT_ULONG): Likewise.
22895 (BT_FN_UINT_LONG): Likewise.
22896 (BT_FN_UINT_PTR): Likewise.
22897 (BT_FN_ULONG_PTR): Likewise.
22898 (BT_FN_INT8_FLOAT): Likewise.
22899 (BT_FN_INT16_FLOAT): Likewise.
22900 (BT_FN_UINT32_FLOAT): Likewise.
22901 (BT_FN_UINT16_FLOAT): Likewise.
22902 (BT_FN_UINT8_FLOAT): Likewise.
22903 (BT_FN_UINT64_FLOAT): Likewise.
22904 (BT_FN_UINT16_UINT32): Likewise.
22905 (BT_FN_UINT32_UINT16): Likewise.
22906 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22907 (BT_FN_INT_PTR_INT): Likewise.
22908 (BT_FN_UINT_PTR_UINT): Likewise.
22909 (BT_FN_LONG_PTR_LONG): Likewise.
22910 (BT_FN_ULONG_PTR_ULONG): Likewise.
22911 (BT_FN_VOID_UINT64_UINT64): Likewise.
22912 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22913 (BT_FN_INT8_INT8_INT8): Likewise.
22914 (BT_FN_INT16_INT16_INT16): Likewise.
22915 (BT_FN_INT_INT_INT): Likewise.
22916 (BT_FN_UINT_FLOAT_UINT): Likewise.
22917 (BT_FN_FLOAT_UINT_UINT): Likewise.
22918 (BT_FN_ULONG_UINT_UINT): Likewise.
22919 (BT_FN_ULONG_UINT_PTR): Likewise.
22920 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22921 (BT_FN_UINT_UINT_UINT): Likewise.
22922 (BT_FN_VOID_UINT_PTR): Likewise.
22923 (BT_FN_UINT_UINT_PTR: Likewise.
22924 (BT_FN_UINT32_UINT64_PTR): Likewise.
22925 (BT_FN_INT_INT_UINT_UINT): Likewise.
22926 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22927 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22928 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22929 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22930 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22931 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22932 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22933 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22934 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22935 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22936 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22937 * doc/frontends.texi: List BRIG FE.
22938 * doc/install.texi (Testing): Add BRIG tesring requirements.
22939 * doc/invoke.texi (Overall Options): Mention BRIG.
22940 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22942 2017-01-24 Richard Biener <rguenther@suse.de>
22944 PR translation/79208
22945 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22947 2017-01-24 Martin Jambor <mjambor@suse.cz>
22950 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22951 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22952 and known_contexts.
22954 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22956 PR middle-end/79123
22957 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22958 casts from signed to unsigned really don't have a range.
22960 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22962 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22963 GMP_RNDx for compatiblity.
22965 2017-01-24 Martin Liska <mliska@suse.cz>
22968 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22969 that would prevent us to call alloca with -1 as argument.
22971 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22973 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22974 Avoid -Wformat-security warning.
22976 2017-01-23 Andrew Pinski <apinski@cavium.com>
22978 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22981 2017-01-23 Martin Sebor <msebor@redhat.com>
22983 PR middle-end/78703
22984 * gimple-ssa-sprintf.c (warn_level): New global.
22985 (format_integer): Use it here and throughout the rest of the file.
22986 Use the same switch to compute sign as base.
22987 (maybe_warn): New function.
22988 (format_directive): Factor out warnings into maybe_warn.
22989 Add debugging output. Use warn_level.
22990 (add_bytes): Use warn_level.
22991 (pass_sprintf_length::compute_format_length): Add debugging output.
22992 (try_substitute_return_value): Same.
22993 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22995 PR middle-end/78703
22996 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22997 (struct fmtresult, format_integer, format_floating): Adjust.
22998 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22999 (get_string_length, format_string,format_directive): Same.
23000 (pass_sprintf_length::compute_format_length): Same.
23001 (try_substitute_return_value): Simplify slightly.
23003 PR middle-end/78703
23004 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23005 (fmtresult::operator+=): Outlined.
23006 (struct fmtresult): Add ctors.
23007 (struct conversion_spec): Rename...
23008 (struct directive): ...to this. Add and remove data members.
23009 (directive::set_width, directive::set_precision): New functions.
23010 (format_percent): Use fmtresult ctor.
23011 (get_width_and_precision): Remove.
23012 (format_integer): Make naming changes. Avoid computing width and
23014 (format_floating): Same. Adjust indentation.
23015 (format_character, format_none): New functions.
23016 (format_string): Moved character handling to format_character.
23017 (format_directive): Remove arguments, change return type.
23018 (parse_directive): New function.
23019 (pass_sprintf_length::compute_format_length): Move directive
23020 parsing to parse_directive.
23022 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23024 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23025 (assign_assembler_name_if_needed): ... this.
23026 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23027 (assign_assembler_name_if_needed): ... this.
23028 (free_lang_data_in_cgraph): Adjust callers.
23029 * cgraphunit.c (cgraph_node::analyze): Likewise.
23030 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23033 2017-01-23 Richard Biener <rguenther@suse.de>
23035 PR tree-optimization/79088
23036 PR tree-optimization/79188
23037 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23038 resetting loop bounds after last path deletion. Reset loop
23039 bounds of the target loop, make code match the comments.
23040 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23041 Make sure loops need no fixups.
23043 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23045 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23046 exponent support with double type for first argument.
23047 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23048 type returned by __builtin_vec_extract_sig,
23049 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23050 functions from "vector int" to "vector unsigned int" or from
23051 "vector long long int" to "vector unsigned long long int".
23052 Changed type returned by __builtin_vec_extract_exp,
23053 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23054 functions from "vector int" to "vector unsigned int" or from
23055 "vector long long int" to "vector unsigned long long int".
23056 Changed return type of __builtin_vec_test_data_class,
23057 __builtin_vec_test_data_class_sp, and
23058 __builtin_vec_test_data_class_dp from "vector int" to
23059 "vector bool int" or from "vector long long int" to "vector bool
23060 long long int" and changed second argument type from "unsigned
23061 int" to "int". Added new overloaded function forms "vector float
23062 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23063 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23064 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23065 double, vector unsigned long long int)" and "vector double
23066 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23067 long int)". Changed return type of
23068 __builtin_scalar_test_data_class and
23069 __builtin_scalar_test_data_class_sp and
23070 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23071 int" and changed second argument from "unsigned int" to "int".
23072 Changed type returned by __builtin_scalar_test_neg,
23073 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23074 from "int" to "bool int". Added new overloaded function form
23075 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23076 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23077 exponent double-precision with floating point first argument.
23078 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23079 documentation of scalar_test_data_class, scalar_test_neg,
23080 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23081 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23082 vec_test_data_class built-in functions to reflect refinements in
23083 their type signatures.
23085 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23087 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23089 (aarch64_elf_asm_destructor): Likewise.
23091 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23093 PR rtl-optimization/78634
23094 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23095 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23096 * ifcvt.c (noce_try_cmove): Add missing cost check.
23098 PR rtl-optimization/71724
23099 * combine.c (if_then_else_cond): Look for situations where it is
23100 beneficial to undo the work of one of the recursive calls.
23102 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23104 PR tree-optimization/70754
23105 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23106 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23107 combined stmt before it if not NULL.
23108 (combine_chains): Process refs reversely and compute dominance point
23111 2017-01-23 Martin Liska <mliska@suse.cz>
23113 PR tree-optimization/79196
23114 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23115 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23117 (strlen_optimize_stmt): Call the renamed function.
23119 2017-01-23 Michael Matz <matz@suse.de>
23121 PR tree-optimization/78384
23122 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23124 2017-01-23 Richard Biener <rguenther@suse.de>
23126 PR tree-optimization/79186
23127 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23128 both incoming edges before moving an assert.
23130 2017-01-23 Martin Jambor <mjambor@suse.cz>
23132 * ipa-prop.c (load_from_param_1): Removed.
23133 (load_from_unmodified_param): Bits from load_from_param_1 put back
23135 (load_from_param): Removed.
23136 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23137 with stmt. Reverted back to use of load_from_unmodified_param.
23139 2017-01-23 Martin Jambor <mjambor@suse.cz>
23142 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23143 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23144 field a pointer to garbage collected vector, mark lattices and
23145 ipcp_orig_node with GTY((skip)).
23146 (ipa_get_param_count): Adjust to descriptors being a pointer.
23147 (ipa_get_param): Likewise.
23148 (ipa_get_type): Likewise.
23149 (ipa_get_param_move_cost): Likewise.
23150 (ipa_set_param_used): Likewise.
23151 (ipa_get_controlled_uses): Likewise.
23152 (ipa_set_controlled_uses): Likewise.
23153 (ipa_is_param_used): Likewise.
23154 (ipa_node_params_t): Move into garbage collector. New methods insert
23156 (ipa_node_params_sum): Annotate wth GTY(()).
23157 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23159 (ipa_load_from_parm_agg): Adjust declaration.
23160 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23161 * ipa-profile.c (ipa_profile): Likewise.
