1 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4 * stor-layout.c (mode_for_size_tree): Likewise.
5 (mode_for_array): Update accordingly.
6 (layout_decl): Likewise.
7 (compute_record_mode): Likewise. Only set the mode once.
9 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11 * target.def (get_mask_mode): Change return type to opt_mode.
13 * doc/tm.texi: Regenerate.
14 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15 * targhooks.c (default_get_mask_mode): Likewise.
16 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17 * optabs-query.c (can_vec_mask_load_store_p): Update use of
18 targetm.get_mask_mode.
19 * tree.c (build_truth_vector_type): Likewise.
21 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
23 * machmode.h (mode_for_vector): Return an opt_mode.
24 * stor-layout.c (mode_for_vector): Likewise.
25 (mode_for_int_vector): Update accordingly.
26 (layout_type): Likewise.
27 * config/i386/i386.c (emit_memmov): Likewise.
28 (ix86_expand_set_or_movmem): Likewise.
29 (ix86_expand_vector_init): Likewise.
30 (ix86_get_mask_mode): Likewise.
31 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
33 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
34 * expmed.c (extract_bit_field_1): Likewise.
35 * expr.c (expand_expr_real_2): Likewise.
36 * optabs-query.c (can_vec_perm_p): Likewise.
37 (can_vec_mask_load_store_p): Likewise.
38 * optabs.c (expand_vec_perm): Likewise.
39 * targhooks.c (default_get_mask_mode): Likewise.
40 * tree-vect-stmts.c (vectorizable_store): Likewise.
41 (vectorizable_load): Likewise.
42 (get_vectype_for_scalar_type_and_size): Likewise.
44 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
46 * machmode.h (mode_for_int_vector): New function.
47 * stor-layout.c (mode_for_int_vector): Likewise.
48 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
49 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
50 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
51 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
52 (s390_expand_vcond): Likewise.
54 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
56 * machmode.h (opt_machine_mode): New type.
57 (opt_mode<T>): Allow construction from anything that can be
59 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
60 (mode_for_size): Return an opt_machine_mode.
61 * stor-layout.c (mode_for_size): Likewise.
62 (mode_for_size_tree): Update call accordingly.
63 (bitwise_mode_for_mode): Likewise.
64 (make_fract_type): Likewise.
65 (make_accum_type): Likewise.
66 * caller-save.c (replace_reg_with_saved_mem): Update call
68 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
69 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
70 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
71 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
72 * expmed.c (extract_bit_field_1): Likewise.
73 * reload.c (get_secondary_mem): Likewise.
74 * varasm.c (assemble_integer): Likewise.
75 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
78 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
80 * machmode.h (decimal_float_mode_for_size): New function.
81 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
82 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
83 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
84 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
85 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
86 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
88 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
90 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
91 (get_builtin_sync_mode): Likewise.
92 (expand_ifn_atomic_compare_exchange): Likewise.
93 (expand_builtin_atomic_clear): Likewise.
94 (expand_builtin_atomic_test_and_set): Likewise.
95 (fold_builtin_atomic_always_lock_free): Likewise.
96 * calls.c (compute_argument_addresses): Likewise.
97 (emit_library_call_value_1): Likewise.
98 (store_one_arg): Likewise.
99 * combine.c (combine_instructions): Likewise.
100 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
101 * config/arm/arm.c (arm_function_value): Likewise.
102 (aapcs_allocate_return_reg): Likewise.
103 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
104 * config/i386/i386.c (construct_container): Likewise.
105 (ix86_gimplify_va_arg): Likewise.
106 (ix86_expand_sse_cmp): Likewise.
107 (emit_memmov): Likewise.
108 (emit_memset): Likewise.
109 (expand_small_movmem_or_setmem): Likewise.
110 (ix86_expand_pextr): Likewise.
111 (ix86_expand_pinsr): Likewise.
112 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
113 * config/microblaze/microblaze.c (microblaze_block_move_straight):
115 * config/mips/mips.c (mips_function_value_1) Likewise.
116 (mips_block_move_straight): Likewise.
117 (mips_expand_ins_as_unaligned_store): Likewise.
118 * config/powerpcspe/powerpcspe.c
119 (rs6000_darwin64_record_arg_advance_flush): Likewise.
120 (rs6000_darwin64_record_arg_flush): Likewise.
121 * config/rs6000/rs6000.c
122 (rs6000_darwin64_record_arg_advance_flush): Likewise.
123 (rs6000_darwin64_record_arg_flush): Likewise.
124 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
125 (sparc_function_value_1): Likewise.
126 * config/spu/spu.c (adjust_operand): Likewise.
127 (spu_emit_branch_or_set): Likewise.
128 (arith_immediate_p): Likewise.
129 * emit-rtl.c (gen_lowpart_common): Likewise.
130 * expr.c (expand_expr_real_1): Likewise.
131 * function.c (assign_parm_setup_block): Likewise.
132 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
133 * reload1.c (alter_reg): Likewise.
134 * stor-layout.c (mode_for_vector): Likewise.
135 (layout_type): Likewise.
137 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
139 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
140 (spu_convert_move): Likewise.
141 * lower-subreg.c (resolve_simple_move): Likewise.
143 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
146 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
147 define_insn to a define_expand.
148 (altivec_vsum2sws_direct): New define_insn.
149 (altivec_vsumsws): Convert from a define_insn to a define_expand.
151 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
153 * config/arm/arm.c (arm_option_params_internal): Improve setting of
156 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
161 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
162 realign stack if stack alignment needed is less than incoming
165 2017-09-05 Marek Polacek <polacek@redhat.com>
168 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
171 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
173 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
175 2017-09-05 Richard Biener <rguenther@suse.de>
177 PR tree-optimization/82084
178 * fold-const.c (can_native_encode_string_p): Handle wide characters.
180 2017-09-05 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/82102
183 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
185 2017-09-05 Martin Liska <mliska@suse.cz>
187 PR tree-optimization/82032
188 * tree-cfg.c (generate_range_test): New function.
189 * tree-cfg.h (generate_range_test): Declared here.
190 * tree-cfgcleanup.c (convert_single_case_switch): New function.
191 (cleanup_control_expr_graph): Use it.
192 * tree-switch-conversion.c (try_switch_expansion): Remove
194 (emit_case_nodes): Use generate_range_test.
196 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
199 * config/i386/i386.md (*<btsc><mode>_mask): Add
200 TARGET_USE_BT to insn constraint.
201 (*btr<mode>_mask): Ditto.
203 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
205 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
206 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
208 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
211 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
212 TARGET_NEON and TARGET_IWMMXT.
213 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
214 TARGET_NEON and TARGET_IWMMXT.
215 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
217 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
219 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
220 (ix86_rewrite_tls_address): Ditto.
221 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
222 (ix86_rewrite_tls_address_1): Ditto.
223 (ix86_rewrite_tls_address): Ditto.
224 * config/i386/predicates.md (tls_address_pattern): New predicate.
225 * config/i386/i386.md (TLS address splitter): New splitter.
227 2017-09-04 Richard Biener <rguenther@suse.de>
229 PR tree-optimization/82084
230 * fold-const.h (can_native_encode_string_p): Declare.
231 * fold-const.c (can_native_encode_string_p): Factor out from ...
232 (native_encode_string): ... here.
233 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
234 vectorizing stores from constants we later cannot handle.
236 2017-09-04 Marek Polacek <polacek@redhat.com>
239 * doc/invoke.texi: Update -Wtautological-compare documentation.
241 2017-09-04 Jeff Law <law@redhat.com>
243 PR tree-optimization/64910
244 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
245 swap the first and last operand if the last is a constant.
247 2017-09-04 Marek Polacek <polacek@redhat.com>
250 * convert.c (do_narrow): When sanitizing signed integer overflows,
251 bail out for signed types.
252 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
254 2017-09-04 Richard Biener <rguenther@suse.de>
256 PR tree-optimization/82060
257 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
258 Move devirtualization after stmt folding and before EH/AB/noreturn
259 cleanup to get the stmt refs canonicalized. Use a bool instead
260 of gimple_modified_p since that doesn't work for NOPs. Schedule
261 NOPs generated by folding for removal.
263 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hayward@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * coretypes.h (pad_direction): New enum.
268 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
269 (FUNCTION_ARG_PADDING): Likewise.
270 * target.def (function_arg_padding): New hook.
271 * targhooks.h (default_function_arg_padding): Declare.
272 * targhooks.c (default_function_arg_padding): New function.
273 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
274 (TARGET_FUNCTION_ARG_PADDING): ...this.
275 * doc/tm.texi: Regenerate.
276 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
277 instead of direction.
278 (compute_argument_addresses): Likewise.
279 (load_register_parameters): Likewise.
280 (emit_library_call_value_1): Likewise.
281 (store_one_arg): Use targetm.calls.function_arg_padding instead
282 of FUNCTION_ARG_PADDING.
283 (must_pass_in_stack_var_size_or_pad): Likewise.
284 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
285 (emit_group_store): Likewise.
286 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
287 instead of FUNCTION_ARG_PADDING.
288 (emit_push_insn): Likewise, and propagate enum change throughout
290 * function.h (direction): Delete.
291 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
293 * function.c (assign_parm_find_stack_rtl): Likewise.
294 (assign_parm_setup_block_p): Likewise.
295 (assign_parm_setup_block): Likewise.
296 (gimplify_parameters): Likewise.
297 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
298 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
300 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
301 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
302 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
303 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
304 (aarch64_function_arg_padding): ...this new function.
305 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
306 (TARGET_FUNCTION_ARG_PADDING): Redefine.
307 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
308 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
309 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
310 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
311 (arm_pad_arg_upward): Replace with...
312 (arm_function_arg_padding): ...this new function.
313 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
315 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
316 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
317 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
318 (ia64_hpux_function_arg_padding): Replace with...
319 (ia64_function_arg_padding): ...this new function. Use pad_direction
320 instead of direction. Check for TARGET_HPUX.
321 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
322 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
323 (iq2000_function_arg_padding): New function.
324 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
325 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
326 (mips_function_arg_padding): ...this new function.
327 (mips_pad_reg_upward): Update accordingly.
328 (TARGET_FUNCTION_ARG_PADDING): Redefine.
329 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
330 targetm.calls.function_arg_padding.
331 (FUNCTION_ARG_PADDING): Delete.
332 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
333 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
334 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
335 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
336 (nios2_block_reg_padding): Return pad_direction instead of direction.
337 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
338 instead of direction.
339 (nios2_function_arg_padding): Likewise. Make static.
340 (TARGET_FUNCTION_ARG_PADDING): Redefine.
341 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
342 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
343 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
344 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
345 (pa_function_arg_padding): Make static. Return pad_direction instead
347 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
348 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
349 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
350 instead of direction. Use targetm.calls.function_arg_padding.
351 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
352 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
353 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
354 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
355 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
357 (function_arg_padding): Rename to...
358 (rs6000_function_arg_padding): ...this. Make static. Return
359 pad_direction instead of direction.
360 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
361 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
362 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
363 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
364 instead of direction. Use targetm.calls.function_arg_padding.
365 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
366 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
367 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
368 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
369 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
370 (function_arg_padding): Rename to...
371 (rs6000_function_arg_padding): ...this. Make static. Return
372 pad_direction instead of direction.
373 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
374 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
375 * config/s390/s390.c (s390_function_arg_padding): New function.
376 (TARGET_FUNCTION_ARG_PADDING): Redefine.
377 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
378 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
379 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
380 (function_arg_padding): Rename to...
381 (sparc_function_arg_padding): ...this. Make static. Return
382 pad_direction instead of direction.
383 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
384 * config/spu/spu.c (spu_function_arg_padding): New function.
385 (TARGET_FUNCTION_ARG_PADDING): Redefine.
386 * system.h (FUNCTION_ARG_PADDING): Poison.
388 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
389 Alan Hayward <alan.hayward@arm.com>
390 David Sherwood <david.sherwood@arm.com>
392 * target.def (modes_tieable_p): New hook.
393 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
394 (TARGET_MODES_TIEABLE_P): ...this.
395 * doc/tm.texi.in: Regenerate.
396 * hooks.h (hook_bool_mode_mode_true): Declare.
397 * hooks.c (hook_bool_mode_mode_true): New function.
398 * combine.c (subst): Use targetm.modes_tieable_p instead of
400 * dse.c (find_shift_sequence): Likewise.
401 * expmed.c (extract_low_bits): Likewise.
402 * lower-subreg.c: Include target.h.
403 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
405 * rtlanal.c (rtx_cost): Likewise.
406 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
407 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
408 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
409 (TARGET_MODES_TIEABLE_P): Redefine.
410 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
411 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
412 (TARGET_MODES_TIEABLE_P): Redefine.
413 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
414 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
415 (arc_modes_tieable_p): New function.
416 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
417 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
418 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
419 (arm_modes_tieable_p): Make static.
420 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
421 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
422 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
423 (TARGET_MODES_TIEABLE_P): Redefine.
424 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
425 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
426 (TARGET_MODES_TIEABLE_P): Redefine.
427 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
428 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
429 (cr16_modes_tieable_p): New function.
430 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
431 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
432 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
433 (TRULY_NOOP_TRUNCATION): Update comment.
434 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
435 (TRULY_NOOP_TRUNCATION): Update comment.
436 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
437 (frv_modes_tieable_p): New function.
438 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
439 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
440 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
441 (TARGET_MODES_TIEABLE_P): Redefine.
442 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
443 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
444 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
445 (TARGET_MODES_TIEABLE_P): Redefine.
446 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
447 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
448 (ia64_modes_tieable_p): New function.
449 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
450 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
451 (iq2000_modes_tieable_p): New function.
452 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
453 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
454 (lm32_modes_tieable_p): New function.
455 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
456 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
457 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
458 (TARGET_MODES_TIEABLE_P): Redefine.
459 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
460 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
461 (m32r_modes_tieable_p): New function.
462 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
463 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
464 (m68k_modes_tieable_p): New function.
465 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
466 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
467 (mcore_modes_tieable_p): New function.
468 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
469 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
471 (TARGET_MODES_TIEABLE_P): Redefine.
472 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
473 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
474 * config/mips/mips.c (mips_modes_tieable_p): Make static.
475 (TARGET_MODES_TIEABLE_P): Redefine.
476 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
477 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
478 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
479 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
480 (mn10300_modes_tieable_p): ...this and make static.
481 (TARGET_MODES_TIEABLE_P): Redefine.