23162 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23163 (ipa_populate_param_decls): Make descriptors parameter garbage
23165 (ipa_dump_param): Adjust to descriptors being a pointer.
23166 (ipa_alloc_node_params): Likewise.
23167 (ipa_initialize_node_params): Likewise.
23168 (load_from_param_1): Make descriptors parameter garbage collected.
23169 (load_from_unmodified_param): Likewise.
23170 (load_from_param): Likewise.
23171 (ipa_load_from_parm_agg): Likewise.
23172 (ipa_node_params::~ipa_node_params): Removed.
23173 (ipa_free_all_node_params): Remove call to delete operator.
23174 (ipa_node_params_t::insert): New.
23175 (ipa_node_params_t::remove): Likewise.
23176 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23177 copy known_csts and known_contexts vectors.
23178 (ipa_read_node_info): Adjust to descriptors being a pointer.
23179 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23181 (ipcp_transform_function): Make descriptors variable garbage
23184 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23186 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23187 * config/i386/avx512dqintrin.h: Ditto.
23188 * config/i386/avx512fintrin.h: Ditto.
23189 * config/i386/i386.c: Handle new builtins.
23190 * config/i386/i386-builtin.def: Add new builtins.
23191 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23192 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23194 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23195 Martin Liska <mliska@suse.cz>
23197 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23198 * asan.c (asan_expand_poison_ifn): Support stores and use
23199 appropriate ASAN report function.
23200 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23201 * internal-fn.def (ASAN_POISON_USE): Declare.
23202 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23203 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23204 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23205 ASAN_POISON calls w/o LHS.
23206 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23207 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23208 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23209 * gimplify.c (asan_poison_variables): Add attribute
23210 use_after_scope_memory to variables that really needs to live
23212 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23213 having the attribute.
23215 2017-01-23 Martin Liska <mliska@suse.cz>
23217 * asan.c (create_asan_shadow_var): New function.
23218 (asan_expand_poison_ifn): Likewise.
23219 * asan.h (asan_expand_poison_ifn): New declaration.
23220 * internal-fn.c (expand_ASAN_POISON): Likewise.
23221 * internal-fn.def (ASAN_POISON): New builtin.
23222 * sanopt.c (pass_sanopt::execute): Expand
23223 asan_expand_poison_ifn.
23224 * tree-inline.c (copy_decl_for_dup_finish): Make function
23226 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23227 * tree-ssa.c (is_asan_mark_p): New function.
23228 (execute_update_addresses_taken): Rewrite local variables
23229 (identified just by use-after-scope as addressable) into SSA.
23231 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23233 * doc/install.texi (Specific): opensource.apple.com uses https
23234 now. Remove trailing slash.
23236 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23238 * README.Portability: Remove note on an Irix compatibility issue.
23240 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23242 * gcov.c (INCLUDE_ALGORITHM): Define.
23243 (INCLUDE_VECTOR): Define.
23244 No longer include <vector> and <algorithm> directly.
23246 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23248 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23250 * doc/invoke.texi (Code Gen Options): Ditto.
23252 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23255 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23257 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23259 rtl-optimization/79125
23260 * cprop.c (local_cprop_pass): Handle cases where we make an
23261 unconditional trap.
23263 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23267 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23268 read from, for big endian.
23270 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23272 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23273 register pauth builtins for LP64 only.
23275 2017-01-20 Marek Polacek <polacek@redhat.com>
23278 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23281 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23283 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23285 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23286 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23287 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23289 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23292 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23293 in big-endian mode when they are not a single duplicated value.
23295 2017-01-20 Richard Biener <rguenther@suse.de>
23297 * BASE-VER: Bump to 7.0.1.
23299 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23301 * omp-low.c (omplow_simd_context): New struct. Use it...
23302 (lower_rec_simd_input_clauses): ...here and...
23303 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23304 references to idx, lane, max_vf, is_simt.
23306 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23308 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23311 2017-01-20 Martin Jambor <mjambor@suse.cz>
23313 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23314 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23316 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23317 (GTFILES): Rename hsa.c to hsa-common.c.
23318 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23319 * hsa-dump.c: Likewise.
23320 * hsa-gen.c: Likewise.
23321 * hsa-regalloc.c: Likewise.
23322 * ipa-hsa.c: Likewise.
23323 * omp-expand.c: Likewise.
23324 * omp-low.c: Likewise.
23325 * toplev.c: Likewise.
23327 2017-01-20 Marek Polacek <polacek@redhat.com>
23330 * doc/invoke.texi: Document -Wduplicated-branches.
23331 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23332 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23333 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23334 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23335 return 0 only when not OEP_LEXICOGRAPHIC.
23336 (fold_build_cleanup_point_expr): Use the expression
23337 location when building CLEANUP_POINT_EXPR.
23338 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23339 * tree.c (add_expr): Handle error_mark_node.
23341 2017-01-20 Martin Liska <mliska@suse.cz>
23344 * tree-profile.c (init_ic_make_global_vars): Do not call
23346 (gimple_init_gcov_profiler): Likewise.
23348 2017-01-20 Martin Liska <mliska@suse.cz>
23351 * cgraph.h (maybe_create_reference): Remove argument and
23353 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23355 * ipa-cp.c (create_specialized_node): Likewise.
23356 * symtab.c (symtab_node::maybe_create_reference): Handle
23357 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23359 2017-01-20 Martin Liska <mliska@suse.cz>
23361 * read-rtl-function.c (function_reader::create_function): Use
23362 build_decl instread of build_decl_stat.
23364 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23366 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23367 * config/i386/avx512dqintrin.h: Ditto.
23368 * config/i386/avx512fintrin.h: Ditto.
23369 * config/i386/i386-builtin-types.def: Add new types.
23370 * config/i386/i386.c: Handle new types.
23371 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23372 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23373 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23374 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23375 (__builtin_ia32_kshiftridi): New.
23376 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23378 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23382 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23383 define to rs6000_init_stack_protect_guard.
23384 (rs6000_init_stack_protect_guard): New function.
23386 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23387 Yunqiang Su <yunqiang.su@imgtec.com>
23389 * config.gcc (supported_defaults): Add madd4.
23390 (with_madd4): Add validation.
23391 (all_defaults): Add madd4.
23392 * config/mips/mips.opt (mmadd4): New option.
23393 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23395 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23397 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23398 (ISA_HAS_FUSED_MADD4): Likewise.
23399 * doc/invoke.texi (-mmadd4): Document the new option.
23400 * doc/install.texi (--with-madd4): Document the new option.
23402 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23404 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23405 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23406 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23407 (aarch64_init_pauth_hint_builtins): New.
23408 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23409 (aarch64_expand_builtin): Expand new builtins.
23411 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23413 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23414 * combine-stack-adj.c (no_unhandled_cfa): Handle
23415 REG_CFA_TOGGLE_RA_MANGLE.
23416 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23417 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23418 info for return address signing.
23419 (aarch64_expand_epilogue): Likewise.
23421 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23423 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23424 * config/aarch64/aarch64-protos.h
23425 (aarch64_return_address_signing_enabled): New declaration.
23426 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23428 (aarch64_expand_prologue): Sign return address before it's pushed onto
23430 (aarch64_expand_epilogue): Authenticate return address fetched from
23432 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23433 (aarch64_attributes): New function attributes for "sign-return-address".
23434 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23435 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23436 ("*do_return"): Generate combined instructions according to key index.
23437 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23438 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23440 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23441 * config/aarch64/aarch64.opt (msign-return-address=): New.
23442 * doc/extend.texi (AArch64 Function Attributes): Documents
23443 "sign-return-address=".
23444 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23446 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23448 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23449 overall option summary.
23451 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23453 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23454 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23455 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23456 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23458 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23460 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23461 -mpower9-minmax by default for -mcpu=power9.
23462 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23463 128-bit floating point.
23465 2017-01-20 Alan Modra <amodra@gmail.com>
23467 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23468 optimizing for size.
23470 2017-01-20 Alan Modra <amodra@gmail.com>
23473 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23474 for strcmp and strncmp from corresponding builtin decl.
23476 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23478 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23479 instead of i386/rtems-64.h.
23480 * config/i386/rtems-64.h: Remove.
23482 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23486 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23488 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23490 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23492 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23493 Change int to HOST_WIDE_INT.
23494 * config/aarch64/aarch64-protos.h
23495 (aarch64_simd_gen_const_vector_dup): Likewise.
23496 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23498 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23500 * langhooks-def.h (lhd_type_for_size): New decl.
23501 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23502 * langhooks.c (lhd_type_for_size): New function, taken from
23505 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23507 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23508 define_bypass for CR latency.
23509 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23510 (power9-alu2): Add define_bypass for CR latency.
23512 (power9-mul): Update insn latency.
23513 (power9-mul-compare): Update insn latency, bypass latency and remove
23516 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23518 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23520 * config/aarch64/aarch64.md
23521 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23522 aarch64_nopcrelative_literal_loads.
23523 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23525 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23527 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23528 TARGET_LOONGSON_3A.
23529 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23531 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23534 * config.gcc (supported_defaults): Add lxc1-sxc1.
23535 (with_lxc1_sxc1): Add validation.
23536 (all_defaults): Add lxc1-sxc1.
23537 * config/mips/mips.opt (mlxc1-sxc1): New option.
23538 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23540 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23541 __mips_no_lxc1_sxc1.
23542 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23543 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23544 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23546 2017-01-19 Richard Biener <rguenther@suse.de>
23548 PR tree-optimization/72488
23549 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23550 sure to restore SSA info.
23551 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23553 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23555 PR rtl-optimization/79121
23556 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23557 of the inner type when shifting an extended value.
23559 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23562 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23563 interposable aliases.
23565 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23568 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23569 Use the evmergelohi instruction.
23570 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23571 (mov_sitf_e500_subreg8_2_be): Likewise.
23572 (mov_sitf_e500_subreg12_2_le): Likewise.
23573 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23574 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23575 (mov_sitf_e500_subreg8_2_le): Likewise.
23576 (mov_sitf_e500_subreg12_2_be): Likewise.
23578 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23580 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23581 attribute from vecsimple to vecperm.
23582 (altivec_vbpermq2): Likewise.
23584 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23587 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23589 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23590 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23591 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23592 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23593 case where N arg is SIZE_MAX.
23594 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23595 (cmpstrsi): Add pattern.
23597 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23599 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23600 __builtin_vec_revb builtins.
23601 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23602 built-in functions to support generation of the ISA 3.0 XXBR<x>
23603 vector byte reverse instructions.
23604 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23605 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23606 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23607 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23608 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23609 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23610 (P9V_BUILTIN_VEC_REVB): Likewise.
23611 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23612 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23613 (p9_xxbrq_v16qi): Likewise.
23614 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23615 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23616 (p9_xxbrh_v8hi): Likewise.
23617 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23618 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23619 vec_revb built-in functions.
23621 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23623 PR rtl-optimization/78952
23624 * config/i386/i386.md (any_extract): New code iterator.
23625 (*insvqi_2): Use any_extract for source operand.
23626 (*insvqi_3): Use any_shiftrt for source operand.
23628 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23630 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23632 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23634 2017-01-18 Matthias Klose <doko@ubuntu.com>
23636 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23638 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23640 * config/rs6000/altivec.h (vec_bperm): Change #define.
23641 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23642 (altivec_vbpermq2): New define_insn.
23643 (altivec_vbpermd): Likewise.
23644 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23645 function interface.
23646 (VBPERMD): Likewise.
23647 (VBPERM): New polymorphic function interface.
23648 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23649 Add entries for P9V_BUILTIN_VEC_VBPERM.
23650 * doc/extend.texi: Add interfaces for vec_bperm.
23652 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23654 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23655 first letter of error messages.
23656 (s390_resolve_overloaded_builtin): Likewise.
23657 * config/s390/s390.c (s390_expand_builtin): Likewise.
23658 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23659 (s390_valid_target_attribute_inner_p): Likewise.
23660 * config/s390/s390.md ("tabort"): Likewise.
23662 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23664 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23665 (ISA_AVOID_DIV_HILO): New macro.
23666 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23667 (ISA_HAS_DDIV): Likewise.
23669 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23671 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23673 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23675 * doc/invoke.texi (fabi-version): Spelling fix.
23677 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23680 * doc/invoke.texi (fabi-version): Mention mangling fix for
23683 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23686 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23688 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23691 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23692 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23694 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23696 (stack_protect_test): Ditto.
23697 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23698 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23700 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23701 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23702 -mstack-protector-guard-offset=.
23703 (RS/6000 and PowerPC Options): Ditto.
23705 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23707 * config/i386/i386.h (MASK_CLASS_P): New define.
23708 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23709 there are no registers from different register sets also when
23710 mask registers are used. Update function comment.
23711 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23712 to (*k/*r) and (*k/*km) alternatives.
23714 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23716 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23717 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23718 (EH_RETURN_HANDLER_RTX): New define.
23719 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23720 Force frame pointer in EH return functions.
23721 (aarch64_expand_epilogue): Add barrier for eh_return.
23722 (aarch64_final_eh_return_addr): Remove.
23723 (aarch64_eh_return_handler_rtx): New function.
23724 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23726 (aarch64_eh_return_handler_rtx): New prototype.
23728 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23730 * config/rs6000/altivec.h (vec_rlmi): New #define.
23731 (vec_vrlnm): Likewise.
23732 (vec_rlnm): Likewise.
23733 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23734 (UNSPEC_VRLNM): Likewise.
23735 (VIlong): New mode iterator.
23736 (altivec_vrl<VI_char>mi): New define_insn.
23737 (altivec_vrl<VI_char>nm): Likewise.
23738 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23740 (VRLDNM): Likewise.
23741 (RLNM): New polymorphic function entry.
23742 (VRLWMI): New monomorphic function entry.
23743 (VRLDMI): Likewise.
23744 (RLMI): New polymorphic function entry.
23745 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23746 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23747 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23750 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23753 * dwarf2out.c (field_byte_offset): Restore the
23754 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23755 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23756 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23759 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23762 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23764 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23767 * dwarf2out.c (add_data_member_location_attribute): For constant
23768 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23769 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23770 DW_AT_byte_size attributes.
23772 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23774 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23775 after forcing to constant memory when the code model is medium.
23777 2017-01-17 Julia Koval <julia.koval@intel.com>
23780 * config/i386/avx512fintrin.h
23781 (_mm512_i32gather_ps): Change __addr type to void const*.
23782 (_mm512_mask_i32gather_ps): Ditto.
23783 (_mm512_i32gather_pd): Ditto.
23784 (_mm512_mask_i32gather_pd): Ditto.
23785 (_mm512_i64gather_ps): Ditto.
23786 (_mm512_mask_i64gather_ps): Ditto.
23787 (_mm512_i64gather_pd): Ditto.
23788 (_mm512_mask_i64gather_pd): Ditto.
23789 (_mm512_i32gather_epi32): Ditto.
23790 (_mm512_mask_i32gather_epi32): Ditto.
23791 (_mm512_i32gather_epi64): Ditto.
23792 (_mm512_mask_i32gather_epi64): Ditto.
23793 (_mm512_i64gather_epi32): Ditto.
23794 (_mm512_mask_i64gather_epi32): Ditto.
23795 (_mm512_i64gather_epi64): Ditto.
23796 (_mm512_mask_i64gather_epi64): Ditto.
23797 (_mm512_i32scatter_ps): Change __addr type to void*.
23798 (_mm512_mask_i32scatter_ps): Ditto.
23799 (_mm512_i32scatter_pd): Ditto.
23800 (_mm512_mask_i32scatter_pd): Ditto.
23801 (_mm512_i64scatter_ps): Ditto.
23802 (_mm512_mask_i64scatter_ps): Ditto.
23803 (_mm512_i64scatter_pd): Ditto.
23804 (_mm512_mask_i64scatter_pd): Ditto.
23805 (_mm512_i32scatter_epi32): Ditto.
23806 (_mm512_mask_i32scatter_epi32): Ditto.
23807 (_mm512_i32scatter_epi64): Ditto.
23808 (_mm512_mask_i32scatter_epi64): Ditto.
23809 (_mm512_i64scatter_epi32): Ditto.
23810 (_mm512_mask_i64scatter_epi32): Ditto.
23811 (_mm512_i64scatter_epi64): Ditto.
23812 (_mm512_mask_i64scatter_epi64): Ditto.
23813 * config/i386/avx512pfintrin.h
23814 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23815 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23816 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23817 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23818 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23819 (_mm512_prefetch_i32scatter_ps): Ditto.
23820 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23821 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23822 (_mm512_prefetch_i64scatter_pd): Ditto.
23823 (_mm512_prefetch_i64scatter_ps): Ditto.
23824 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23825 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23826 * config/i386/avx512vlintrin.h
23827 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23828 (_mm_mmask_i32gather_ps): Ditto.
23829 (_mm256_mmask_i32gather_pd): Ditto.
23830 (_mm_mmask_i32gather_pd): Ditto.
23831 (_mm256_mmask_i64gather_ps): Ditto.
23832 (_mm_mmask_i64gather_ps): Ditto.
23833 (_mm256_mmask_i64gather_pd): Ditto.
23834 (_mm_mmask_i64gather_pd): Ditto.
23835 (_mm256_mmask_i32gather_epi32): Ditto.
23836 (_mm_mmask_i32gather_epi32): Ditto.
23837 (_mm256_mmask_i32gather_epi64): Ditto.
23838 (_mm_mmask_i32gather_epi64): Ditto.
23839 (_mm256_mmask_i64gather_epi32): Ditto.
23840 (_mm_mmask_i64gather_epi32): Ditto.
23841 (_mm256_mmask_i64gather_epi64): Ditto.
23842 (_mm_mmask_i64gather_epi64): Ditto.
23843 (_mm256_i32scatter_ps): Change __addr type to void*.
23844 (_mm256_mask_i32scatter_ps): Ditto.
23845 (_mm_i32scatter_ps): Ditto.
23846 (_mm_mask_i32scatter_ps): Ditto.
23847 (_mm256_i32scatter_pd): Ditto.