482 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
483 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
484 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
485 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
486 (msp430_modes_tieable_p): Make static.
487 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
488 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
489 (TARGET_MODES_TIEABLE_P): Redefine.
490 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
491 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
492 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
493 (TARGET_MODES_TIEABLE_P): Redefine.
494 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
495 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
496 * config/pa/pa.c (pa_modes_tieable_p): Make static.
497 (TARGET_MODES_TIEABLE_P): Redefine.
498 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
499 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
500 (pdp11_modes_tieable_p): New function.
501 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
502 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
503 (rs6000_modes_tieable_p): New function.
504 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
505 * config/powerpcspe/powerpcspe.md: Update comment.
506 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
507 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
508 (TARGET_MODES_TIEABLE_P): Redefine.
509 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
510 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
511 (rl78_modes_tieable_p): New function.
512 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
513 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
514 (rs6000_modes_tieable_p): New function.
515 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
516 * config/rs6000/rs6000.md: Update comment.
517 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
518 * config/rx/rx.c (rx_modes_tieable_p): New function.
519 (TARGET_MODES_TIEABLE_P): Redefine.
520 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
521 * config/s390/s390.c (s390_modes_tieable_p): New function.
522 (TARGET_MODES_TIEABLE_P): Redefine.
523 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
524 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
525 (sh_modes_tieable_p): New function.
526 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
527 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
528 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
529 (sparc_modes_tieable_p): Make static.
530 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
531 * config/spu/spu.c (spu_modes_tieable_p): New function.
532 (TARGET_MODES_TIEABLE_P): Redefine.
533 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
534 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
535 (TARGET_MODES_TIEABLE_P): Redefine.
536 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
537 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
538 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
539 * config/v850/v850.c (v850_modes_tieable_p): New function.
540 (TARGET_MODES_TIEABLE_P): Redefine.
541 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
542 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
543 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
544 (visium_modes_tieable_p): New function.
545 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
546 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
547 (xtensa_modes_tieable_p): New function.
548 * system.h (MODES_TIEABLE_P): Poison.
550 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
551 Alan Hayward <alan.hayward@arm.com>
552 David Sherwood <david.sherwood@arm.com>
554 * target.def (hard_regno_mode_ok): New hook.
555 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
556 (TARGET_HARD_REGNO_MODE_OK): ...this.
557 * doc/tm.texi.in: Regenerate.
558 * hooks.h (hook_bool_uint_mode_true): Declare.
559 * hooks.c (hook_bool_uint_mode_true): New function.
560 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
562 * genpreds.c (write_insn_preds_c): Add an include of target.h.
563 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
564 instead of HARD_REGNO_MODE_OK.
565 * caller-save.c: Include target.h.
566 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
568 * combine.c (can_combine_p): Likewise.
569 (combinable_i3pat): Likewise.
570 (can_change_dest_mode): Likewise.
571 * expr.c (init_expr_target): Likewise.
572 (convert_move): Likewise.
573 (convert_modes): Likewise.
574 * ira.c (setup_prohibited_class_mode_regs): Likewise.
575 (setup_prohibited_mode_move_regs): Likewise.
576 * ira.h (target_ira): Likewise.
577 * lra-assigns.c (find_hard_regno_for_1): Likewise.
578 * lra-constraints.c (process_alt_operands): Likewise.
579 (split_reg): Likewise.
580 * recog.c (peep2_find_free_register): Likewise.
581 * ree.c (combine_reaching_defs): Likewise.
582 * regcprop.c (maybe_mode_change): Likewise.
583 * reginfo.c (init_reg_sets_1): Likewise.
584 (choose_hard_reg_mode): Likewise.
585 (simplifiable_subregs): Likewise.
586 * regrename.c (check_new_reg_p): Likewise.
587 * reload.c (find_valid_class): Likewise.
588 (find_valid_class_1): Likewise.
589 (reload_inner_reg_of_subreg): Likewise.
590 (push_reload): Likewise.
591 (combine_reloads): Likewise.
592 (find_dummy_reload): Likewise.
593 (find_reloads): Likewise.
594 * reload1.c (find_reg): Likewise.
595 (set_reload_reg): Likewise.
596 (allocate_reload_reg): Likewise.
597 (choose_reload_regs): Likewise.
598 (reload_adjust_reg_for_temp): Likewise.
599 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
600 (simplify_subreg_regno): Likewise.
601 * sel-sched.c (init_regs_for_mode): Likewise.
602 * varasm.c (make_decl_rtl): Likewise.
603 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
604 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
606 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
607 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
608 (TARGET_HARD_REGNO_MODE_OK): Redefine.
609 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
610 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
612 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
613 (arc_mode_class): Delete.
614 (HARD_REGNO_MODE_OK): Delete.
615 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
616 (arc_hard_regno_mode_ok): Rename old array to...
617 (arc_hard_regno_mode_ok_modes): ...this.
618 (arc_conditional_register_usage): Update accordingly.
619 (arc_mode_class): Make static.
620 (arc_hard_regno_mode_ok): New function.
621 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
622 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
623 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
624 (arm_hard_regno_mode_ok): Make static.
625 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
627 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
628 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
629 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
632 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
633 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
634 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
635 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
636 (TARGET_HARD_REGNO_MODE_OK): Redefine.
637 * config/bfin/predicates.md (valid_reg_operand): Use
638 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
639 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
640 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
641 (TARGET_HARD_REGNO_MODE_OK): Redefine.
642 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
643 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
644 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
645 (cr16_hard_regno_mode_ok): Make static and return a bool.
646 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
647 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
648 (cris_hard_regno_mode_ok): New function.
649 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
650 (epiphany_mode_class): Delete.
651 (HARD_REGNO_MODE_OK): Delete.
652 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
653 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
654 (hard_regno_mode_ok): Rename to...
655 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
656 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
657 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
659 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
660 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
661 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
662 (frv_hard_regno_mode_ok): Make static and return a bool.
663 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
665 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
666 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
667 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
668 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
670 (TARGET_HARD_REGNO_MODE_OK): Redefine.
671 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
672 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
673 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
675 (TARGET_HARD_REGNO_MODE_OK): Redefine.
676 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
677 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
678 (ia64_hard_regno_mode_ok): New function.
679 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
680 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
681 (iq2000_hard_regno_mode_ok): New function.
682 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
683 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
684 (lm32_hard_regno_mode_ok): New function.
685 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
686 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
687 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
688 instead of HARD_REGNO_MODE_OK.
689 (m32c_hard_regno_ok): Rename to...
690 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
691 (m32c_cannot_change_mode_class): Update accordingly.
692 (TARGET_HARD_REGNO_MODE_OK): Redefine.
693 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
694 (m32r_mode_class): Delete.
695 (HARD_REGNO_MODE_OK): Delete.
696 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
697 (m32r_hard_regno_mode_ok): Rename to...
698 (m32r_hard_regno_modes): ...this.
699 (m32r_mode_class): Make static.
700 (m32r_hard_regno_mode_ok): New function.
701 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
702 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
703 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
704 (m68k_hard_regno_mode_ok): Make static.
705 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
706 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
707 (mcore_hard_regno_mode_ok): New function.
708 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
709 (HARD_REGNO_MODE_OK): Delete.
710 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
712 (microblaze_hard_regno_mode_ok_p): ...this and make static.
713 (microblaze_hard_regno_mode_ok): New function.
714 (TARGET_HARD_REGNO_MODE_OK): Redefine.
715 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
716 (mips_hard_regno_mode_ok): Delete.
717 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
718 (mips_hard_regno_mode_ok_p): ...this and make static.
719 (mips_hard_regno_mode_ok_p): Rename to...
720 (mips_hard_regno_mode_ok_uncached): ...this.
721 (mips_hard_regno_mode_ok): New function.
722 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
723 of HARD_REGNO_MODE_OK.
724 (mips_option_override): Update after above name changes.
725 (TARGET_HARD_REGNO_MODE_OK): Redefine.
726 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
727 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
728 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
729 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
730 (TARGET_HARD_REGNO_MODE_OK): Redefine.
731 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
732 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
733 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
734 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
735 (msp430_hard_regno_mode_ok): Make static and return a bool.
736 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
737 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
738 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
741 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
742 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
743 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
744 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
745 (PA_HARD_REGNO_MODE_OK): ...this
746 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
747 (PA_HARD_REGNO_MODE_OK): ...this.
748 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
749 (pa_hard_regno_mode_ok): New function.
750 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
751 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
752 (pdp11_hard_regno_mode_ok): New function.
753 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
754 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
756 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
758 (TARGET_HARD_REGNO_MODE_OK): Redefine.
759 (rs6000_hard_regno_mode_ok): Rename to...
760 (rs6000_hard_regno_mode_ok_uncached): ...this.
761 (rs6000_init_hard_regno_mode_ok): Update accordingly.
762 (rs6000_hard_regno_mode_ok): New function.
763 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
764 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
765 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
766 (riscv_hard_regno_mode_ok): ...this and make static.
767 (TARGET_HARD_REGNO_MODE_OK): Redefine.
768 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
769 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
770 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
771 (rl78_hard_regno_mode_ok): Make static and return bool.
772 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
773 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
775 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
776 (TARGET_HARD_REGNO_MODE_OK): Redefine.
777 (rs6000_hard_regno_mode_ok): Rename to...
778 (rs6000_hard_regno_mode_ok_uncached): ...this.
779 (rs6000_init_hard_regno_mode_ok): Update accordingly.
780 (rs6000_hard_regno_mode_ok): New function.
781 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
782 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
783 (TARGET_HARD_REGNO_MODE_OK): Redefine.
784 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
785 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
786 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
787 (TARGET_HARD_REGNO_MODE_OK): Redefine.
788 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
789 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
790 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
791 (sh_hard_regno_mode_ok): Make static.
792 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
793 instead of HARD_REGNO_MODE_OK.
794 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
795 (sparc_mode_class): Delete.
796 (HARD_REGNO_MODE_OK): Delete.
797 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
798 (hard_regno_mode_classes): Make static.
799 (sparc_mode_class): Likewise.
800 (sparc_hard_regno_mode_ok): New function.
801 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
802 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
803 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
805 (TARGET_HARD_REGNO_MODE_OK): Redefine.
806 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
807 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
808 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
809 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
811 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
812 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
813 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
814 (visium_hard_regno_mode_ok): New function.
815 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
816 instead of HARD_REGNO_MODE_OK.
817 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
818 (HARD_REGNO_MODE_OK): Delete.
819 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
820 (xtensa_hard_regno_mode_ok_p): ...this and make static.
821 (xtensa_option_override): Update accordingly.
822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
823 (xtensa_hard_regno_mode_ok): New function.
824 * system.h (HARD_REGNO_MODE_OK): Poison.
826 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
827 Alan Hayward <alan.hayward@arm.com>
828 David Sherwood <david.sherwood@arm.com>
830 * target.def (hard_regno_call_part_clobbered): New hook.
831 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
832 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
833 * doc/tm.texi: Regenerate.
834 * hooks.h (hook_bool_uint_mode_false): Declare.
835 * hooks.c (hook_bool_uint_mode_false): New function.
836 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
837 * cselib.c (cselib_process_insn): Use
838 targetm.hard_regno_call_part_clobbered instead of
839 HARD_REGNO_CALL_PART_CLOBBERED.
840 * ira-conflicts.c (ira_build_conflicts): Likewise.
841 * ira-costs.c (ira_tune_allocno_costs): Likewise.
842 * lra-constraints.c (need_for_call_save_p): Likewise.
843 * lra-lives.c: Include target.h.
844 (check_pseudos_live_through_calls): Use
845 targetm.hard_regno_call_part_clobbered instead of
846 HARD_REGNO_CALL_PART_CLOBBERED.
847 * regcprop.c: Include target.h.
848 (copyprop_hardreg_forward_1): Use
849 targetm.hard_regno_call_part_clobbered instead of
850 HARD_REGNO_CALL_PART_CLOBBERED.
851 * reginfo.c (choose_hard_reg_mode): Likewise.
852 * regrename.c (check_new_reg_p): Likewise.
853 * reload.c (find_equiv_reg): Likewise.
854 * reload1.c (emit_reload_insns): Likewise.
855 * sched-deps.c (deps_analyze_insn): Likewise.
856 * sel-sched.c (init_regs_for_mode): Likewise.
857 (mark_unavailable_hard_regs): Likewise.
858 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
859 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
860 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
862 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
863 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
864 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
866 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
868 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
869 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
870 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
872 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
873 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
874 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
876 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
877 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
879 * config/powerpcspe/powerpcspe.c
880 (rs6000_hard_regno_call_part_clobbered): New function.
881 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
882 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
883 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
885 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
886 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
887 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
889 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
890 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
891 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
893 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
894 Alan Hayward <alan.hayward@arm.com>
895 David Sherwood <david.sherwood@arm.com>
897 * rtl.h (subreg_memory_offset): Declare.
898 * emit-rtl.c (subreg_memory_offset): New function.
899 * expmed.c (store_bit_field_1): Use it.
900 * expr.c (undefined_operand_subword_p): Likewise.
901 * simplify-rtx.c (simplify_subreg): Likewise.
903 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
905 PR rtl-optimization/57448
908 * optabs.c (expand_atomic_load): Place compiler memory barriers if
909 using atomic_load pattern.
910 (expand_atomic_store): Likewise.
912 2017-09-04 Jakub Jelinek <jakub@redhat.com>
915 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
916 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
917 handling. Use replace_call_with_value with NULL instead of
918 gsi_replace, unlink_stmt_vdef and release_defs.
920 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
923 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
925 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
928 * rtl.h (emit_library_call_value_1): Declare.
929 (emit_library_call): Replace declaration with a series of overloads.
930 Remove the parameter count argument.
931 (emit_library_call_value): Likewise.
932 * calls.c (emit_library_call_value_1): Make global. Replace varargs
933 with an "rtx_mode_t *".
934 (emit_library_call_value): Delete.
935 (emit_library_call): Likewise.
936 * asan.c (asan_emit_stack_protection): Update calls accordingly.
937 (asan_emit_allocas_unpoison): Likewise.
938 * builtins.c (expand_builtin_powi): Likewise.
939 (expand_asan_emit_allocas_unpoison): Likewise.
940 * cfgexpand.c (expand_main_function): Likewise.
941 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
942 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
943 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
944 * config/arm/arm.c (arm_trampoline_init): Likewise.
945 (arm_call_tls_get_addr): Likewise.
946 (arm_expand_divmod_libfunc): Likewise.
947 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
948 (smulsi3_highpart): Likewise.
949 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
950 (c6x_expand_compare): Likewise.
951 (c6x_expand_movmem): Likewise.
952 * config/frv/frv.c (frv_trampoline_init): Likewise.
953 * config/i386/i386.c (ix86_trampoline_init): Likewise.
954 (ix86_expand_divmod_libfunc): Likewise.
955 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
956 (ia64_expand_compare): Likewise.
957 (ia64_profile_hook): Likewise.
958 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
959 (nonlocal_goto): Likewise.
960 (restore_stack_nonlocal): Likewise.
961 * config/m32r/m32r.c (block_move_call): Likewise.
962 (m32r_trampoline_init): Likewise.
963 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
964 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
965 (m68k_call_m68k_read_tp): Likewise.
966 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
967 (microblaze_expand_divide): Likewise.
968 * config/mips/mips.h (mips_args): Likewise.
969 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
970 (MIPS_ICACHE_SYNC): Likewise.
971 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
972 (nios2_trampoline_init): Likewise.
973 * config/pa/pa.c (hppa_tls_call): Likewise.
974 (pa_trampoline_init): Likewise.
975 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
976 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
977 (expand_strn_compare): Likewise.
978 (rs6000_generate_compare): Likewise.
979 (rs6000_expand_float128_convert): Likewise.
980 (output_profile_hook): Likewise.
981 (rs6000_trampoline_init): Likewise.
982 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
983 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
984 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
985 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
986 (rs6000_generate_compare): Likewise.
987 (rs6000_expand_float128_convert): Likewise.
988 (output_profile_hook): Likewise.
989 (rs6000_trampoline_init): Likewise.
990 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
991 * config/sh/sh.c (sh_trampoline_init): Likewise.
992 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
993 (sparc_emit_float_lib_cmp): Likewise.
994 (sparc32_initialize_trampoline): Likewise.
995 (sparc64_initialize_trampoline): Likewise.
996 (sparc_profile_hook): Likewise.
997 * config/spu/spu.c (ea_load_store): Likewise.
998 * config/spu/spu.md (floatunssidf2): Likewise.
999 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1000 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1001 * config/visium/visium.c (expand_block_move_4): Likewise.
1002 (expand_block_move_2): Likewise.
1003 (expand_block_move_1): Likewise.
1004 (expand_block_set_4): Likewise.
1005 (expand_block_set_2): Likewise.
1006 (expand_block_set_1): Likewise.
1007 (visium_trampoline_init): Likewise.
1008 (visium_profile_hook): Likewise.
1009 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1010 (xtensa_setup_frame_addresses): Likewise.
1011 (xtensa_trampoline_init): Likewise.
1012 * except.c (sjlj_emit_function_enter): Likewise.
1013 (sjlj_emit_function_exit): Likewise.
1014 * explow.c (allocate_dynamic_stack_space): Likewise.
1015 (probe_stack_range): Likewise.
1016 * expr.c (convert_mode_scalar): Likewise.
1017 * optabs.c (expand_binop): Likewise.
1018 (expand_twoval_binop_libfunc): Likewise.
1019 (expand_unop): Likewise.
1020 (prepare_cmp_insn): Likewise.
1021 (prepare_float_lib_cmp): Likewise.
1022 (expand_float): Likewise.
1023 (expand_fix): Likewise.
1024 (expand_fixed_convert): Likewise.
1025 (maybe_emit_sync_lock_test_and_set): Likewise.
1026 (expand_atomic_compare_and_swap): Likewise.
1027 (expand_mem_thread_fence): Likewise.
1028 (expand_atomic_fetch_op): Likewise.
1030 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1032 * doc/generic.texi (OpenACC): Adjust URL.
1033 * doc/invoke.texi (C Dialect Options): Ditto.
1035 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1037 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1038 predicate for operand 1. Add (m,<S>) constraint.
1039 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1040 Prevent memory operand 1 with register operand 2.
1042 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1044 PR rtl-optimization/82024
1045 * combine.c (try_combine): If the combination result is a PARALLEL,
1046 and we only need to retain the SET in there that would be placed
1047 at I2, check that we can place that at I3 instead, before doing so.
1049 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1052 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1054 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1055 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1058 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1059 Jeff Law <law@redhat.com>
1061 * varasm.c (bss_initializer_p): Do not put constants into .bss
1062 (categorize_decl_for_section): Handle bss_initializer_p returning
1063 false when DECL_INITIAL is NULL.
1065 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1068 * config/s390/s390.c (s390_can_inline_p): New function.
1070 2017-09-01 Jeff Law <law@redhat.com>
1072 PR tree-optimization/82052
1073 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1074 Always initialize the returned slot after a hash table miss
1075 when INSERT is true.
1077 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1079 * config/s390/s390.md (mem_signal_fence): Remove.
1080 * doc/md.texi (mem_signal_fence): Remove.
1081 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1083 * target-insns.def (mem_signal_fence): Remove.
1085 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1088 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1091 * asan.c (create_odr_indicator): Strip name encoding from assembler
1092 name before appending it after __odr_asan_.
1094 2017-09-01 Martin Liska <mliska@suse.cz>
1096 PR tree-optimization/82059
1097 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1098 frequency only when an edge is redirected.
1100 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1102 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1103 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1104 (arc_conditional_register_usage): Remove ARC600 lp_count
1106 (arc_file_start): Emit Tag_ARC_CPU_variation.
1107 (arc_can_use_doloop_p): New conditions to use ZOLs.
1108 (hwloop_fail): New function.
1109 (hwloop_optimize): Likewise.
1110 (hwloop_pattern_reg): Likewise.
1111 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1112 (arc_reorg_loops): New function, calls reorg_loops.
1113 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1114 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1116 (arc_loop_hazard): Remove function, functionality moved into
1118 (arc_hazard): Remove arc_loop_hazard call.
1119 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1120 into hwloop_optimize.
1121 (arc_label_align): Remove ZOL handling.
1122 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1123 * config/arc/arc.md (doloop_begin): Remove pattern.
1124 (doloop_begin_i): Likewise.
1125 (doloop_end_i): Likewise.
1126 (doloop_fallback): Likewise.
1127 (doloop_fallback_m): Likewise.
1128 (doloop_end): Reimplement expand.
1129 (arc_lp): New pattern for LP instruction.
1130 (loop_end): New pattern.
1131 (loop_fail): Likewise.
1132 (decrement_and_branch_until_zero): Likewise.
1133 * config/arc/arc.opt (mlpc-width): New option.
1134 * doc/invoke.texi (mlpc-width): Document option.
1136 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1138 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1139 (arc_ccfsm_advance): Fix checking for delay slots.
1140 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1142 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1144 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1146 (movhi_insn): Update store instruction constraint which are saving
1147 6-bit short immediates.
1148 (movsi_insn): Consider also short scaled load operations.
1149 (zero_extendhisi2_i): Use Usd constraint instead of T.
1150 (extendhisi2_i): Add q constraint.
1151 (arc_clzsi2): Add type and length attributes.
1152 (arc_ctzsi2): Likewise.
1153 * config/arc/constraints.md (Usc): Update constraint, the
1154 assembler can parse two relocations for a single instruction.
1156 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1158 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1159 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1161 2017-08-31 Olivier Hainque <hainque@adacore.com>
1163 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1164 match as powerpc-wrs-vxworks*.
1166 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1168 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1169 register constraint for by-element operand.
1170 (aarch64_mls_elt_merge<mode>): Likewise.
1172 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1174 * config/arc/arc.c (arc_can_follow_jump): Check for short
1177 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1179 * config.gcc: Use g.opt for arc.
1180 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1181 functionality moved to ...
1182 (legitimate_scaled_address_p): New function, ...here.
1183 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1184 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1185 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1187 (arc_override_options): Handle G option.
1188 (arc_output_pic_addr_const): Correct function definition.
1189 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1190 (arc_decl_anon_ns_mem_p): Delete.
1191 (arc_in_small_data_p): Overhaul this function to take into
1192 consideration the value given via G option.
1193 (arc_rewrite_small_data_1): Renamed and corrected old
1194 arc_rewrite_small_data function.
1195 (arc_rewrite_small_data): New function.
1196 (small_data_pattern): Don't use pic_offset_table_rtx.
1197 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1198 * config/arc/simdext.md (movmisalignv2hi): Use
1199 prepare_move_operands function.
1201 (movmisalign*): Likewise.
1202 * doc/invoke.texi (ARC options): Document -G option.
1204 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1206 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1208 * config/arc/arc.c (arc_print_operand): Output scalled address for
1209 sdata whenever is possible.
1210 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1211 load/stores are available.
1212 (compact_sda_memory_operand): Check for the alignment required by
1213 code density instructions.
1214 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1216 * config/arc/constraints.md (Usd): Update constraint.
1217 (Us0): New constraint.
1218 (Usc): Update constraint.
1220 2017-08-31 Richard Biener <rguenther@suse.de>
1223 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1226 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1228 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1229 Resize type_signature.
1231 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1232 Alan Hayward <alan.hayward@arm.com>
1233 David Sherwood <david.sherwood@arm.com>
1235 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1236 subregs whose inner modes can be stored in GPRs.
1237 (aarch64_classify_index): Likewise.
1239 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1240 Alan Hayward <alan.hayward@arm.com>
1241 David Sherwood <david.sherwood@arm.com>
1243 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1244 (V_INT_EQUIV): ...this.
1245 (v_cmp_result): Rename to...
1246 (v_int_equiv): ...this.
1247 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1248 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1249 (copysign<mode>3): Likewise.
1250 (aarch64_simd_bsl<mode>_internal): Likewise.
1251 (aarch64_simd_bsl<mode>): Likewise.
1252 (vec_cmp<mode><mode>): Likewise.
1253 (vcond<mode><mode>): Likewise.
1254 (vcond<v_cmp_mixed><mode>): Likewise.
1255 (vcondu<mode><v_cmp_mixed>): Likewise.
1256 (aarch64_cm<optab><mode>): Likewise.
1257 (aarch64_cmtst<mode>): Likewise.
1258 (aarch64_fac<optab><mode>): Likewise.
1259 (vec_perm_const<mode>): Likewise.
1260 (vcond_mask_<mode><v_cmp_result>): Rename to...
1261 (vcond_mask_<mode><v_int_equiv>): ...this.
1262 (vec_cmp<mode><v_cmp_result>): Rename to...
1263 (vec_cmp<mode><v_int_equiv>): ...this.
1265 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1266 Alan Hayward <alan.hayward@arm.com>
1267 David Sherwood <david.sherwood@arm.com>
1269 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1271 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1272 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1273 (UNSPEC_LD4_DREG): New unspecs.
1274 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1275 (aarch64_ld2<mode>_dreg_be): Replace with...
1276 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1278 (aarch64_ld3<mode>_dreg_le)
1279 (aarch64_ld3<mode>_dreg_be): Replace with...
1280 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1282 (aarch64_ld4<mode>_dreg_le)
1283 (aarch64_ld4<mode>_dreg_be): Replace with...
1284 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1287 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1289 PR tree-optimization/81987
1290 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1291 insert an initializer in a location not dominated by the stride
1294 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1296 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1297 on the entire header of the finally block in the fallthru case.
1299 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1301 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1303 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1305 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1306 rs6000_emit_move_from_cr and call renamed function.
1307 (rs6000_emit_prologue): Call renamed functions.
1308 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1309 movesi_from_cr, remove volatile CRs.
1311 2017-08-30 Jon Beniston <jon@beniston.com>
1312 Richard Biener <rguenther@suse.de>
1314 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1315 of VECTOR_MODE_P check.
1316 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1317 element vector types.
1319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1321 * df.h (df_read_modify_subreg_p): Remove in favor of...
1322 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1323 const_rtx instead of an rtx.
1324 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1325 * df-problems.c (df_word_lr_mark_ref): Likewise.
1326 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1327 (mark_pseudo_reg_dead): Likewise.
1328 (mark_ref_dead): Likewise.
1329 * reginfo.c (init_subregs_of_mode): Likewise.
1330 * sched-deps.c (sched_analyze_1): Likewise.
1331 * df-scan.c (df_def_record_1): Likewise.
1332 (df_uses_record): Likewise.
1333 (df_read_modify_subreg_p): Remove in favor of...
1334 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1335 const_rtx instead of an rtx.
1337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1338 Alan Hayward <alan.hayward@arm.com>
1339 David Sherwood <david.sherwood@arm.com>
1341 * rtl.h (partial_subreg_p): New function.
1342 * caller-save.c (save_call_clobbered_regs): Use it.
1343 * calls.c (expand_call): Likewise.
1344 * combine.c (combinable_i3pat): Likewise.
1345 (simplify_set): Likewise.
1346 (make_extraction): Likewise.
1347 (make_compound_operation_int): Likewise.
1348 (gen_lowpart_or_truncate): Likewise.
1349 (force_to_mode): Likewise.
1350 (make_field_assignment): Likewise.
1351 (reg_truncated_to_mode): Likewise.
1352 (record_truncated_value): Likewise.
1353 (move_deaths): Likewise.
1354 * cse.c (record_jump_cond): Likewise.
1355 (cse_insn): Likewise.
1356 * cselib.c (cselib_lookup_1): Likewise.
1357 * expmed.c (extract_bit_field_using_extv): Likewise.
1358 * function.c (assign_parm_setup_reg): Likewise.
1359 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1360 * ira-build.c (create_insn_allocnos): Likewise.
1361 * lra-coalesce.c (merge_pseudos): Likewise.
1362 * lra-constraints.c (match_reload): Likewise.
1363 (simplify_operand_subreg): Likewise.
1364 (curr_insn_transform): Likewise.
1365 * lra-lives.c (process_bb_lives): Likewise.
1366 * lra.c (new_insn_reg): Likewise.
1367 (lra_substitute_pseudo): Likewise.
1368 * regcprop.c (mode_change_ok): Likewise.
1369 (maybe_mode_change): Likewise.
1370 (copyprop_hardreg_forward_1): Likewise.
1371 * reload.c (push_reload): Likewise.
1372 (find_reloads): Likewise.
1373 (find_reloads_subreg_address): Likewise.
1374 * reload1.c (alter_reg): Likewise.
1375 (eliminate_regs_1): Likewise.
1376 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1378 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1380 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1381 back to if statements, including unpack.
1383 2017-08-30 Martin Liska <mliska@suse.cz>
1386 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1388 (func_checker::compare_asm_inputs_outputs): ... this function.