23848 (_mm256_mask_i32scatter_pd): Ditto.
23849 (_mm_i32scatter_pd): Ditto.
23850 (_mm_mask_i32scatter_pd): Ditto.
23851 (_mm256_i64scatter_ps): Ditto.
23852 (_mm256_mask_i64scatter_ps): Ditto.
23853 (_mm_i64scatter_ps): Ditto.
23854 (_mm_mask_i64scatter_ps): Ditto.
23855 (_mm256_i64scatter_pd): Ditto.
23856 (_mm256_mask_i64scatter_pd): Ditto.
23857 (_mm_i64scatter_pd): Ditto.
23858 (_mm_mask_i64scatter_pd): Ditto.
23859 (_mm256_i32scatter_epi32): Ditto.
23860 (_mm256_mask_i32scatter_epi32): Ditto.
23861 (_mm_i32scatter_epi32): Ditto.
23862 (_mm_mask_i32scatter_epi32): Ditto.
23863 (_mm256_i32scatter_epi64): Ditto.
23864 (_mm256_mask_i32scatter_epi64): Ditto.
23865 (_mm_i32scatter_epi64): Ditto.
23866 (_mm_mask_i32scatter_epi64): Ditto.
23867 (_mm256_i64scatter_epi32): Ditto.
23868 (_mm256_mask_i64scatter_epi32): Ditto.
23869 (_mm_i64scatter_epi32): Ditto.
23870 (_mm_mask_i64scatter_epi32): Ditto.
23871 (_mm256_i64scatter_epi64): Ditto.
23872 (_mm256_mask_i64scatter_epi64): Ditto.
23873 (_mm_i64scatter_epi64): Ditto.
23874 (_mm_mask_i64scatter_epi64): Ditto.
23875 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
23876 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
23877 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
23878 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
23879 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
23880 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
23881 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
23882 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
23883 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
23884 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
23885 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
23886 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
23887 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
23888 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
23889 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
23890 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
23891 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
23892 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
23893 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
23894 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
23895 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
23896 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
23897 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
23898 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23899 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23900 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23901 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23902 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23903 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23904 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23905 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23906 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23907 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23908 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23909 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23910 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23911 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23912 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23913 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23914 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23915 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23916 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23917 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23918 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23919 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23920 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23921 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23922 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23923 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23924 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23925 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23926 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23927 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23928 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23929 definitions accordingly.
23931 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23932 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23935 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23938 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23941 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23942 hard regs for inner regno.
23944 2017-01-17 Martin Liska <mliska@suse.cz>
23947 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23948 assumption and add comment.
23950 2017-01-17 Nathan Sidwell <nathan@acm.org>
23952 * ipa-visibility.c (localize_node): New function, broken out of ...
23953 (function_and_variable_visibility): ... here. Call it.
23955 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23957 PR middle-end/77445
23958 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23959 correctly set frequency of oudgoing edge.
23960 (duplicate_thread_path): Fix profile updating.
23962 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23965 * configure.ac: Add GCC_BASE_VER.
23966 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23967 version from BASE-VER file.
23968 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23969 (gcc.o): Depend on $(BASEVER).
23970 * common.opt (dumpfullversion): New option.
23971 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23972 * doc/invoke.texi: Document -dumpfullversion.
23973 * doc/install.texi: Document --with-gcc-major-version-only.
23974 * configure: Regenerated.
23976 2017-01-17 Richard Biener <rguenther@suse.de>
23978 PR tree-optimization/71433
23979 * tree-vrp.c (register_new_assert_for): Merge same asserts
23980 on all incoming edges.
23981 (process_assert_insertions_for): Handle insertions at the
23984 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23986 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23987 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23989 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23992 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23995 2017-01-17 Alan Modra <amodra@gmail.com>
23998 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23999 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24000 symbolic stack limit when pic.
24002 2017-01-16 Martin Sebor <msebor@redhat.com>
24004 PR tree-optimization/78608
24005 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24007 2017-01-16 Jeff Law <law@redhat.com>
24010 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24011 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24012 for several include directories that may be relative to sysroot.
24013 * config/i386/x-mingw32 (gplus_includedir): Define.
24014 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24015 (native_system_includedir): Likewise.
24016 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24017 override if TARGET_SYSTEM_ROOT is defined.
24018 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24020 PR tree-optimization/79090
24021 PR tree-optimization/33562
24022 PR tree-optimization/61912
24023 PR tree-optimization/77485
24024 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24025 and computed trims into the dump file.
24027 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24029 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24031 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24034 * gimplify.c (gimplify_init_constructor): If want_value and
24035 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24039 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24040 sequence. Formatting fixes.
24041 (doloop_optimize): Formatting fixes.
24044 * gcc.c (debug_level_greater_than_spec_func): New function.
24045 (static_spec_functions): Add debug-level-gt spec function.
24046 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24048 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24049 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24050 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24051 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24052 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24053 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24055 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24057 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24058 QImode fixups to general and mask registers only.
24060 2017-01-16 Carl Love <cel@us.ibm.com>
24062 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24063 for built-in functions
24064 vector signed char vec_nabs (vector signed char)
24065 vector signed short vec_nabs (vector signed short)
24066 vector signed int vec_nabs (vector signed int)
24067 vector signed long long vec_nabs (vector signed long long)
24068 vector float vec_nabs (vector float)
24069 vector double vec_nabs (vector double)
24070 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24072 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24073 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24074 * doc/extend.texi: Update the documentation file for the new built-in
24077 2017-01-16 Martin Sebor <msebor@redhat.com>
24079 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24082 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24084 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24085 UNSPEC_VSX__XXSPLTD to require special splat handling.
24087 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24090 * system.h: Poison strndup.
24092 2017-01-16 Alan Modra <amodra@gmail.com>
24095 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24098 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24100 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24102 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24104 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24105 call recog here. Assert that INSN_CODE (insn) is non-negative.
24107 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24110 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24112 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24113 in the currently scheduled RTL fragment.
24115 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24117 PR rtl-optimization/78751
24118 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24121 2017-01-14 Jeff Law <law@redhat.com>
24123 PR tree-optimization/79090
24124 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24125 variable length stores.
24126 (compute_trims): Delete dead assignment to *trim_tail.
24127 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24130 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24132 PR rtl-optimization/78626
24133 PR rtl-optimization/78727
24134 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24135 of a block, and split such blocks after everything else is finished.
24137 2017-01-14 Alan Modra <amodra@gmail.com>
24140 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24141 target legitimate_combined_insn.
24142 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24143 (rs6000_legitimate_combined_insn): New function.
24144 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24146 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24147 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24148 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24150 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24152 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24154 2017-01-13 Jeff Law <law@redhat.com>
24156 PR tree-optimization/33562
24157 PR tree-optimization/61912
24158 PR tree-optimization/77485
24159 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24161 (delete_dead_assignment): Likewise.
24162 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24163 statement to delete_dead_call and delete_dead_assignment.
24165 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24168 * substring-locations.c (format_warning_va): Strengthen case 1 so
24169 that both endpoints of the substring must be within the format
24170 range for just the substring to be printed.
24172 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24174 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24175 * config/i386/i386.c (ix86_target_string): Add missing options
24176 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24177 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24178 flags_other and ix86_target_other to flags2_other. Display unknown
24180 (ix86_valid_target_attribute_inner_p): Add missing options and
24181 reorder options by implied ISAs, as in ix86_target_string.
24183 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24185 * hash-table.h (hash_table::too_empty_p): New function.
24186 (hash_table::expand): Use it.
24187 (hash_table::traverse): Likewise.
24188 (hash_table::empty_slot): Use sizeof (value_type) instead of
24189 sizeof (PTR) to convert bytes to elements. Shrink the table
24190 if the current size is excessive for the current number of
24193 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24195 * ira-costs.c (record_reg_classes): Break from the inner loop
24196 early once alt_fail is known to be true. Update outer loop
24197 handling accordingly.
24199 2017-01-13 Jeff Law <law@redhat.com>
24201 * tree-ssa-dse.c (decrement_count): New function.
24202 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24203 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24204 when we know the partially dead statement is a mem* function.
24206 PR tree-optimization/61912
24207 PR tree-optimization/77485
24208 * tree-ssa-dse.c: Include expr.h.
24209 (maybe_trim_constructor_store): New function.
24210 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24212 PR tree-optimization/33562
24213 PR tree-optimization/61912
24214 PR tree-optimization/77485
24215 * doc/invoke.texi: Document new dse-max-object-size param.
24216 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24217 * tree-ssa-dse.c: Include params.h.
24218 (dse_store_status): New enum.
24219 (initialize_ao_ref_for_dse): New, partially extracted from
24221 (valid_ao_ref_for_dse, normalize_ref): New.
24222 (setup_live_bytes_from_ref, compute_trims): Likewise.
24223 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24224 (maybe_trim_partially_dead_store): Likewise.
24225 (maybe_trim_complex_store): Likewise.
24226 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24227 Track what bytes live from the original store. Return tri-state
24228 for dead, partially dead or live.
24229 (dse_dom_walker): Add constructor, destructor and new private members.