1389 (func_checker::compare_gimple_asm): Use the function to compare
1390 also ASM constrains.
1391 * ipa-icf-gimple.h: Rename the function.
1393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1394 Alan Hayward <alan.hayward@arm.com>
1395 David Sherwood <david.sherwood@arm.com>
1397 * coretypes.h (complex_mode): New type.
1398 * gdbhooks.py (build_pretty_printer): Handle it.
1399 * machmode.h (complex_mode): New class.
1400 (complex_mode::includes_p): New function.
1401 (is_complex_int_mode): Likewise.
1402 (is_complex_float_mode): Likewise.
1403 * genmodes.c (get_mode_class): Handle complex mode classes.
1404 * function.c (expand_function_end): Use is_complex_int_mode.
1406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1407 Alan Hayward <alan.hayward@arm.com>
1408 David Sherwood <david.sherwood@arm.com>
1410 * coretypes.h (scalar_mode_pod): New typedef.
1411 * gdbhooks.py (build_pretty_printer): Handle it.
1412 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1413 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1414 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1415 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1419 Alan Hayward <alan.hayward@arm.com>
1420 David Sherwood <david.sherwood@arm.com>
1422 * machmode.h (mode_for_vector): Take a scalar_mode instead
1424 * stor-layout.c (mode_for_vector): Likewise.
1425 * explow.c (promote_mode): Use as_a <scalar_mode>.
1426 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1429 Alan Hayward <alan.hayward@arm.com>
1430 David Sherwood <david.sherwood@arm.com>
1432 * target.def (preferred_simd_mode): Take a scalar_mode
1433 instead of a machine_mode.
1434 * targhooks.h (default_preferred_simd_mode): Likewise.
1435 * targhooks.c (default_preferred_simd_mode): Likewise.
1436 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1437 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1438 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1439 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1440 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1441 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1442 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1443 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1445 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1446 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1447 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1448 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1449 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1450 * doc/tm.texi: Regenerate.
1451 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1455 Alan Hayward <alan.hayward@arm.com>
1456 David Sherwood <david.sherwood@arm.com>
1458 * target.def (scalar_mode_supported_p): Take a scalar_mode
1459 instead of a machine_mode.
1460 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1461 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1462 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1463 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1464 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1465 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1466 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1467 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1468 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1469 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1470 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1471 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1472 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1473 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1475 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1476 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1477 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1478 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1479 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1481 * doc/tm.texi: Regenerate.
1483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1484 Alan Hayward <alan.hayward@arm.com>
1485 David Sherwood <david.sherwood@arm.com>
1487 * coretypes.h (opt_scalar_mode): New typedef.
1488 * gdbhooks.py (build_pretty_printers): Handle it.
1489 * machmode.h (mode_iterator::get_2xwider): Add overload for
1491 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1493 * expr.c (convert_mode_scalar): Likewise.
1494 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1495 * optabs.c (expand_float): Likewise.
1496 (expand_fix): Likewise.
1497 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1500 Alan Hayward <alan.hayward@arm.com>
1501 David Sherwood <david.sherwood@arm.com>
1503 * optabs.c (expand_float): Explicitly check for scalars before
1504 using a branching expansion.
1505 (expand_fix): Likewise.
1507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1508 Alan Hayward <alan.hayward@arm.com>
1509 David Sherwood <david.sherwood@arm.com>
1511 * expr.c (convert_mode): Split scalar handling out into...
1512 (convert_mode_scalar): ...this new function. Treat the modes
1515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1521 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1524 Alan Hayward <alan.hayward@arm.com>
1525 David Sherwood <david.sherwood@arm.com>
1527 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1528 rather than a machine_mode.
1529 (fixed_from_string): Likewise.
1530 (fixed_convert): Likewise.
1531 (fixed_convert_from_int): Likewise.
1532 (fixed_convert_from_real): Likewise.
1533 (real_convert_from_fixed): Likewise.
1534 * fixed-value.c (fixed_from_double_int): Likewise.
1535 (fixed_from_string): Likewise.
1536 (fixed_convert): Likewise.
1537 (fixed_convert_from_int): Likewise.
1538 (fixed_convert_from_real): Likewise.
1539 (real_convert_from_fixed): Likewise.
1540 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1543 Alan Hayward <alan.hayward@arm.com>
1544 David Sherwood <david.sherwood@arm.com>
1546 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1547 of separate mode class checks. Do not allow vector modes here.
1548 (immed_wide_int_const): Use as_a <scalar_mode>.
1549 * explow.c (trunc_int_for_mode): Likewise.
1550 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1551 (wi::shwi): Likewise.
1552 (wi::min_value): Likewise.
1553 (wi::max_value): Likewise.
1554 * dwarf2out.c (loc_descriptor): Likewise.
1555 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * tree.h (SCALAR_TYPE_MODE): New macro.
1563 * expr.c (expand_expr_addr_expr_1): Use it.
1564 (expand_expr_real_2): Likewise.
1565 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1566 (fold_convert_const_fixed_from_int): Likewise.
1567 (fold_convert_const_fixed_from_real): Likewise.
1568 (native_encode_fixed): Likewise
1569 (native_encode_complex): Likewise
1570 (native_encode_vector): Likewise.
1571 (native_interpret_fixed): Likewise.
1572 (native_interpret_real): Likewise.
1573 (native_interpret_complex): Likewise.
1574 (native_interpret_vector): Likewise.
1575 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1576 (simd_clone_adjust_argument_types): Likewise.
1577 (simd_clone_init_simd_arrays): Likewise.
1578 (simd_clone_adjust): Likewise.
1579 * stor-layout.c (layout_type): Likewise.
1580 * tree.c (build_minus_one_cst): Likewise.
1581 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1582 * tree-inline.c (estimate_move_cost): Likewise.
1583 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1584 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1585 (vectorizable_reduction): Likewise.
1586 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1587 (vect_recog_mixed_size_cond_pattern): Likewise.
1588 (check_bool_pattern): Likewise.
1589 (adjust_bool_pattern): Likewise.
1590 (search_type_for_mask_1): Likewise.
1591 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1592 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1593 (vectorizable_load): Likewise.
1594 (vectorizable_store): Likewise.
1595 * ubsan.c (ubsan_encode_value): Likewise.
1596 * varasm.c (output_constant): Likewise.
1598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1599 Alan Hayward <alan.hayward@arm.com>
1600 David Sherwood <david.sherwood@arm.com>
1602 * coretypes.h (scalar_mode): New class.
1603 * machmode.h (scalar_mode): Likewise.
1604 (scalar_mode::includes_p): New function.
1605 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1606 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1607 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1608 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1609 * expmed.c (store_bit_field_1): Likewise.
1610 (extract_bit_field_1): Likewise.
1611 * expr.c (write_complex_part): Likewise.
1612 (read_complex_part): Likewise.
1613 (emit_move_complex_push): Likewise.
1614 (expand_expr_real_2): Likewise.
1615 * function.c (assign_parm_setup_reg): Likewise.
1616 (assign_parms_unsplit_complex): Likewise.
1617 * optabs.c (expand_binop): Likewise.
1618 * rtlanal.c (subreg_get_info): Likewise.
1619 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1620 * varasm.c (output_constant_pool_2): Likewise.
1622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1623 Alan Hayward <alan.hayward@arm.com>
1624 David Sherwood <david.sherwood@arm.com>
1626 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1628 (expmed_mult_highpart_optab): Likewise.
1629 (expmed_mult_highpart): Likewise.
1631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1632 Alan Hayward <alan.hayward@arm.com>
1633 David Sherwood <david.sherwood@arm.com>
1635 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1636 instead of a machine_mode.
1637 (builtin_memset_read_str): Likewise.
1638 * builtins.c (c_readstr): Likewise.
1639 (builtin_memcpy_read_str): Likewise.
1640 (builtin_strncpy_read_str): Likewise.
1641 (builtin_memset_read_str): Likewise.
1642 (builtin_memset_gen_str): Likewise.
1643 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1644 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1645 instead of a machine_mode.
1646 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1648 (make_extraction): Likewise.
1649 (try_widen_shift_mode): Take and return scalar_int_modes instead
1651 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1652 a scalar_int_mode instead of a machine_mode.
1653 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1654 (avr_addr_space_pointer_mode): Likewise.
1655 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1656 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1657 (msp430_unwind_word_mode): Likewise.
1658 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1659 (spu_addr_space_pointer_mode): Likewise.
1660 (spu_addr_space_address_mode): Likewise.
1661 (spu_libgcc_cmp_return_mode): Likewise.
1662 (spu_libgcc_shift_count_mode): Likewise.
1663 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1664 (rl78_addr_space_pointer_mode): Likewise.
1665 (fl78_unwind_word_mode): Likewise.
1666 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1668 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1669 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1670 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1671 (mips_valid_pointer_mode): Likewise.
1672 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1673 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1674 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1676 (ft32_addr_space_address_mode): Likewise.
1677 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1678 scalar_int_mode instead of a machine_mode.
1679 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1681 (m32c_addr_space_address_mode): Likewise.
1682 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1683 (rs6000_eh_return_filter_mode): Likewise.
1684 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1685 (rs6000_eh_return_filter_mode): Likewise.
1686 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1687 (s390_libgcc_shift_count_mode): Likewise.
1688 (s390_unwind_word_mode): Likewise.
1689 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1691 * target.def (mode_rep_extended): Likewise.
1692 (valid_pointer_mode): Likewise.
1693 (addr_space.valid_pointer_mode): Likewise.
1694 (eh_return_filter_mode): Return a scalar_int_mode rather than
1696 (libgcc_cmp_return_mode): Likewise.
1697 (libgcc_shift_count_mode): Likewise.
1698 (unwind_word_mode): Likewise.
1699 (addr_space.pointer_mode): Likewise.
1700 (addr_space.address_mode): Likewise.
1701 * doc/tm.texi: Regenerate.
1702 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1704 (do_jump): Use scalar_int_mode for local variables.
1705 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1706 rather than a machine_mode.
1707 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1708 (scompare_loc_descriptor_wide): Likewise.
1709 (scompare_loc_descriptor_narrow): Likewise.
1710 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1712 * except.c (sjlj_emit_dispatch_table): Likewise.
1713 (expand_builtin_eh_copy_values): Likewise.
1714 * explow.c (convert_memory_address_addr_space_1): Likewise.
1715 Take a scalar_int_mode rather than a machine_mode.
1716 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1717 than a machine_mode.
1718 (memory_address_addr_space): Use scalar_int_mode for local variables.
1719 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1720 rather than a machine_mode.
1721 * expmed.c (mask_rtx): Likewise.
1722 (init_expmed_one_conv): Likewise.
1723 (expand_mult_highpart_adjust): Likewise.
1724 (extract_high_half): Likewise.
1725 (expmed_mult_highpart_optab): Likewise.
1726 (expmed_mult_highpart): Likewise.
1727 (expand_smod_pow2): Likewise.
1728 (expand_sdiv_pow2): Likewise.
1729 (emit_store_flag_int): Likewise.
1730 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1732 (extract_low_bits): Likewise.
1733 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1735 * expr.c (pieces_addr::adjust): Likewise.
1736 (can_store_by_pieces): Likewise.
1737 (store_by_pieces): Likewise.
1738 (clear_by_pieces_1): Likewise.
1739 (expand_expr_addr_expr_1): Likewise.
1740 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1741 (expand_expr_real_1): Likewise.
1742 (try_casesi): Likewise.
1743 * final.c (shorten_branches): Likewise.
1744 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1745 type of "mode" to machine_mode.
1746 * internal-fn.c (expand_arith_overflow_result_store): Take a
1747 scalar_int_mode rather than a machine_mode.
1748 (expand_mul_overflow): Use scalar_int_mode for local variables.
1749 * loop-doloop.c (doloop_modify): Likewise.
1750 (doloop_optimize): Likewise.
1751 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1752 than a machine_mode.
1753 (expand_doubleword_shift_condmove): Likewise.
1754 (expand_doubleword_shift): Likewise.
1755 (expand_doubleword_clz): Likewise.
1756 (expand_doubleword_popcount): Likewise.
1757 (expand_doubleword_parity): Likewise.
1758 (expand_absneg_bit): Use scalar_int_mode for local variables.
1759 (prepare_float_lib_cmp): Likewise.
1760 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1761 rather than a machine_mode.
1762 (convert_memory_address_addr_space): Likewise.
1763 (get_mode_bounds): Likewise.
1764 (get_address_mode): Return a scalar_int_mode rather than a
1766 * rtlanal.c (get_address_mode): Likewise.
1767 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1768 than a machine_mode.
1769 * targhooks.c (default_mode_rep_extended): Likewise.
1770 (default_valid_pointer_mode): Likewise.
1771 (default_addr_space_valid_pointer_mode): Likewise.
1772 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1773 than a machine_mode.
1774 (default_libgcc_cmp_return_mode): Likewise.
1775 (default_libgcc_shift_count_mode): Likewise.
1776 (default_unwind_word_mode): Likewise.
1777 (default_addr_space_pointer_mode): Likewise.
1778 (default_addr_space_address_mode): Likewise.
1779 * targhooks.h (default_eh_return_filter_mode): Likewise.
1780 (default_libgcc_cmp_return_mode): Likewise.
1781 (default_libgcc_shift_count_mode): Likewise.
1782 (default_unwind_word_mode): Likewise.
1783 (default_addr_space_pointer_mode): Likewise.
1784 (default_addr_space_address_mode): Likewise.
1785 (default_mode_rep_extended): Take a scalar_int_mode rather than
1787 (default_valid_pointer_mode): Likewise.
1788 (default_addr_space_valid_pointer_mode): Likewise.
1789 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1791 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
1792 rather than a machine_mode.
1793 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
1794 for local variables.
1795 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1796 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
1797 than a machine_mode.
1799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
1804 the mode argument to scalar_int_mode.
1805 (do_jump_by_parts_zero_rtx): Likewise.
1806 (do_jump_by_parts_equality_rtx): Likewise.
1807 (do_jump_by_parts_greater): Take a mode argument.
1808 (do_jump_by_parts_equality): Likewise.
1809 (do_jump_1): Update calls accordingly.
1811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1812 Alan Hayward <alan.hayward@arm.com>
1813 David Sherwood <david.sherwood@arm.com>
1815 * is-a.h (safe_dyn_cast): New function.
1816 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
1817 (jump_table_for_label): Likewise.
1818 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
1819 instead of an rtx_insn *.
1820 (shorten_branches): Use dyn_cast instead of LABEL_P and
1821 JUMP_TABLE_DATA_P. Use jump_table_for_label and
1822 rtx_jump_table_data::get_data_mode.
1823 (final_scan_insn): Likewise.
1825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1826 Alan Hayward <alan.hayward@arm.com>
1827 David Sherwood <david.sherwood@arm.com>
1829 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1830 trying to combine a full-register integer set with a subreg
1833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1834 Alan Hayward <alan.hayward@arm.com>
1835 David Sherwood <david.sherwood@arm.com>
1837 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1838 that is always either address_mode or pointer_mode.
1840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1841 Alan Hayward <alan.hayward@arm.com>
1842 David Sherwood <david.sherwood@arm.com>
1844 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1845 when the two are known to be equal.
1847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1848 Alan Hayward <alan.hayward@arm.com>
1849 David Sherwood <david.sherwood@arm.com>
1851 * simplify-rtx.c (simplify_const_unary_operation): Use
1852 is_a <scalar_int_mode> instead of checking for a nonzero
1853 precision. Forcibly convert op_mode to a scalar_int_mode
1854 in that case. More clearly differentiate the operand and
1855 result modes and use the former when deciding what the value
1856 of a count-bits operation should be. Use is_int_mode instead
1857 of checking for a MODE_INT. Remove redundant check for whether
1858 this mode has a zero precision.
1860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * optabs.c (widen_leading): Change the type of the mode argument
1865 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1866 (widen_bswap): Likewise.
1867 (expand_parity): Likewise.
1868 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1869 (expand_ffs): Likewise.
1870 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1879 (expand_expr_real_1): Update call accordingly.
1881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1882 Alan Hayward <alan.hayward@arm.com>
1883 David Sherwood <david.sherwood@arm.com>
1885 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1886 value_mode arguments. Use scalar_int_mode internally.
1887 (store_bit_field_1): Rename the new integer mode from imode
1888 to op0_mode and use it instead of GET_MODE (op0). Update calls
1889 to store_split_bit_field, store_bit_field_using_insv and
1890 store_fixed_bit_field.
1891 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1892 Use scalar_int_mode internally. Use a bit count rather than a mode
1893 when calculating the largest bit size for get_best_mode.
1894 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1895 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1896 Remove assertion that OP0 has a scalar integer mode.
1897 (store_split_bit_field): Add op0_mode and value_mode arguments.
1898 Update calls to extract_fixed_bit_field.
1899 (extract_bit_field_using_extv): Add an op0_mode argument.
1900 Use scalar_int_mode internally.
1901 (extract_bit_field_1): Rename the new integer mode from imode to
1902 op0_mode and use it instead of GET_MODE (op0). Update calls to
1903 extract_split_bit_field, extract_bit_field_using_extv and
1904 extract_fixed_bit_field.
1905 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1906 to extract_split_bit_field and extract_fixed_bit_field_1.
1907 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1908 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1910 (extract_split_bit_field): Add an op0_mode argument. Update call
1911 to extract_fixed_bit_field.
1913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1914 Alan Hayward <alan.hayward@arm.com>
1915 David Sherwood <david.sherwood@arm.com>
1917 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1918 * explow.c (hard_function_value): Likewise.
1919 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1920 convert_to_mode call outside the loop.
1921 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1922 for the mode iterator. Require the mode specified by max_pieces
1924 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1926 (copy_blkmode_to_reg): Likewise.
1927 (set_storage_via_setmem): Likewise.
1928 * optabs.c (prepare_cmp_insn): Likewise.
1929 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1930 * stor-layout.c (finish_bitfield_representative): Likewise.
1932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1933 Alan Hayward <alan.hayward@arm.com>
1934 David Sherwood <david.sherwood@arm.com>
1936 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1937 * expr.c (convert_move): Use them.
1938 (convert_modes): Likewise.
1939 (store_expr_with_bounds): Likewise.
1941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1942 Alan Hayward <alan.hayward@arm.com>
1943 David Sherwood <david.sherwood@arm.com>
1945 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1946 parameter for the mode of "x". Remove the "known_x", "known_mode"
1947 and "known_ret" arguments. Change the type of the mode argument
1949 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1950 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1951 (reg_num_sign_bit_copies_for_combine): Likewise.
1952 * rtlanal.c (nonzero_bits1): Likewise.
1953 (num_sign_bit_copies1): Likewise.
1954 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1955 (reg_num_sign_bit_copies_general): Likewise.
1956 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1957 (reg_nonzero_bits_general): Likewise.
1959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1960 Alan Hayward <alan.hayward@arm.com>
1961 David Sherwood <david.sherwood@arm.com>
1963 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1964 than in subroutines. Return 1 for non-integer modes.
1965 (cached_num_sign_bit_copies): Change the type of the mode parameter
1967 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1968 classes. Handle CONST_INT_P first and then check whether X also
1969 has a scalar integer mode. Check the same thing for inner registers
1970 of a SUBREG and for values that are being extended or truncated.
1972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1973 Alan Hayward <alan.hayward@arm.com>
1974 David Sherwood <david.sherwood@arm.com>
1976 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1977 in subroutines. Return the mode mask for non-integer modes.
1978 (cached_nonzero_bits): Change the type of the mode parameter
1980 (nonzero_bits1): Likewise. Remove early exit for other mode
1981 classes. Handle CONST_INT_P first and then check whether X
1982 also has a scalar integer mode.
1984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1985 Alan Hayward <alan.hayward@arm.com>
1986 David Sherwood <david.sherwood@arm.com>
1988 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1989 Return a scalar_int_mode and assert that the size is greater than
1991 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1992 (op_by_pieces_d::op_by_pieces_d): Likewise.
1993 (op_by_pieces_d::run): Likewise.
1994 (can_store_by_pieces): Likewise.
1996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1997 Alan Hayward <alan.hayward@arm.com>
1998 David Sherwood <david.sherwood@arm.com>
2000 * combine.c (extract_left_shift): Add a mode argument and update
2002 (make_compound_operation_int): Change the type of the mode parameter
2003 to scalar_int_mode and update the call to extract_left_shift.
2005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2006 Alan Hayward <alan.hayward@arm.com>
2007 David Sherwood <david.sherwood@arm.com>
2009 * combine.c (simplify_and_const_int): Change the type of the mode
2010 parameter to scalar_int_mode.
2011 (simplify_and_const_int_1): Likewise. Update recursive call.
2013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2014 Alan Hayward <alan.hayward@arm.com>
2015 David Sherwood <david.sherwood@arm.com>
2017 * combine.c (simplify_compare_const): Check that the mode is a
2018 scalar_int_mode (rather than VOIDmode) before testing its
2020 (simplify_comparison): Move COMPARISON_P handling out of the
2021 loop and restrict the latter part of the loop to scalar_int_modes.
2022 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2023 and when considering SUBREG_REGs. Use is_int_mode instead of
2024 checking GET_MODE_CLASS against MODE_INT.
2026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2027 Alan Hayward <alan.hayward@arm.com>
2028 David Sherwood <david.sherwood@arm.com>
2030 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2031 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2032 shift_unit_mode and for modes involved in scalar shifts.
2034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2035 Alan Hayward <alan.hayward@arm.com>
2036 David Sherwood <david.sherwood@arm.com>
2038 * combine.c (force_int_to_mode): New function, split out from...
2039 (force_to_mode): ...here. Keep xmode up-to-date and use it
2040 instead of GET_MODE (x).
2042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2043 Alan Hayward <alan.hayward@arm.com>
2044 David Sherwood <david.sherwood@arm.com>
2046 * optabs-query.h (extraction_insn::struct_mode): Change type to
2047 opt_scalar_int_mode and update comment.
2048 (extraction_insn::field_mode): Change type to scalar_int_mode.
2049 (extraction_insn::pos_mode): Likewise.
2050 * combine.c (make_extraction): Update accordingly.
2051 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2052 (get_optab_extraction_insn): Likewise.
2053 * recog.c (simplify_while_replacing): Likewise.
2054 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2055 parameter to opt_scalar_int_mode.
2057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2058 Alan Hayward <alan.hayward@arm.com>
2059 David Sherwood <david.sherwood@arm.com>
2061 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2062 to a scalar_int_mode instead of a machine_mode.
2063 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2064 (get_best_mode): Return a boolean and use a pointer argument to store
2065 the selected mode. Replace the limit mode parameter with a bit limit.
2066 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2067 for the values returned by bit_field_mode_iterator::next_mode.
2068 (store_bit_field): Update call to get_best_mode.
2069 (store_fixed_bit_field): Likewise.
2070 (extract_fixed_bit_field): Likewise.
2071 * expr.c (optimize_bitfield_assignment_op): Likewise.
2072 * fold-const.c (optimize_bit_field_compare): Likewise.
2073 (fold_truth_andor_1): Likewise.
2074 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2075 Update for new type of m_mode.
2076 (get_best_mode): As above.
2078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2079 Alan Hayward <alan.hayward@arm.com>
2080 David Sherwood <david.sherwood@arm.com>
2082 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2083 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2084 (store_bit_field): Check is_a <scalar_int_mode> before calling
2085 strict_volatile_bitfield_p.
2086 (extract_bit_field): Likewise.
2088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2089 Alan Hayward <alan.hayward@arm.com>
2090 David Sherwood <david.sherwood@arm.com>
2092 * target.def (cstore_mode): Return a scalar_int_mode.
2093 * doc/tm.texi: Regenerate.
2094 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2095 * targhooks.h (default_cstore_mode): Likewise.
2096 * targhooks.c (default_cstore_mode): Likewise, using a forced
2098 * expmed.c (emit_cstore): Expect the target of the cstore to be
2101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2102 Alan Hayward <alan.hayward@arm.com>
2103 David Sherwood <david.sherwood@arm.com>
2105 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2107 (niter_desc): Likewise mode.
2108 (iv_analyze): Add a mode parameter.
2110 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2111 and change its type to scalar_int_mode.
2112 * loop-iv.c: Update commentary at head of file.
2113 (iv_constant): Pass the mode paraeter before the rtx it describes
2114 and change its type to scalar_int_mode. Remove VOIDmode handling.
2115 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2116 (iv_extend): Likewise.
2117 (shorten_into_mode): Likewise.
2118 (iv_add): Use scalar_int_mode.
2119 (iv_mult): Likewise.
2120 (iv_shift): Likewise.
2121 (canonicalize_iv_subregs): Likewise.
2122 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2123 it describes and change its mode to scalar_int_mode. Also change
2124 the type of the returned inner_mode to scalar_int_mode.
2125 (get_biv_step): Likewise, turning outer_mode from a pointer
2126 into a direct parameter. Update call to get_biv_step_1.
2127 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2128 iv_constant and get_biv_step.
2129 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2130 and change its type to scalar_int_mode. Don't initialise iv->mode
2131 to VOIDmode and remove later checks for its still being VOIDmode.
2132 Update calls to iv_analyze_op and iv_analyze_expr. Check
2133 is_a <scalar_int_mode> when changing the mode under consideration.
2134 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2135 Update call to iv_analyze_expr.
2136 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2137 inner register is not also a scalar_int_mode. Update call to
2139 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2140 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2141 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2142 separate mode class checks. Update calls to iv_analyze. Remove
2143 fix-up of VOIDmodes after iv_analyze_biv.
2144 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2145 don't have a scalar_int_mode. Update call to biv_p.
2147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2148 Alan Hayward <alan.hayward@arm.com>
2149 David Sherwood <david.sherwood@arm.com>
2151 * cfgexpand.c (convert_debug_memory_address): Use
2152 as_a <scalar_int_mode>.
2153 * combine.c (expand_compound_operation): Likewise.
2154 (make_extraction): Likewise.
2155 (change_zero_ext): Likewise.
2156 (simplify_comparison): Likewise.
2157 * cse.c (cse_insn): Likewise.
2158 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2159 (mem_loc_descriptor): Likewise.
2160 (loc_descriptor): Likewise.
2161 * expmed.c (init_expmed_one_mode): Likewise.
2162 (synth_mult): Likewise.
2163 (emit_store_flag_1): Likewise.
2164 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2165 of a comparison with size.
2166 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2167 (reduce_to_bit_field_precision): Likewise.
2168 * function.c (expand_function_end): Likewise.
2169 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2170 * loop-doloop.c (doloop_modify): Likewise.
2171 * optabs.c (expand_binop): Likewise.
2172 (expand_unop): Likewise.
2173 (expand_copysign_absneg): Likewise.
2174 (prepare_cmp_insn): Likewise.
2175 (maybe_legitimize_operand): Likewise.
2176 * recog.c (const_scalar_int_operand): Likewise.
2177 * rtlanal.c (get_address_mode): Likewise.
2178 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2179 (simplify_cond_clz_ctz): Likewise.
2180 * tree-nested.c (get_nl_goto_field): Likewise.
2181 * tree.c (build_vector_type_for_mode): Likewise.
2182 * var-tracking.c (use_narrower_mode): Likewise.
2184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2185 Alan Hayward <alan.hayward@arm.com>
2186 David Sherwood <david.sherwood@arm.com>
2188 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2189 * builtins.c (expand_builtin_signbit): Use it.
2190 * cfgexpand.c (expand_debug_expr): Likewise.
2191 * dojump.c (do_jump): Likewise.
2192 (do_compare_and_jump): Likewise.
2193 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2194 * expmed.c (make_tree): Likewise.
2195 * expr.c (expand_expr_real_2): Likewise.
2196 (expand_expr_real_1): Likewise.
2197 (try_casesi): Likewise.
2198 * fold-const-call.c (fold_const_call_ss): Likewise.
2199 * fold-const.c (unextend): Likewise.
2200 (extract_muldiv_1): Likewise.
2201 (fold_single_bit_test): Likewise.
2202 (native_encode_int): Likewise.
2203 (native_encode_string): Likewise.
2204 (native_interpret_int): Likewise.
2205 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2206 * internal-fn.c (expand_addsub_overflow): Likewise.
2207 (expand_neg_overflow): Likewise.
2208 (expand_mul_overflow): Likewise.
2209 (expand_arith_overflow): Likewise.
2210 * match.pd: Likewise.
2211 * stor-layout.c (layout_type): Likewise.
2212 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2213 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2214 * tree-ssanames.c (get_range_info): Likewise.
2215 * tree-switch-conversion.c (array_value_type) Likewise.
2216 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2217 (vect_recog_divmod_pattern): Likewise.