24230 (delete_dead_call, delete_dead_assignment): New extracted from
24232 (dse_optimize_stmt): Make a member of dse_dom_walker.
24233 Use initialize_ao_ref_for_dse.
24235 PR tree-optimization/33562
24236 PR tree-optimization/61912
24237 PR tree-optimization/77485
24238 * sbitmap.h (bitmap_count_bits): Prototype.
24239 (bitmap_clear_range, bitmap_set_range): Likewise.
24240 * sbitmap.c (bitmap_clear_range): New function.
24241 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24243 2017-01-13 Martin Liska <mliska@suse.cz>
24246 * function.c (set_cfun): Add new argument force.
24247 * function.h (set_cfun): Likewise.
24248 * ipa-inline-transform.c (inline_call): Use the function when
24249 strict alising from is dropped for function we inline to.
24251 2017-01-13 Richard Biener <rguenther@suse.de>
24253 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24254 for dumping GIMPLE INTEGER_CSTs.
24256 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24258 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24259 to 201112L since C++17.
24261 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24264 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24265 if -fsanitize=kernel-address is present.
24267 2017-01-13 Richard Biener <rguenther@suse.de>
24269 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24270 as _Literal ( type ) number in case usual suffixes do not
24271 preserve all information.
24273 2017-01-13 Richard Biener <rguenther@suse.de>
24275 PR tree-optimization/77283
24276 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24277 and ssa-iterators.h.
24278 (is_feasible_trace): Implement a cost model based on joiner
24281 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24284 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24285 char or short to __float128/_Float128 directly.
24287 2017-01-12 Martin Sebor <msebor@redhat.com>
24289 to -Wformat-overflow.
24290 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24291 (min_bytes_remaining): Same.
24292 (get_string_length): Same.
24293 (format_string): Same.
24294 (format_directive): Same.
24296 (pass_sprintf_length::handle_gimple_call): Same.
24298 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24300 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24301 info.nowrite calls with no lhs that can't throw. Return bool
24302 whether gsi_remove has been called or not.
24303 (pass_sprintf_length::handle_gimple_call): Return bool whether
24304 try_substitute_return_value called gsi_remove. Formatting fix.
24305 (pass_sprintf_length::execute): Don't use gsi_remove if
24306 handle_gimple_call returned true.
24309 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24310 be removed due to side-effects, don't remove following barrier nor
24311 turn the successor edge into fallthru edge.
24313 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24316 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24317 element-reversing loads and stores as not swappable.
24319 2017-01-12 Nathan Sidwell <nathan@acm.org>
24320 Nicolai Stange <nicstange@gmail.com>
24322 * combine.c (try_combine): Don't ignore result of overlap checking
24323 loop. Combine overlap & asm check into single loop.
24325 2017-01-12 Richard Biener <rguenther@suse.de>
24327 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24328 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24330 2017-01-12 Richard Biener <rguenther@suse.de>
24332 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24333 and TS_TARGET_OPTION directly derive from TS_BASE.
24334 * tree-core.h (tree_optimization_option): Derive from tree_base.
24335 (tree_target_option): Likewise.
24337 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24339 * config/i386/i386.c (memory_address_length): Increase len
24340 only when rip_relative_addr_p returns false.
24342 2017-01-11 Julia Koval <julia.koval@intel.com>
24344 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24345 (OPTION_MASK_ISA_SGX_SET): New.
24346 (ix86_handle_option): Handle OPT_msgx.
24347 * config.gcc: Added sgxintrin.h.
24348 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24349 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24350 * config/i386/i386.c (ix86_target_string): Add -msgx.
24352 (ix86_option_override_internal): Handle new options.
24353 (ix86_valid_target_attribute_inner_p): Add sgx.
24354 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24355 * config/i386/i386.opt: Add msgx.
24356 * config/i386/sgxintrin.h: New file.
24357 * config/i386/x86intrin.h: Add sgxintrin.h.
24359 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24362 * fold-const.c (maybe_nonzero_address): Return 1 for function
24364 (tree_single_nonzero_warnv_p): Don't handle function local objects
24368 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24371 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24374 * opts.c: Include "spellcheck.h"
24375 (struct string_fragment): New struct.
24376 (struct edit_distance_traits<const string_fragment &>): New
24378 (get_closest_sanitizer_option): New function.
24379 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24381 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24383 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24385 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24386 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24387 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24388 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24389 for initial die_offset if dwarf_split_debug_info.
24390 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24391 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24393 (output_skeleton_debug_sections): Formatting fix. Use
24394 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24395 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24397 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24399 * config/arm/cortex-a53.md: Add bypasses for
24400 cortex_a53_r2f_cvt.
24401 (cortex_a53_r2f): Only use for transfers.
24402 (cortex_a53_f2r): Likewise.
24403 (cortex_a53_r2f_cvt): Add reservation for conversions.
24404 (cortex_a53_f2r_cvt): Likewise.
24406 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24408 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24409 to all inlined functions, change static to extern.
24411 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24414 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24416 (arm_assemble_integer): Likewise.
24418 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24420 * config.gcc: Use new awk script to check CPU, FPU and architecture
24421 parameters for --with-... options.
24422 * config/arm/parsecpu.awk: New file
24423 * config/arm/arm-cpus.in: New file.
24424 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24426 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24428 * config/arm/t-arm: Update dependency rules.
24429 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24430 of processing .def files.
24431 * config/arm/genopt.sh: Deleted.
24432 * config/arm/gentune.sh: Deleted.
24433 * config/arm/arm-cores.def: Deleted.
24434 * config/arm/arm-arches.def: Deleted.
24435 * config/arm/arm-fpus.def: Deleted.
24436 * config/arm/arm-tune.md: Regenerated.
24437 * config/arm/arm-tables.opt: Regenerated.
24438 * config/arm/arm-cpu.h: New generated file.
24439 * config/arm/arm-cpu-data.h: New generated file.
24440 * config/arm/arm-cpu-cdata.h: New generated file.
24442 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24445 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24447 (input_varpool_node): Unpack dynamically_initialized bit.
24449 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24451 PR rtl-optimization/79032
24452 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24453 the alignment of the adjusted memory reference against that of MODE,
24454 instead of the alignment of the original memory reference.
24456 2017-01-11 Martin Jambor <mjambor@suse.cz>
24458 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24460 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24461 decorated functions.
24463 2017-01-11 Richard Biener <rguenther@suse.de>
24465 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24466 set range/nonnull info for PHI results. Do not set it on
24467 stmts marked for removal.
24469 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24471 * expr.c (store_field): In the bitfield case, fetch the return value
24472 from the registers before applying a single big-endian adjustment.
24473 Always do a final load for a BLKmode value not larger than a word.
24475 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24478 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24479 that we correctly handle column numbers greater than
24480 LINE_MAP_MAX_COLUMN_NUMBER.
24482 2017-01-10 Martin Sebor <msebor@redhat.com>
24484 PR middle-end/78245
24485 * gimple-ssa-sprintf.c (get_destination_size): Call
24486 {init,fini}object_sizes.
24487 * tree-object-size.c (addr_object_size): Adjust.
24488 (pass_through_call): Adjust.
24489 (pass_object_sizes::execute): Adjust.
24490 * tree-object-size.h (fini_object_sizes): Declare.
24492 2017-01-10 Martin Sebor <msebor@redhat.com>
24494 PR tree-optimization/78775
24495 * builtins.c (get_size_range): Move...
24496 * calls.c: ...to here.
24497 (alloc_max_size): Accept zero argument.
24498 (operand_signed_p): Remove.
24499 (maybe_warn_alloc_args_overflow): Call get_size_range.
24500 * calls.h (get_size_range): Declare.
24502 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24504 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24505 from TI's devices.csv file as of September 2016.
24506 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24508 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24510 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24511 * doc/invoke.texi: Likewise.
24512 * doc/md.texi: Likewise.
24513 * doc/objc.texi: Likewise.
24515 2017-01-10 Joshua Conner <joshconner@google.com>
24517 * config/arm/fuchsia-elf.h: New file.
24518 * config/fuchsia.h: New file.
24519 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24520 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24522 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24524 2016-01-10 Richard Biener <rguenther@suse.de>
24526 PR tree-optimization/79034
24527 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24528 Propagate out degenerate PHIs in the joiner.
24530 2017-01-10 Martin Liska <mliska@suse.cz>
24532 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24533 (sort_congruence_classes_by_decl_uid): Likewise.
24534 (sort_congruence_class_groups_by_decl_uid): Likewise.
24535 (sem_item_optimizer::merge_classes): Sort class, groups in these
24536 classes and members in the groups by DECL_UID of declarations.
24537 This would make merge operations stable.
24539 2017-01-10 Martin Liska <mliska@suse.cz>
24541 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24542 usage of m_classes_vec.
24543 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24544 (sem_item_optimizer::get_group_by_hash): Likewise.
24545 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24546 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24547 (sem_item_optimizer::verify_classes): Likewise.
24548 (sem_item_optimizer::process_cong_reduction): Likewise.
24549 (sem_item_optimizer::dump_cong_classes): Likewise.
24550 (sem_item_optimizer::merge_classes): Likewise.