2218 (vect_recog_mixed_size_cond_pattern): Likewise.
2219 * tree-vrp.c (extract_range_basic): Likewise.
2220 (simplify_float_conversion_using_ranges): Likewise.
2221 * tree.c (int_fits_type_p): Likewise.
2222 * ubsan.c (instrument_bool_enum_load): Likewise.
2223 * varasm.c (mergeable_string_section): Likewise.
2224 (narrowing_initializer_constant_valid_p): Likewise.
2225 (output_constant): Likewise.
2227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2228 Alan Hayward <alan.hayward@arm.com>
2229 David Sherwood <david.sherwood@arm.com>
2231 * machmode.h (NARROWEST_INT_MODE): New macro.
2232 * expr.c (alignment_for_piecewise_move): Use it instead of
2233 GET_CLASS_NARROWEST_MODE (MODE_INT).
2234 (push_block): Likewise.
2235 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2237 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2240 Alan Hayward <alan.hayward@arm.com>
2241 David Sherwood <david.sherwood@arm.com>
2243 * postreload.c (move2add_valid_value_p): Change the type of the
2244 mode parameter to scalar_int_mode.
2245 (move2add_use_add2_insn): Add a mode parameter and use it instead
2247 (move2add_use_add3_insn): Likewise.
2248 (reload_cse_move2add): Update accordingly.
2250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2251 Alan Hayward <alan.hayward@arm.com>
2252 David Sherwood <david.sherwood@arm.com>
2254 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2256 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2257 * optabs.c (expand_unop): Likewise.
2259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2260 Alan Hayward <alan.hayward@arm.com>
2261 David Sherwood <david.sherwood@arm.com>
2263 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2264 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2265 (popcount_loc_descriptor): Likewise.
2266 (bswap_loc_descriptor): Likewise.
2267 (rotate_loc_descriptor): Likewise.
2268 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2269 calling the functions above.
2271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2277 (try_combine): Likewise.
2278 (simplify_if_then_else): Likewise.
2279 * cse.c (cse_insn): Likewise.
2280 * dwarf2out.c (mem_loc_descriptor): Likewise.
2281 * emit-rtl.c (gen_lowpart_common): Likewise.
2282 * simplify-rtx.c (simplify_truncation): Likewise.
2283 (simplify_binary_operation_1): Likewise.
2284 (simplify_const_relational_operation): Likewise.
2285 (simplify_ternary_operation): Likewise.
2286 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2289 Alan Hayward <alan.hayward@arm.com>
2290 David Sherwood <david.sherwood@arm.com>
2292 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2293 * reload.c (push_reload): Likewise.
2294 (find_reloads): Likewise.
2296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2297 Alan Hayward <alan.hayward@arm.com>
2298 David Sherwood <david.sherwood@arm.com>
2300 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2301 (make_compound_operation_int): Likewise.
2302 (change_zero_ext): Likewise.
2303 * expr.c (convert_move): Likewise.
2304 (convert_modes): Likewise.
2305 * fwprop.c (forward_propagate_subreg): Likewise.
2306 * loop-iv.c (get_biv_step_1): Likewise.
2307 * optabs.c (widen_operand): Likewise.
2308 * postreload.c (move2add_valid_value_p): Likewise.
2309 * recog.c (simplify_while_replacing): Likewise.
2310 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2311 (simplify_binary_operation_1): Likewise. Remove redundant
2312 mode equality check.
2314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2315 Alan Hayward <alan.hayward@arm.com>
2316 David Sherwood <david.sherwood@arm.com>
2318 * combine.c (combine_simplify_rtx): Add checks for
2319 is_a <scalar_int_mode>.
2320 (simplify_if_then_else): Likewise.
2321 (make_field_assignment): Likewise.
2322 (simplify_comparison): Likewise.
2323 * ifcvt.c (noce_try_bitop): Likewise.
2324 * loop-invariant.c (canonicalize_address_mult): Likewise.
2325 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2328 Alan Hayward <alan.hayward@arm.com>
2329 David Sherwood <david.sherwood@arm.com>
2331 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2332 is_a <scalar_int_mode> instead of != BLKmode.
2334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2335 Alan Hayward <alan.hayward@arm.com>
2336 David Sherwood <david.sherwood@arm.com>
2338 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2339 instead of != VOIDmode.
2340 * combine.c (if_then_else_cond): Likewise.
2341 (change_zero_ext): Likewise.
2342 * dwarf2out.c (mem_loc_descriptor): Likewise.
2343 (loc_descriptor): Likewise.
2344 * rtlanal.c (canonicalize_condition): Likewise.
2345 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2348 Alan Hayward <alan.hayward@arm.com>
2349 David Sherwood <david.sherwood@arm.com>
2351 * simplify-rtx.c (simplify_binary_operation_1): Use
2352 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2355 Alan Hayward <alan.hayward@arm.com>
2356 David Sherwood <david.sherwood@arm.com>
2358 * wide-int.h (int_traits<unsigned char>) New class.
2359 (int_traits<unsigned short>) Likewise.
2360 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2361 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2363 * combine.c (set_nonzero_bits_and_sign_copies): Use
2364 is_a <scalar_int_mode>.
2365 (find_split_point): Likewise.
2366 (combine_simplify_rtx): Likewise.
2367 (simplify_logical): Likewise.
2368 (expand_compound_operation): Likewise.
2369 (expand_field_assignment): Likewise.
2370 (make_compound_operation): Likewise.
2371 (extended_count): Likewise.
2372 (change_zero_ext): Likewise.
2373 (simplify_comparison): Likewise.
2374 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2375 (ucompare_loc_descriptor): Likewise.
2376 (minmax_loc_descriptor): Likewise.
2377 (mem_loc_descriptor): Likewise.
2378 (loc_descriptor): Likewise.
2379 * expmed.c (init_expmed_one_mode): Likewise.
2380 * lra-constraints.c (lra_constraint_offset): Likewise.
2381 * optabs.c (prepare_libcall_arg): Likewise.
2382 * postreload.c (move2add_note_store): Likewise.
2383 * reload.c (operands_match_p): Likewise.
2384 * rtl.h (load_extend_op): Likewise.
2385 * rtlhooks.c (gen_lowpart_general): Likewise.
2386 * simplify-rtx.c (simplify_truncation): Likewise.
2387 (simplify_unary_operation_1): Likewise.
2388 (simplify_binary_operation_1): Likewise.
2389 (simplify_const_binary_operation): Likewise.
2390 (simplify_const_relational_operation): Likewise.
2391 (simplify_subreg): Likewise.
2392 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2393 * var-tracking.c (adjust_mems): Likewise.
2394 (prepare_call_arguments): Likewise.
2396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2397 Alan Hayward <alan.hayward@arm.com>
2398 David Sherwood <david.sherwood@arm.com>
2400 * machmode.h (is_int_mode): New fuction.
2401 * combine.c (find_split_point): Use it.
2402 (combine_simplify_rtx): Likewise.
2403 (simplify_if_then_else): Likewise.
2404 (simplify_set): Likewise.
2405 (simplify_shift_const_1): Likewise.
2406 (simplify_comparison): Likewise.
2407 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2408 * cse.c (notreg_cost): Likewise.
2409 (cse_insn): Likewise.
2410 * cselib.c (cselib_lookup_1): Likewise.
2411 * dojump.c (do_jump_1): Likewise.
2412 (do_compare_rtx_and_jump): Likewise.
2413 * dse.c (get_call_args): Likewise.
2414 * dwarf2out.c (rtl_for_decl_init): Likewise.
2415 (native_encode_initializer): Likewise.
2416 * expmed.c (emit_store_flag_1): Likewise.
2417 (emit_store_flag): Likewise.
2418 * expr.c (convert_modes): Likewise.
2419 (store_field): Likewise.
2420 (expand_expr_real_1): Likewise.
2421 * fold-const.c (fold_read_from_constant_string): Likewise.
2422 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2423 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2424 * optabs.c (expand_binop): Likewise.
2425 (expand_unop): Likewise.
2426 (expand_abs_nojump): Likewise.
2427 (expand_one_cmpl_abs_nojump): Likewise.
2428 * simplify-rtx.c (mode_signbit_p): Likewise.
2429 (val_signbit_p): Likewise.
2430 (val_signbit_known_set_p): Likewise.
2431 (val_signbit_known_clear_p): Likewise.
2432 (simplify_relational_operation_1): Likewise.
2433 * tree.c (vector_type_mode): Likewise.
2435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * machmode.h (smallest_mode_for_size): Fix formatting.
2440 (smallest_int_mode_for_size): New function.
2441 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2442 instead of smallest_mode_for_size.
2443 * combine.c (make_extraction): Likewise.
2444 * config/arc/arc.c (arc_expand_movmem): Likewise.
2445 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2446 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2447 * config/s390/s390.c (s390_expand_insv): Likewise.
2448 * config/sparc/sparc.c (assign_int_registers): Likewise.
2449 * config/spu/spu.c (spu_function_value): Likewise.
2450 (spu_function_arg): Likewise.
2451 * coverage.c (get_gcov_type): Likewise.
2452 (get_gcov_unsigned_t): Likewise.
2453 * dse.c (find_shift_sequence): Likewise.
2454 * expmed.c (store_bit_field_1): Likewise.
2455 * expr.c (convert_move): Likewise.
2456 (store_field): Likewise.
2457 * internal-fn.c (expand_arith_overflow): Likewise.
2458 * optabs-query.c (get_best_extraction_insn): Likewise.
2459 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2460 * stor-layout.c (layout_type): Likewise.
2461 (initialize_sizetypes): Likewise.
2462 * targhooks.c (default_get_mask_mode): Likewise.
2463 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2466 Alan Hayward <alan.hayward@arm.com>
2467 David Sherwood <david.sherwood@arm.com>
2469 * machmode.h (opt_mode::else_blk): New function.
2470 (int_mode_for_mode): Declare.
2471 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2472 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2474 * cfgexpand.c (expand_debug_expr): Likewise.
2475 * combine.c (gen_lowpart_or_truncate): Likewise.
2476 (gen_lowpart_for_combine): Likewise.
2477 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2478 * config/avr/avr.c (avr_to_int_mode): Likewise.
2479 (avr_out_plus_1): Likewise.
2480 (avr_out_plus): Likewise.
2481 (avr_out_round): Likewise.
2482 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2483 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2484 (s390_expand_vcond): Likewise.
2485 * config/spu/spu.c (spu_split_immediate): Likewise.
2486 (spu_expand_mov): Likewise.
2487 * dse.c (get_stored_val): Likewise.
2488 * expmed.c (store_bit_field_1): Likewise.
2489 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2491 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2492 (extract_low_bits): Likewise.
2493 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2494 handling rather than repeating the check.
2495 (emit_group_store): Likewise.
2496 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2497 * optabs.c (expand_absneg_bit): Likewise.
2498 (expand_copysign_absneg): Likewise.
2499 (expand_copysign_bit): Likewise.
2500 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2501 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2502 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2503 * var-tracking.c (prepare_call_arguments): Likewise.
2504 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2505 int_mode_for_mode instead of mode_for_size.
2506 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2509 Alan Hayward <alan.hayward@arm.com>
2510 David Sherwood <david.sherwood@arm.com>
2512 * machmode.h (int_mode_for_size): New function.
2513 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2514 instead of mode_for_size.
2515 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2517 * combine.c (expand_field_assignment): Use int_mode_for_size
2518 instead of mode_for_size.
2519 (make_extraction): Likewise.
2520 (simplify_shift_const_1): Likewise.
2521 (simplify_comparison): Likewise.
2522 * dojump.c (do_jump): Likewise.
2523 * dwarf2out.c (mem_loc_descriptor): Likewise.
2524 * emit-rtl.c (init_derived_machine_modes): Likewise.
2525 * expmed.c (flip_storage_order): Likewise.
2526 (convert_extracted_bit_field): Likewise.
2527 * expr.c (copy_blkmode_from_reg): Likewise.
2528 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2529 * internal-fn.c (expand_mul_overflow): Likewise.
2530 * lower-subreg.c (simple_move): Likewise.
2531 * optabs-libfuncs.c (init_optabs): Likewise.
2532 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2533 * tree.c (vector_type_mode): Likewise.
2534 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2535 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2536 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2537 * tree-vect-stmts.c (vectorizable_load): Likewise.
2538 (vectorizable_store): Likewise.
2540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2541 Alan Hayward <alan.hayward@arm.com>
2542 David Sherwood <david.sherwood@arm.com>
2544 * coretypes.h (pod_mode): New type.
2545 (scalar_int_mode_pod): New typedef.
2546 * machmode.h (pod_mode): New class.
2547 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2548 * genmodes.c (emit_mode_int_n): Update accordingly.
2549 * lower-subreg.h (target_lower_subreg): Change type to
2550 scalar_int_mode_pod.
2551 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2552 scalar_int_mode_pod.
2554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2555 Alan Hayward <alan.hayward@arm.com>
2556 David Sherwood <david.sherwood@arm.com>
2558 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2559 machine_mode to scalar_int_mode.
2560 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2561 (rs6000_option_override_internal): Remove cast to int.
2562 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2563 machine_mode to scalar_int_mode.
2564 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2565 (rs6000_option_override_internal): Remove cast to int.
2566 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2567 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2569 * config/s390/s390.c (s390_expand_builtin): Likewise.
2570 * coretypes.h (scalar_int_mode): New type.
2571 (opt_scalar_int_mode): New typedef.
2572 * machmode.h (scalar_int_mode): New class.
2573 (scalar_int_mode::includes_p): New function.
2574 (byte_mode): Change type to scalar_int_mode.
2575 (word_mode): Likewise.
2576 (ptr_mode): Likewise.
2577 * emit-rtl.c (byte_mode): Likewise.
2578 (word_mode): Likewise.
2579 (ptr_mode): Likewise.
2580 (init_derived_machine_modes): Update accordingly.
2581 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2582 and MODE_PARTIAL_INT.
2583 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2584 opt_scalar_int_mode.
2586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2587 Alan Hayward <alan.hayward@arm.com>
2588 David Sherwood <david.sherwood@arm.com>
2590 * target.def (libgcc_floating_mode_supported_p): Take a
2592 * doc/tm.texi: Regenerate.
2593 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2595 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2596 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2600 Alan Hayward <alan.hayward@arm.com>
2601 David Sherwood <david.sherwood@arm.com>
2603 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2604 * doc/tm.texi: Regenerate.