24551 * ipa-icf.h (congruence_class_hash): Rename from
24552 congruence_class_group_hash. Remove declaration of m_classes_vec.
24554 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24556 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24557 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24558 * config.gcc: Add avx512vpopcntdqintrin.h.
24559 * config/i386/avx512vpopcntdqintrin.h: New.
24560 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24561 * config/i386/i386-builtin-types.def: Add new types.
24562 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24563 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24564 __builtin_ia32_vpopcountq_v8di_mask): New.
24565 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24566 __AVX512VPOPCNTDQ__.
24567 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24568 (PTA_AVX512VPOPCNTDQ): Define.
24569 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24570 TARGET_AVX512VPOPCNTDQ_P): Define.
24571 * config/i386/i386.opt: Add mavx512vpopcntdq.
24572 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24573 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24575 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24577 PR middle-end/77484
24578 * predict.def (PRED_CALL): Set to 67.
24580 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24582 * expr.c (store_field): In the bitfield case, if the value comes from
24583 a function call and is of an aggregate type returned in registers, do
24584 not modify the field mode; extract the value in all cases if the mode
24585 is BLKmode and the size is not larger than a word.
24587 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24590 * config/i386/cpuid.h: Fix undefined behavior.
24592 2017-01-04 Jeff Law <law@redhat.com>
24594 PR tree-optimization/79007
24595 PR tree-optimization/67955
24596 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24597 conservative for pt.null when flag_non_call_exceptions is on.
24599 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24601 PR translation/79019
24602 PR translation/79020
24603 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24604 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24605 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24607 * config/avr/avr.opt (maccumulate-args): Likewise.
24608 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24609 * common.opt (freport-bug): Likewise.
24610 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24611 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24612 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24613 translatable string.
24614 * config/i386/i386.c (function_value_32): Likewise.
24615 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24616 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24618 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24619 * common/config/msp430/msp430-common.c (msp430_handle_option):
24621 * symtab.c (symtab_node::verify_base): Likewise.
24622 * opts.c (set_debug_level): Likewise.
24623 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24624 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24625 missing whitespace to translatable strings.
24626 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24627 * config/sh/superh.h: Likewise.
24628 * config/i386/xopintrin.h: Likewise.
24629 * config/i386/znver1.md: Likewise.
24630 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24631 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24632 * double-int.h (struct double_int): Likewise.
24633 * double-int.c (div_and_round_double): Likewise.
24634 * wide-int.cc: Likewise.
24635 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24636 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24637 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24638 (crossjumps_occurred): ... this.
24639 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24642 PR tree-optimization/78899
24643 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24644 returning bool return struct loop *, NULL for failure and the new
24646 (versionable_outer_loop_p): Don't version outer loop if it has
24647 dont_vectorized bit set.
24648 (tree_if_conversion): When versioning outer loop, ensure
24649 tree_if_conversion is performed also on the inner loop of the
24650 non-vectorizable outer loop copy.
24651 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24652 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24653 prevent vectorization of it.
24654 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24655 the outer loop vectorization of the non-scalar version is attempted
24656 before vectorization of the inner loop in scalar version. If
24657 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24658 vectorization of its inner loop.
24659 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24660 has 2 inner loops, rename also on edges from bb whose single pred
24661 is outer_loop->header. Fix typo in function comment.
24663 2017-01-09 Martin Sebor <msebor@redhat.com>
24666 * asan.c (asan_emit_stack_protection): Increase local buffer size
24667 to avoid snprintf truncation warning.
24669 2017-01-09 Andrew Pinski <apinski@cavium.com>
24671 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24672 to reference thunderx2t99 for the tuning structure
24673 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24675 (thunderx2t99_extra_costs): This.
24676 * config/aarch64/aarch64-tune.md: Regenerate.
24677 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24678 (vulcan_addrcost_table): This.
24679 (vulcan_regmove_cost): Rename to ...
24680 (thunderx2t99_regmove_cost): This.
24681 (vulcan_vector_cost): Rename to ...
24682 (thunderx2t99_vector_cost): this.
24683 (vulcan_branch_cost): Rename to ...
24684 (thunderx2t99_branch_cost): This.
24685 (vulcan_tunings): Rename to ...
24686 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24687 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24689 2017-01-09 Martin Jambor <mjambor@suse.cz>
24693 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24694 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24695 (propagate_vr_accross_jump_function): Use the above function for all
24696 value range computations for pass-through jump functions and type
24697 converasion from explicit value range values.
24698 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24699 parameters from TYPE_ARG_TYPES.
24700 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24701 (ipa_write_node_info): Stream type of the actual argument.
24702 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24704 2017-01-09 Martin Liska <mliska@suse.cz>
24707 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24708 (lookup_compiler): Do not show error message with have_E.
24710 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24712 PR tree-optimization/78938
24713 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24714 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24715 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24716 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24719 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24721 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24724 2017-01-09 Richard Biener <rguenther@suse.de>
24726 PR tree-optimization/78997
24727 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24728 name condition properly.
24730 2017-01-09 Richard Biener <rguenther@suse.de>
24733 * dwarf2out.c (is_cxx): New overload with context.
24734 (is_naming_typedef_decl): Use it.
24736 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24738 * invoke.texi (Option Summary): Correct spacing in option lists
24739 and add line breaks to fix over-long lines.
24741 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24743 PR middle-end/17660
24745 * extend.texi (Common Variable Attributes): Add xref to GCC
24746 Internals manual to explain mode attribute keywords.
24748 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24751 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24752 and Preprocessor Options.
24753 (Options for Linking): Document -pthread here....
24754 (RS/6000 and PowerPC Options): ...not here.
24755 (Solaris 2 Options): ...or here.
24756 * doc/cppopts.texi: Document -pthread.
24758 2017-01-08 Martin Sebor <msebor@redhat.com>
24760 PR middle-end/77708
24761 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24762 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24763 New member functions.
24764 (format_directive): Used them.
24766 (pass_sprintf_length::handle_gimple_call): Same.
24767 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24768 to avoid truncation for any argument.
24769 (extract_affine_mul): Same.
24770 * tree.c (get_file_function_name): Same.
24772 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24774 PR middle-end/77484
24775 * predict.def (PRED_INDIR_CALL): Set to 86.
24777 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24779 PR preprocessor/54124
24780 * doc/cppopts.texi: Reformat -d subtable to list the full name
24781 of the options. Add cross-reference to the docs for the general
24782 compiler -d options.
24783 * doc/invoke.texi (Developer Options): Add cross-reference to the
24784 preprocessor-specific -d option documentation.
24786 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24788 PR preprocessor/13498
24789 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24790 redudant material, and reflect new command-line options.
24791 (System Headers): Likewise.
24793 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24795 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24796 -isystem, and -idirafter. Copy-edit.
24797 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24798 default for -ftrack-macro-expansion. Delete obsolete and
24799 badly-formatted implementation details about -fdebug-cpp output.
24800 * doc/cppwarnopts.texi: Copy-edit.
24802 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24805 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24806 that the transition from a max line width >= 1<<10 to narrower
24807 lines works correctly.
24809 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24811 * doc/options.texi (PerFunction): New.
24812 * opt-functions.awk (switch_flags): Map both Optimization and
24813 PerFunction to CL_OPTIMIZATION.
24814 * opth-gen.awk: Test for PerFunction flag along with
24816 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24817 it only when the latter is present. Skip those that don't in
24818 the hash function generator.
24819 * common.opt (fvar-tracking): Mark as PerFunction instead of
24821 (fvar-tracking-assignments): Likewise.
24822 (fvar-tracking-assignments-toggle): Likewise.
24823 (fvar-tracking-uninit): Likewise.
24825 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24827 PR translation/79018
24828 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24831 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24834 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24835 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24836 TARGET_LONG_JUMP_TABLE_OFFSETS.
24837 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24838 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24839 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24840 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24841 * config/m68k/m68k.md (tablejump expander): Likewise.
24842 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24843 TARGET_LONG_JUMP_TABLE_OFFSETS.
24844 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24845 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24847 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24848 David Holsgrove <david.holsgrove@xilinx.com>
24850 * common/config/microblaze/microblaze-common.c
24851 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24852 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24854 * config/microblaze/microblaze.c (microblaze_must_save_register)
24855 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24857 (microblaze_eh_return): New function.
24858 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24859 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24860 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24861 * config/microblaze/microblaze.md (eh_return): New pattern.
24863 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24865 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
24866 GCC_DIAGNOSTIC_STRINGIFY): Define.
24868 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
24870 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24872 * config/arm/arm.md (<mcrr>): New.
24874 * config/arm/arm.c (arm_arch5te): New.
24875 (arm_option_override): Set arm_arch5te.
24876 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
24878 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
24879 (arm_mcrr_qualifiers): ... this. New.
24880 (MRRC_QUALIFIERS): Define to...
24881 (arm_mrrc_qualifiers): ... this. New.
24882 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
24884 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
24885 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
24886 (MRRCI, mrrc, MRRC): New.
24887 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
24888 VUNSPEC_MRRC2): New.
24890 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24892 * config/arm/arm.md (<mcr>): New.