2605 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2606 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2607 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2608 * targhooks.h (default_floatn_mode): Likewise.
2609 * targhooks.c (default_floatn_mode): Likewise.
2610 * tree.c (build_common_tree_nodes): Update accordingly.
2612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2613 Alan Hayward <alan.hayward@arm.com>
2614 David Sherwood <david.sherwood@arm.com>
2616 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2617 (mode_iterator::iterate_p): Likewise.
2618 (mode_iterator::get_wider): Likewise.
2619 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2622 Alan Hayward <alan.hayward@arm.com>
2623 David Sherwood <david.sherwood@arm.com>
2625 * coretypes.h (opt_scalar_float_mode): New typedef.
2626 * machmode.h (float_mode_for_size): New function.
2627 * emit-rtl.c (double_mode): Delete.
2628 (init_emit_once): Use float_mode_for_size.
2629 * stor-layout.c (layout_type): Likewise.
2630 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2633 Alan Hayward <alan.hayward@arm.com>
2634 David Sherwood <david.sherwood@arm.com>
2636 * output.h (assemble_real): Take a scalar_float_mode.
2637 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2638 * config/arm/arm.md (consttable_4): Likewise.
2639 (consttable_8): Likewise.
2640 (consttable_16): Likewise.
2641 * config/mips/mips.md (consttable_float): Likewise.
2642 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2643 * varasm.c (assemble_real): Take a scalar_float_mode.
2644 (output_constant_pool_2): Update accordingly.
2645 (output_constant): Likewise.
2647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2648 Alan Hayward <alan.hayward@arm.com>
2649 David Sherwood <david.sherwood@arm.com>
2651 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2652 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2653 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2654 (native_encode_real): Likewise.
2655 (native_interpret_real): Likewise.
2656 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2657 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2660 Alan Hayward <alan.hayward@arm.com>
2661 David Sherwood <david.sherwood@arm.com>
2663 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2664 <scalar_float_mode>. Simplify.
2665 (gen_extend_conv_libfunc): Likewise.
2667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2668 Alan Hayward <alan.hayward@arm.com>
2669 David Sherwood <david.sherwood@arm.com>
2671 * coretypes.h (scalar_float_mode): New type.
2672 * machmode.h (mode_traits::from_int): Use machine_mode if
2673 USE_ENUM_MODES is defined.
2674 (is_a): New function.
2676 (dyn_cast): Likewise.
2677 (scalar_float_mode): New class.
2678 (scalar_float_mode::includes_p): New function.
2679 (is_float_mode): Likewise.
2680 * gdbhooks.py (MachineModePrinter): New class.
2681 (build_pretty_printer): Use it for scalar_float_mode.
2682 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2683 (format_helper::format_helper): Turn into a template.
2684 * genmodes.c (get_mode_class): New function.
2685 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2686 or machine_mode if none.
2687 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2688 as_a <scalar_float_mode>.
2689 * dwarf2out.c (mem_loc_descriptor): Likewise.
2690 (insert_float): Likewise.
2691 (add_const_value_attribute): Likewise.
2692 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2693 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2694 (expand_unop): Update accordingly.
2695 (expand_abs_nojump): Likewise.
2696 (expand_copysign_absneg): Take a scalar_float_mode.
2697 (expand_copysign_bit): Likewise.
2698 (expand_copysign): Update accordingly.
2700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2701 Alan Hayward <alan.hayward@arm.com>
2702 David Sherwood <david.sherwood@arm.com>
2704 * coretypes.h (opt_mode): New class.
2705 * machmode.h (opt_mode): Likewise.
2706 (opt_mode::else_void): New function.
2707 (opt_mode::require): Likewise.
2708 (opt_mode::exists): Likewise.
2709 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2710 (GET_MODE_2XWIDER_MODE): Likewise.
2711 (mode_iterator::get_wider): Update accordingly.
2712 (mode_iterator::get_2xwider): Likewise.
2713 (mode_iterator::get_known_wider): Likewise, turning into a template.
2714 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2715 forcing a wider mode to exist.
2716 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2717 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2718 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2719 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2720 * lower-subreg.c (init_lower_subreg): Likewise.
2721 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2722 on the final iteration.
2723 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2724 a wider mode exists before asking for a move pattern.
2725 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2726 forcing a wider mode to exist.
2727 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2728 returning false if no such mode exists.
2729 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2730 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2731 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2732 Avoid checking for a MODE_INT if we already know the mode is not a
2734 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2735 forcing a wider mode to exist.
2736 (expmed_mult_highpart_optab): Likewise.
2737 (expmed_mult_highpart): Likewise.
2738 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2740 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2741 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2742 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2743 * internal-fn.c (expand_mul_overflow): Update use of
2744 GET_MODE_2XWIDER_MODE.
2745 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2746 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2747 GET_MODE_WIDER_MODE.
2748 (convert_plusminus_to_widen): Likewise.
2749 * tree-switch-conversion.c (array_value_type): Likewise.
2750 * var-tracking.c (emit_note_insn_var_location): Likewise.
2751 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2752 Return false inside rather than outside the loop if no wider mode
2754 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2755 and GET_MODE_2XWIDER_MODE
2756 (can_compare_p): Use else_void.
2757 * gdbhooks.py (OptMachineModePrinter): New class.
2758 (build_pretty_printer): Use it for opt_mode.
2760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2761 Alan Hayward <alan.hayward@arm.com>
2762 David Sherwood <david.sherwood@arm.com>
2764 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2765 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2768 Alan Hayward <alan.hayward@arm.com>
2769 David Sherwood <david.sherwood@arm.com>
2771 * machmode.h (mode_traits): New structure.
2772 (get_narrowest_mode): New function.
2773 (mode_iterator::start): Likewise.
2774 (mode_iterator::iterate_p): Likewise.
2775 (mode_iterator::get_wider): Likewise.
2776 (mode_iterator::get_known_wider): Likewise.
2777 (mode_iterator::get_2xwider): Likewise.
2778 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2779 (FOR_EACH_MODE): Likewise.
2780 (FOR_EACH_MODE_FROM): Likewise.
2781 (FOR_EACH_MODE_UNTIL): Likewise.
2782 (FOR_EACH_WIDER_MODE): Likewise.
2783 (FOR_EACH_2XWIDER_MODE): Likewise.
2784 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2785 * combine.c (simplify_comparison): Likewise
2786 * config/i386/i386.c (type_natural_mode): Likewise.
2787 * cse.c (cse_insn): Likewise.
2788 * dse.c (find_shift_sequence): Likewise.
2789 * emit-rtl.c (init_derived_machine_modes): Likewise.
2790 (init_emit_once): Likewise.
2791 * explow.c (hard_function_value): Likewise.
2792 * expmed.c (extract_fixed_bit_field_1): Likewise.
2793 (extract_bit_field_1): Likewise.
2794 (expand_divmod): Likewise.
2795 (emit_store_flag_1): Likewise.
2796 * expr.c (init_expr_target): Likewise.
2797 (convert_move): Likewise.
2798 (alignment_for_piecewise_move): Likewise.
2799 (widest_int_mode_for_size): Likewise.
2800 (emit_block_move_via_movmem): Likewise.
2801 (copy_blkmode_to_reg): Likewise.
2802 (set_storage_via_setmem): Likewise.
2803 (compress_float_constant): Likewise.
2804 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2805 * optabs-query.c (get_best_extraction_insn): Likewise.
2806 * optabs.c (expand_binop): Likewise.
2807 (expand_twoval_unop): Likewise.
2808 (expand_twoval_binop): Likewise.
2809 (widen_leading): Likewise.
2810 (widen_bswap): Likewise.
2811 (expand_parity): Likewise.
2812 (expand_unop): Likewise.
2813 (prepare_cmp_insn): Likewise.
2814 (prepare_float_lib_cmp): Likewise.
2815 (expand_float): Likewise.
2816 (expand_fix): Likewise.
2817 (expand_sfix_optab): Likewise.
2818 * postreload.c (move2add_use_add2_insn): Likewise.
2819 * reg-stack.c (reg_to_stack): Likewise.
2820 * reginfo.c (choose_hard_reg_mode): Likewise.
2821 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2822 * stor-layout.c (mode_for_size): Likewise.
2823 (smallest_mode_for_size): Likewise.
2824 (mode_for_vector): Likewise.
2825 (finish_bitfield_representative): Likewise.
2826 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2827 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2828 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2829 * var-tracking.c (prepare_call_arguments): Likewise.
2831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2832 Alan Hayward <alan.hayward@arm.com>
2833 David Sherwood <david.sherwood@arm.com>
2835 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2836 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2837 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2838 * machmode.h (mode_size): Move earlier in file.
2839 (mode_precision): Likewise.
2840 (mode_inner): Likewise.
2841 (mode_nunits): Likewise.
2842 (mode_unit_size): Likewise.
2843 (unit_unit_precision): Likewise.
2844 (mode_wider): Likewise.
2845 (mode_2xwider): Likewise.
2846 (machine_mode): New class.
2847 (mode_to_bytes): New function.
2848 (mode_to_bits): Likewise.
2849 (mode_to_precision): Likewise.
2850 (mode_to_inner): Likewise.
2851 (mode_to_unit_size): Likewise.
2852 (mode_to_unit_precision): Likewise.
2853 (mode_to_nunits): Likewise.
2854 (GET_MODE_SIZE): Use mode_to_bytes.
2855 (GET_MODE_BITSIZE): Use mode_to_bits.
2856 (GET_MODE_PRECISION): Use mode_to_precision.
2857 (GET_MODE_INNER): Use mode_to_inner.
2858 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2859 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2860 (GET_MODE_NUNITS): Use mode_to_nunits.
2861 * system.h (ALWAYS_INLINE): New macro.
2862 * config/powerpcspe/powerpcspe-c.c
2863 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2864 int for arg1_mode and arg2_mode.
2866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2867 Alan Hayward <alan.hayward@arm.com>
2868 David Sherwood <david.sherwood@arm.com>
2870 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2871 Prefix mode names with E_ in case statements.
2872 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2873 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2874 (aarch64_split_simd_move): Likewise.
2875 (aarch64_gen_storewb_pair): Likewise.
2876 (aarch64_gen_loadwb_pair): Likewise.
2877 (aarch64_gen_store_pair): Likewise.
2878 (aarch64_gen_load_pair): Likewise.
2879 (aarch64_get_condition_code_1): Likewise.
2880 (aarch64_constant_pool_reload_icode): Likewise.
2881 (get_rsqrte_type): Likewise.
2882 (get_rsqrts_type): Likewise.
2883 (get_recpe_type): Likewise.
2884 (get_recps_type): Likewise.
2885 (aarch64_gimplify_va_arg_expr): Likewise.
2886 (aarch64_simd_container_mode): Likewise.
2887 (aarch64_emit_load_exclusive): Likewise.
2888 (aarch64_emit_store_exclusive): Likewise.
2889 (aarch64_expand_compare_and_swap): Likewise.
2890 (aarch64_gen_atomic_cas): Likewise.
2891 (aarch64_emit_bic): Likewise.
2892 (aarch64_emit_atomic_swap): Likewise.
2893 (aarch64_emit_atomic_load_op): Likewise.
2894 (aarch64_evpc_trn): Likewise.
2895 (aarch64_evpc_uzp): Likewise.
2896 (aarch64_evpc_zip): Likewise.
2897 (aarch64_evpc_ext): Likewise.
2898 (aarch64_evpc_rev): Likewise.
2899 (aarch64_evpc_dup): Likewise.
2900 (aarch64_gen_ccmp_first): Likewise.
2901 (aarch64_gen_ccmp_next): Likewise.
2902 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2903 (alpha_emit_xfloating_libcall): Likewise.
2904 (emit_insxl): Likewise.
2905 (alpha_arg_type): Likewise.
2906 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2907 (arc_preferred_simd_mode): Likewise.
2908 (arc_secondary_reload): Likewise.
2909 (get_arc_condition_code): Likewise.
2910 (arc_print_operand): Likewise.
2911 (arc_legitimate_constant_p): Likewise.
2912 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2913 * config/arc/arc.md (casesi_load): Likewise.
2914 (casesi_compact_jump): Likewise.
2915 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2916 (cc_use_register): Likewise.
2917 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2918 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2919 (arm_init_iwmmxt_builtins): Likewise.
2920 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2921 (neon_expand_vector_init): Likewise.
2922 (arm_attr_length_move_neon): Likewise.
2923 (maybe_get_arm_condition_code): Likewise.
2924 (arm_emit_vector_const): Likewise.
2925 (arm_preferred_simd_mode): Likewise.
2926 (arm_output_iwmmxt_tinsr): Likewise.
2927 (thumb1_output_casesi): Likewise.
2928 (thumb2_output_casesi): Likewise.
2929 (arm_emit_load_exclusive): Likewise.
2930 (arm_emit_store_exclusive): Likewise.
2931 (arm_expand_compare_and_swap): Likewise.
2932 (arm_evpc_neon_vuzp): Likewise.
2933 (arm_evpc_neon_vzip): Likewise.
2934 (arm_evpc_neon_vrev): Likewise.
2935 (arm_evpc_neon_vtrn): Likewise.
2936 (arm_evpc_neon_vext): Likewise.
2937 (arm_validize_comparison): Likewise.
2938 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2939 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2940 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2941 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2942 (c6x_preferred_simd_mode): Likewise.
2943 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2944 (epiphany_rtx_costs): Likewise.
2945 * config/epiphany/predicates.md (proper_comparison_operator):
2947 * config/frv/frv.c (condexec_memory_operand): Likewise.
2948 (frv_emit_move): Likewise.
2949 (output_move_single): Likewise.
2950 (output_condmove_single): Likewise.
2951 (frv_hard_regno_mode_ok): Likewise.
2952 (frv_matching_accg_mode): Likewise.
2953 * config/h8300/h8300.c (split_adds_subs): Likewise.
2954 (h8300_rtx_costs): Likewise.
2955 (h8300_print_operand): Likewise.
2956 (compute_mov_length): Likewise.
2957 (output_logical_op): Likewise.
2958 (compute_logical_op_length): Likewise.
2959 (compute_logical_op_cc): Likewise.
2960 (h8300_shift_needs_scratch_p): Likewise.
2961 (output_a_shift): Likewise.
2962 (compute_a_shift_length): Likewise.
2963 (compute_a_shift_cc): Likewise.
2964 (expand_a_rotate): Likewise.
2965 (output_a_rotate): Likewise.