24894 * config/arm/arm.c (arm_coproc_builtin_available): Add
24895 support for mcr, mrc, mcr2 and mrc2.
24896 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
24897 (arm_mcr_qualifiers): ... this. New.
24898 (MRC_QUALIFIERS): Define to ...
24899 (arm_mrc_qualifiers): ... this. New.
24900 (MCR_QUALIFIERS): Define to ...
24901 (arm_mcr_qualifiers): ... this. New.
24902 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24904 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24905 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24906 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24907 VUNSPEC_MRC2): New.
24909 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24911 * config/arm/arm.md (*ldc): New.
24915 * config/arm/arm.c (arm_coproc_builtin_available): Add
24916 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24917 (arm_coproc_ldc_stc_legitimate_address): New.
24918 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24919 'qualifier_const_pointer'.
24920 (LDC_QUALIFIERS): Define to...
24921 (arm_ldc_qualifiers): ... this. New.
24922 (STC_QUALIFIERS): Define to...
24923 (arm_stc_qualifiers): ... this. New.
24924 * config/arm/arm-protos.h
24925 (arm_coproc_ldc_stc_legitimate_address): New.
24926 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24927 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24928 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24929 stc2, stcl, stc2l): New.
24930 * config/arm/constraints.md (Uz): New.
24931 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24932 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24933 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24934 VUNSPEC_STC2L): New.
24936 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24938 * config/arm/arm.md (<cdp>): New.
24939 * config/arm/arm.c (neon_const_bounds): Rename this ...
24940 (arm_const_bounds): ... this.
24941 (arm_coproc_builtin_available): New.
24942 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24943 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24944 (CDP_QUALIFIERS): Define to...
24945 (arm_cdp_qualifiers): ... this. New.
24947 (arm_expand_builtin_args): Add case for 6 arguments.
24948 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24949 (arm_const_bounds): ... this.
24950 (arm_coproc_builtin_available): New.
24951 * config/arm/arm_acle.h (__arm_cdp): New.
24953 * config/arm/arm_acle_builtins.def (cdp): New.
24955 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24956 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24957 'arm_const_bounds'.
24958 * config/arm/types.md (coproc): New.
24959 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24960 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24961 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24962 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24964 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24966 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24967 (UBINOP_QUALIFIERS): New.
24969 (acle_builtin_data): New. Change comment.
24970 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24971 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24972 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24973 arm_acle_builtins.def.
24974 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24975 (arm_init_acle_builtins): New.
24976 (CRC32_BUILTIN): Remove.
24977 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24978 crc32cb, crc32ch and crc32cw.
24979 (arm_init_crc32_builtins): Remove.
24980 (arm_init_builtins): Use arm_init_acle_builtins rather
24981 than arm_init_crc32_builtins.
24982 (arm_expand_acle_builtin): New.
24983 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24984 * config/arm/arm_acle_builtins.def: New.
24986 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24988 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24989 (arm_builtin_datum): ... this.
24990 (arm_init_neon_builtin): Rename to ...
24991 (arm_init_builtin): ... this. Add a new parameters PREFIX
24992 and USE_SIG_IN_NAME.
24993 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24994 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24995 'arm_builtin_datum'.
24996 (arm_init_vfp_builtins): Likewise.
24997 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24998 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24999 (arm_expand_neon_args): Rename to ...
25000 (arm_expand_builtin_args): ... this. Rename builtin_arg
25001 enum values and differentiate between ARG_BUILTIN_MEMORY
25002 and ARG_BUILTIN_NEON_MEMORY.
25003 (arm_expand_neon_builtin_1): Rename to ...
25004 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25005 values, arm_expand_builtin_args and add bool parameter NEON.
25006 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25007 (arm_expand_vfp_builtin): Likewise.
25008 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25010 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25012 PR middle-end/77484
25013 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25014 * predict.c (tree_estimate_probability_bb): Reverse direction of
25015 polymorphic call predictor.
25017 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25019 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25020 (determine_pass_name_match): ...this new function and...
25021 (should_skip_pass_p): ...this new function.
25023 2017-01-06 Nathan Sidwell <nathan@acm.org>
25025 * ipa-visibility.c (function_and_variable_visibility): Reformat
25026 comments and long lines. Remove extrneous if.
25027 * symtab.c (symtab_node::make_decl_local): Fix code format.
25028 (symtab_node::set_section_for_node): Fix comment typo.
25030 2017-01-06 Martin Liska <mliska@suse.cz>
25033 * lra-constraints.c: Rename invariant to lra_invariant.
25034 * predict.c (set_even_probabilities): Initialize e to NULL.
25036 2017-01-05 Martin Sebor <msebor@redhat.com>
25038 PR tree-optimization/78910
25039 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25040 (format_integer): Correct off-by-one error in the handling
25041 of precision with negative numbers in signed conversions..
25043 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25045 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25047 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25049 PR tree-optimization/71016
25050 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25051 factor_out_conditional_conversion. Formatting fix.
25052 (factor_out_conditional_conversion): Add cond_stmt argument.
25053 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25054 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25057 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25059 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25060 read-rtl-function.o, and selftest-rtl.o.
25061 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25062 (selftest::aarch64_test_loading_full_dump): New function.
25063 (selftest::aarch64_run_selftests): New function.
25064 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25065 selftest::aarch64_run_selftests.
25066 * config/i386/i386.c
25067 (selftest::ix86_test_loading_dump_fragment_1): New function.
25068 (selftest::ix86_test_loading_call_insn): New function.
25069 (selftest::ix86_test_loading_full_dump): New function.
25070 (selftest::ix86_test_loading_unspec): New function.
25071 (selftest::ix86_run_selftests): Call the new functions.
25072 * emit-rtl.c (maybe_set_max_label_num): New function.
25073 * emit-rtl.h (maybe_set_max_label_num): New decl.
25074 * function.c (instantiate_decls): Guard call to
25075 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25076 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25078 * gensupport.c (gen_reader::gen_reader): Pass "false"
25079 for new "compact" param of rtx_reader.
25080 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25081 rather than an empty string for NULL strings.
25082 * read-md.c: Potentially include config.h rather than bconfig.h.
25083 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25084 (have_error): New global, copied from errors.c.
25085 (md_reader::read_name): Rename to...
25086 (md_reader::read_name_1): ...this, adding "out_loc" param,
25087 and converting "missing name or number" to returning false, rather
25089 (md_reader::read_name): Reimplement in terms of read_name_1.
25090 (md_reader::read_name_or_nil): New function.
25091 (md_reader::read_string): Handle "(nil)" by returning NULL.
25092 (md_reader::md_reader): Add new param "compact".
25093 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25094 (md_reader::read_file): New method.
25095 * read-md.h (md_reader::md_reader): Add new param "compact".
25096 (md_reader::read_file): New method.
25097 (md_reader::is_compact): New accessor.
25098 (md_reader::read_name): Convert return type from void to file_location.
25099 (md_reader::read_name_or_nil): New decl.
25100 (md_reader::read_name_1): New decl.
25101 (md_reader::m_compact): New field.
25102 (noop_reader::noop_reader): Pass "false" for new "compact" param
25104 (rtx_reader::rtx_reader): Add new "compact" param.
25105 (rtx_reader::read_rtx_operand): Make virtual and convert return
25106 type from void to rtx.
25107 (rtx_reader::read_until): New decl.
25108 (rtx_reader::handle_any_trailing_information): New virtual function.
25109 (rtx_reader::postprocess): New virtual function.
25110 (rtx_reader::finalize_string): New virtual function.
25111 (rtx_reader::m_in_call_function_usage): New field.
25112 (rtx_reader::m_reuse_rtx_by_id): New field.
25113 * read-rtl-function.c: New file.
25114 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25115 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25116 (selftest::verify_three_block_rtl_cfg): New decl.
25117 * read-rtl-function.h: New file.
25118 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25119 For host, include function.h, memmodel.h, and emit-rtl.h.
25120 (one_time_initialization): New function.
25121 (struct compact_insn_name): New struct.
25122 (compact_insn_names): New array.
25123 (find_code): Handle insn codes in compact dumps.
25124 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25125 (bind_subst_iter_and_attr): Likewise.
25126 (add_condition_to_string): Likewise.
25127 (add_condition_to_rtx): Likewise.
25128 (apply_attribute_uses): Likewise.
25129 (add_current_iterators): Likewise.
25130 (apply_iterators): Likewise.
25131 (initialize_iterators): Guard usage of apply_subst_iterator with
25132 #ifdef GENERATOR_FILE.
25133 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25134 (md_reader::read_mapping): Likewise.
25135 (add_define_attr_for_define_subst): Likewise.
25136 (add_define_subst_attr): Likewise.
25137 (read_subst_mapping): Likewise.
25138 (check_code_iterator): Likewise.
25139 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25141 (one_time_initialization): New function.
25142 (rtx_reader::read_until): New method.
25143 (read_flags): New function.
25144 (parse_reg_note_name): New function.
25145 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25146 Handle reuse_rtx ids.
25147 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25148 Add parsing support for RTL dumps, mirroring the special-cases in
25149 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25150 values, and calling handle_any_trailing_information.
25151 (rtx_reader::read_rtx_operand): Convert return type from void
25152 to rtx, returning return_rtx. Handle case 'e'. Call
25153 finalize_string on XSTR and XTMPL fields.
25154 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25155 "(nil)" values were omitted. Call the postprocess vfunc on the
25157 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25158 class ctor. Initialize m_in_call_function_usage. Call
25159 one_time_initialization.
25160 * rtl-tests.c (selftest::test_uncond_jump): Call
25161 set_new_first_and_last_insn.
25162 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25163 * selftest-rtl.c: New file.
25164 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25165 (selftest::get_insn_by_uid): New decl.
25166 * selftest-run-tests.c (selftest::run_tests): Call
25167 read_rtl_function_c_tests.
25168 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25169 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25172 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25174 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25175 operands in a special way. Assert that pos+len <= mode precision.
25177 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25179 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25180 3 argument Alias with unlimited for the negative form.
25181 (fno-vect-cost-model): Removed.
25183 2017-01-05 Martin Liska <mliska@suse.cz>
25185 * hsa-gen.c (gen_hsa_divmod): New function.
25186 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25188 2017-01-05 Martin Liska <mliska@suse.cz>
25191 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25194 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25196 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25197 small constant length operands.
25199 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25201 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25202 between loop iterations.
25204 2017-01-05 Martin Liska <mliska@suse.cz>
25207 * gimplify.c (gimplify_decl_expr): Compare to
25208 asan_poisoned_variables instread of checking flags.
25209 (gimplify_target_expr): Likewise.
25210 (gimplify_expr): Likewise.
25211 (gimplify_function_tree): Conditionally initialize
25212 asan_poisoned_variables.
25214 2017-01-04 Jeff Law <law@redhat.com>
25216 PR tree-optimizatin/78812
25217 * rtl.h (contains_mem_rtx_p): Prototype.
25218 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25219 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25220 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25221 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25222 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25224 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25226 * input.c (assert_char_at_range): Default-initialize actual_range.
25228 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25230 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25231 to match the caller.
25233 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25235 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25236 insns after final jump in test to emit dummy move.
25238 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25240 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25241 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25243 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25245 * multiple_target.c (create_dispatcher_calls): Init e_next.
25246 * tree-ssa-loop-split.c (split_loop): Init border.
25247 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25250 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25255 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25256 (altivec_register_operand): Do not return true if the operand
25257 contains a SUBREG mixing SImode and SFmode.
25258 (vsx_register_operand): Likewise.
25259 (vsx_reg_sfsubreg_ok): New predicate.
25260 (vfloat_operand): Do not return true if the operand contains a
25261 SUBREG mixing SImode and SFmode.
25262 (vint_operand): Likewise.
25263 (vlogical_operand): Likewise.
25264 (gpc_reg_operand): Likewise.
25265 (int_reg_operand): Likewise.
25266 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25267 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25268 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25270 (rs6000_emit_move_si_sf_subreg): New helper function.
25271 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25272 fixup SUBREGs involving SImode and SFmode.
25273 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25274 numbers for the new peephole2 optimization.
25275 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25276 the GLIBC math library that do AND/IOR/XOR operations on single
25277 precision floating point.
25278 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25279 target macros to say whether we need to avoid SUBREGs mixing
25281 (TARGET_ALLOW_SF_SUBREG): Likewise.
25282 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25283 (UNSPEC_SI_FROM_SF): Likewise.
25284 (iorxor): Change spacing.
25285 (and_ior_xor): New iterator for AND, IOR, and XOR.
25286 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25287 (movdi_from_sf_zero_ext): Likewise.
25288 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25289 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25290 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25291 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25292 (fms<mode>4): Likewise.
25293 (fnma<mode>4): Likewise.
25294 (fnms<mode>4): Likewise.
25295 (nfma<mode>4): Likewise.
25296 (nfms<mode>4): Likewise.
25298 2017-01-04 Marek Polacek <polacek@redhat.com>
25301 * doc/invoke.texi: Document -Wpointer-compare.
25303 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25305 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25308 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25309 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25312 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25315 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25316 documentation of the powerpc_popcntb_ok attribute.
25317 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25318 code to issue warning messages if a requested CPU configuration is
25319 not supported by the binary (assembler and loader) toolchain.
25320 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25321 made to define a built-in function that has been disabled.
25322 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25323 made to define a built-in function that has been disabled.
25324 (altivec_init_builtins): Add comment explaining why definition
25325 of the DST built-in functions is not preceded by an assertion
25326 check. Add assertions to prevent ICE if attempts are made to
25327 define an altivec predicate or an abs* built-in function that has
25329 (htm_init_builtins): Add comment explaining why definition of the
25330 htm built-in functions is not preceded by an assertion check.
25332 2017-01-04 Jeff Law <law@redhat.com>
25334 PR tree-optimizatin/67955
25335 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25336 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25337 the points-to solution does not include pt_null. Use DECL_PT_UID
25340 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25342 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25343 Use gen_int_mode instead of gen_lopwart for const_int operands.
25345 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25347 PR tree-optimization/71563
25348 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25349 out of range value - has low bits known to be zero.
25351 2017-01-04 Alan Modra <amodra@gmail.com>
25353 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25354 * configure: Regenerate.
25355 * config.in: Regenerate.
25357 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25360 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25361 a substring of the message, but strcmp with the whole message. Ifdef
25362 ENABLE_NLS, translate the message first using dgettext.
25364 2017-01-03 Jeff Law <law@redhat.com>
25366 PR tree-optimizatin/78856
25367 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25368 (mark_threaded_blocks): Remove code to truncate thread paths that
25369 cross multiple loop headers. Instead invalidate the cached loop
25370 iteration information and handle case of a thread path walking
25371 into an irreducible region.
25373 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25376 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25377 assertions. Add support for doing the signbit if the IEEE 128-bit
25378 floating point value is in a GPR.
25379 * config/rs6000/rs6000.md (Fsignbit): Delete.
25380 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25381 Update the length attribute if the value is in a GPR.
25382 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25383 the sign or zero extension instruction, since the value is always 0/1.
25384 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25387 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25388 extracting SImode to a GPR register so that we can generate a
25389 store, limit the vector to be in a traditional Altivec register
25390 for the vextuwrx instruction.
25392 2017-01-03 Ian Lance Taylor <iant@google.com>
25394 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25396 2017-01-03 Martin Sebor <msebor@redhat.com>
25398 PR tree-optimization/78696
25399 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25400 precision. Use MPFR for %f for greater fidelity. Correct handling
25402 (pass_sprintf_length::compute_format_length): Set width and precision
25403 specified by asrerisk to void_node for vararg functions.
25404 (try_substitute_return_value): Adjust dump output.
25406 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25408 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25410 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25412 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25413 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25414 -mlra/-mno-lra was passed to the compiler.
25416 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25418 PR rtl-optimization/65618
25419 * emit-rtl.c (try_split): Move initialization of "before" and
25420 "after" to just before the call to emit_insn_after_setloc.
25422 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25424 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25426 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25428 * dwarf2out.c (gen_enumeration_type_die): When
25429 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25431 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25433 PR tree-optimization/78965
25434 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25435 Change first argument from const call_info & to call_info &. For %n
25436 set info.nowrite to false.
25438 PR middle-end/78901
25439 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25440 possibly throwing calls.
25442 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25443 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25444 and fns handling, rather than in a separate case SSA_NAME.
25446 2017-01-02 Jeff Law <law@redhat.com>
25448 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25449 fixes for first_period and second_period variables.
25451 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25454 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25455 (*insvqi_1): New insn pattern.
25456 (*insvqi_1_mem_rex64): Ditto.
25457 (*insvqi_2): Ditto.
25458 (*insvqi_3): Rename from *insvqi.
25460 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25462 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25464 * doc/cfg.texi (Edges): Remove reference to Java.
25465 (Maintaining the CFG): Ditto.
25467 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25469 PR middle-end/77674
25470 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25471 transparent aliases.
25473 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25475 PR middle-end/77484
25476 * predict.def (PRED_CALL): Update hitrate.
25477 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25478 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25479 into direct/indirect/polymorphic variants.
25481 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25483 Update copyright years.
25485 * gcc.c (process_command): Update copyright notice dates.
25486 * gcov-dump.c (print_version): Ditto.
25487 * gcov.c (print_version): Ditto.
25488 * gcov-tool.c (print_version): Ditto.
25489 * gengtype.c (create_file): Ditto.
25490 * doc/cpp.texi: Bump @copying's copyright year.
25491 * doc/cppinternals.texi: Ditto.
25492 * doc/gcc.texi: Ditto.
25493 * doc/gccint.texi: Ditto.
25494 * doc/gcov.texi: Ditto.
25495 * doc/install.texi: Ditto.
25496 * doc/invoke.texi: Ditto.
25498 Copyright (C) 2017 Free Software Foundation, Inc.
25500 Copying and distribution of this file, with or without modification,
25501 are permitted in any medium without royalty provided the copyright
25502 notice and this notice are preserved.