2966 * config/i386/i386.c (classify_argument): Likewise.
2967 (function_arg_advance_32): Likewise.
2968 (function_arg_32): Likewise.
2969 (function_arg_64): Likewise.
2970 (function_value_64): Likewise.
2971 (ix86_gimplify_va_arg): Likewise.
2972 (ix86_legitimate_constant_p): Likewise.
2973 (put_condition_code): Likewise.
2974 (split_double_mode): Likewise.
2975 (ix86_avx256_split_vector_move_misalign): Likewise.
2976 (ix86_expand_vector_logical_operator): Likewise.
2977 (ix86_split_idivmod): Likewise.
2978 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2979 (ix86_build_const_vector): Likewise.
2980 (ix86_build_signbit_mask): Likewise.
2981 (ix86_match_ccmode): Likewise.
2982 (ix86_cc_modes_compatible): Likewise.
2983 (ix86_expand_branch): Likewise.
2984 (ix86_expand_sse_cmp): Likewise.
2985 (ix86_expand_sse_movcc): Likewise.
2986 (ix86_expand_int_sse_cmp): Likewise.
2987 (ix86_expand_vec_perm_vpermi2): Likewise.
2988 (ix86_expand_vec_perm): Likewise.
2989 (ix86_expand_sse_unpack): Likewise.
2990 (ix86_expand_int_addcc): Likewise.
2991 (ix86_split_to_parts): Likewise.
2992 (ix86_vectorize_builtin_gather): Likewise.
2993 (ix86_vectorize_builtin_scatter): Likewise.
2994 (avx_vpermilp_parallel): Likewise.
2995 (inline_memory_move_cost): Likewise.
2996 (ix86_tieable_integer_mode_p): Likewise.
2997 (x86_maybe_negate_const_int): Likewise.
2998 (ix86_expand_vector_init_duplicate): Likewise.
2999 (ix86_expand_vector_init_one_nonzero): Likewise.
3000 (ix86_expand_vector_init_one_var): Likewise.
3001 (ix86_expand_vector_init_concat): Likewise.
3002 (ix86_expand_vector_init_interleave): Likewise.
3003 (ix86_expand_vector_init_general): Likewise.
3004 (ix86_expand_vector_set): Likewise.
3005 (ix86_expand_vector_extract): Likewise.
3006 (emit_reduc_half): Likewise.
3007 (ix86_emit_i387_round): Likewise.
3008 (ix86_mangle_type): Likewise.
3009 (ix86_expand_round_sse4): Likewise.
3010 (expand_vec_perm_blend): Likewise.
3011 (canonicalize_vector_int_perm): Likewise.
3012 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3013 (expand_vec_perm_1): Likewise.
3014 (expand_vec_perm_interleave3): Likewise.
3015 (expand_vec_perm_even_odd_pack): Likewise.
3016 (expand_vec_perm_even_odd_1): Likewise.
3017 (expand_vec_perm_broadcast_1): Likewise.
3018 (ix86_vectorize_vec_perm_const_ok): Likewise.
3019 (ix86_expand_vecop_qihi): Likewise.
3020 (ix86_expand_mul_widen_hilo): Likewise.
3021 (ix86_expand_sse2_abs): Likewise.
3022 (ix86_expand_pextr): Likewise.
3023 (ix86_expand_pinsr): Likewise.
3024 (ix86_preferred_simd_mode): Likewise.
3025 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3026 * config/i386/sse.md (*andnot<mode>3): Likewise.
3027 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3028 (*<code><mode>3): Likewise.
3029 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3030 (ia64_expand_atomic_op): Likewise.
3031 (ia64_arg_type): Likewise.
3032 (ia64_mode_to_int): Likewise.
3033 (ia64_scalar_mode_supported_p): Likewise.
3034 (ia64_vector_mode_supported_p): Likewise.
3035 (expand_vec_perm_broadcast): Likewise.
3036 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3037 (iq2000_function_arg_advance): Likewise.
3038 (iq2000_function_arg): Likewise.
3039 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3040 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3041 (m68k_libcall_value): Likewise.
3042 (m68k_function_value): Likewise.
3043 (sched_attr_op_type): Likewise.
3044 * config/mcore/mcore.c (mcore_output_move): Likewise.
3045 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3047 (microblaze_function_arg): Likewise.
3048 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3049 (mips_print_operand): Likewise.
3050 (mips_mode_ok_for_mov_fmt_p): Likewise.
3051 (mips_vector_mode_supported_p): Likewise.
3052 (mips_preferred_simd_mode): Likewise.
3053 (mips_expand_vpc_loongson_even_odd): Likewise.
3054 (mips_expand_vec_unpack): Likewise.
3055 (mips_expand_vi_broadcast): Likewise.
3056 (mips_expand_vector_init): Likewise.
3057 (mips_expand_vec_reduc): Likewise.
3058 (mips_expand_msa_cmp): Likewise.
3059 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3060 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3061 (cc_flags_for_mode): Likewise.
3062 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3063 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3064 (nds32_output_casesi_pc_relative): Likewise.
3065 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3066 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3067 (nvptx_gen_unpack): Likewise.
3068 (nvptx_gen_pack): Likewise.
3069 (nvptx_gen_shuffle): Likewise.
3070 (nvptx_gen_wcast): Likewise.
3071 (nvptx_preferred_simd_mode): Likewise.
3072 * config/pa/pa.c (pa_secondary_reload): Likewise.
3073 * config/pa/predicates.md (base14_operand): Likewise.
3074 * config/powerpcspe/powerpcspe-c.c
3075 (altivec_resolve_overloaded_builtin): Likewise.
3076 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3078 (rs6000_preferred_simd_mode): Likewise.
3079 (output_vec_const_move): Likewise.
3080 (rs6000_expand_vector_extract): Likewise.
3081 (rs6000_split_vec_extract_var): Likewise.
3082 (reg_offset_addressing_ok_p): Likewise.
3083 (rs6000_legitimate_offset_address_p): Likewise.
3084 (rs6000_legitimize_address): Likewise.
3085 (rs6000_emit_set_const): Likewise.
3086 (rs6000_const_vec): Likewise.
3087 (rs6000_emit_move): Likewise.
3088 (spe_build_register_parallel): Likewise.
3089 (rs6000_darwin64_record_arg_recurse): Likewise.
3090 (swap_selector_for_mode): Likewise.
3091 (spe_init_builtins): Likewise.
3092 (paired_init_builtins): Likewise.
3093 (altivec_init_builtins): Likewise.
3094 (do_load_for_compare): Likewise.
3095 (rs6000_generate_compare): Likewise.
3096 (rs6000_expand_float128_convert): Likewise.
3097 (emit_load_locked): Likewise.
3098 (emit_store_conditional): Likewise.
3099 (rs6000_output_function_epilogue): Likewise.
3100 (rs6000_handle_altivec_attribute): Likewise.
3101 (rs6000_function_value): Likewise.
3102 (emit_fusion_gpr_load): Likewise.
3103 (emit_fusion_p9_load): Likewise.
3104 (emit_fusion_p9_store): Likewise.
3105 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3106 (fusion_gpr_mem_load): Likewise.
3107 (fusion_addis_mem_combo_load): Likewise.
3108 (fusion_addis_mem_combo_store): Likewise.
3109 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3110 (fusion_gpr_mem_load): Likewise.
3111 (fusion_addis_mem_combo_load): Likewise.
3112 (fusion_addis_mem_combo_store): Likewise.
3113 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3115 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3116 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3117 (rs6000_preferred_simd_mode): Likewise.
3118 (output_vec_const_move): Likewise.
3119 (rs6000_expand_vector_extract): Likewise.
3120 (rs6000_split_vec_extract_var): Likewise.
3121 (reg_offset_addressing_ok_p): Likewise.
3122 (rs6000_legitimate_offset_address_p): Likewise.
3123 (rs6000_legitimize_address): Likewise.
3124 (rs6000_emit_set_const): Likewise.
3125 (rs6000_const_vec): Likewise.
3126 (rs6000_emit_move): Likewise.
3127 (rs6000_darwin64_record_arg_recurse): Likewise.
3128 (swap_selector_for_mode): Likewise.
3129 (paired_init_builtins): Likewise.
3130 (altivec_init_builtins): Likewise.
3131 (rs6000_expand_float128_convert): Likewise.
3132 (emit_load_locked): Likewise.
3133 (emit_store_conditional): Likewise.
3134 (rs6000_output_function_epilogue): Likewise.
3135 (rs6000_handle_altivec_attribute): Likewise.
3136 (rs6000_function_value): Likewise.
3137 (emit_fusion_gpr_load): Likewise.
3138 (emit_fusion_p9_load): Likewise.
3139 (emit_fusion_p9_store): Likewise.
3140 * config/rx/rx.c (rx_gen_move_template): Likewise.
3141 (flags_from_mode): Likewise.
3142 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3143 (s390_slb_comparison): Likewise.
3144 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3145 (s390_vector_mode_supported_p): Likewise.
3146 (s390_cc_modes_compatible): Likewise.
3147 (s390_match_ccmode_set): Likewise.
3148 (s390_canonicalize_comparison): Likewise.
3149 (s390_emit_compare_and_swap): Likewise.
3150 (s390_branch_condition_mask): Likewise.
3151 (s390_rtx_costs): Likewise.
3152 (s390_secondary_reload): Likewise.
3153 (__SECONDARY_RELOAD_CASE): Likewise.
3154 (s390_expand_cs): Likewise.
3155 (s390_preferred_simd_mode): Likewise.
3156 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3157 * config/sh/sh.c (sh_print_operand): Likewise.
3158 (dump_table): Likewise.
3159 (sh_secondary_reload): Likewise.
3160 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3161 * config/sh/sh.md (casesi_worker_1): Likewise.
3162 (casesi_worker_2): Likewise.
3163 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3164 (fcc_comparison_operator): Likewise.
3165 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3166 (emit_soft_tfmode_cvt): Likewise.
3167 (sparc_preferred_simd_mode): Likewise.
3168 (output_cbranch): Likewise.
3169 (sparc_print_operand): Likewise.
3170 (sparc_expand_vec_perm_bmask): Likewise.
3171 (vector_init_bshuffle): Likewise.
3172 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3173 (spu_vector_mode_supported_p): Likewise.
3174 (spu_expand_insv): Likewise.
3175 (spu_emit_branch_or_set): Likewise.
3176 (spu_handle_vector_attribute): Likewise.
3177 (spu_builtin_splats): Likewise.
3178 (spu_builtin_extract): Likewise.
3179 (spu_builtin_promote): Likewise.
3180 (spu_expand_sign_extend): Likewise.
3181 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3182 (tilegx_simd_int): Likewise.
3183 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3184 (tilepro_simd_int): Likewise.
3185 * config/v850/v850.c (const_double_split): Likewise.
3186 (v850_print_operand): Likewise.
3187 (ep_memory_offset): Likewise.
3188 * config/vax/vax.c (vax_rtx_costs): Likewise.
3189 (vax_output_int_move): Likewise.
3190 (vax_output_int_add): Likewise.
3191 (vax_output_int_subtract): Likewise.
3192 * config/visium/predicates.md (visium_branch_operator): Likewise.
3193 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3194 (visium_print_operand_address): Likewise.
3195 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3196 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3197 (xtensa_expand_conditional_branch): Likewise.
3198 (xtensa_copy_incoming_a7): Likewise.
3199 (xtensa_output_literal): Likewise.
3200 * dfp.c (decimal_real_maxval): Likewise.
3201 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3204 Alan Hayward <alan.hayward@arm.com>
3205 David Sherwood <david.sherwood@arm.com>
3207 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3208 (mode_nunits_inline): Likewise.
3209 (mode_inner_inline): Likewise.
3210 (mode_unit_size_inline): Likewise.
3211 (mode_unit_precision_inline): Likewise.
3212 (emit_insn_modes_h): Likewise. Also emit a #define of the
3214 (emit_mode_wider): Add an E_ prefix to mode names.
3215 (emit_mode_complex): Likewise.
3216 (emit_mode_inner): Likewise.
3217 (emit_mode_adjustments): Likewise.
3218 (emit_mode_int_n): Likewise.
3219 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3220 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3221 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3222 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3223 (CRC32_BUILTIN, ENTRY): Likewise.
3224 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3225 (aarch64_pop_regs): Likewise.
3226 (aarch64_process_components): Likewise.
3227 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3228 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3229 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3230 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3231 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3232 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3234 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3235 (pcmpistr): Likewise.
3236 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3237 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3238 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3240 * config/rl78/rl78.c (mduc_regs): Likewise.
3241 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3242 (htm_expand_builtin): Likewise.
3243 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3244 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3245 * config/xtensa/xtensa.c (print_operand): Likewise.
3246 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3247 (NUM_MODE_VECTOR_INT): Likewise.
3248 * genoutput.c (null_operand): Likewise.
3249 (output_operand_data): Likewise.
3250 * genrecog.c (print_parameter_value): Likewise.
3251 * lra.c (debug_operand_data): Likewise.
3253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3255 * dwarf2out.c (scompare_loc_descriptor_wide)
3256 (scompare_loc_descriptor_narrow): New functions, split out from...
3257 (scompare_loc_descriptor): ...here.
3258 * expmed.c (emit_store_flag_int): New function, split out from...
3259 (emit_store_flag): ...here.
3261 2017-08-30 Richard Biener <rguenther@suse.de>
3263 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3264 (dwarf2out_early_finish): Move setting of AT_pubnames from
3265 early debug output to early finish.
3267 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3269 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3270 and -mdata-region to the assembler.
3272 2017-08-30 Richard Biener <rguenther@suse.de>
3274 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3276 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3277 (dwarf2out_early_global_decl): Only generate a DIE for the
3278 abstract origin if it doesn't already exist or is a declaration DIE.
3279 (resolve_addr): Do not add the linkage name twice when
3280 generating a stub DIE for the DW_TAG_GNU_call_site target.
3282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3284 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3285 Use machine_mode rather than int for arg1_mode.
3287 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3290 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3291 that the second argument of the built-in functions to unpack
3292 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3293 switch statement instead a lot of if statements.
3294 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3295 Allow 64-bit values to be in Altivec registers as well as
3296 traditional floating point registers.
3297 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3299 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3301 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3302 MAX_REGS_PER_ADDRESS == 1.
3304 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3306 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3307 * config/i386/i386.c (ix86_option_override_internal): Simplify
3308 setting of opts->x_flag_entry.
3310 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3311 Jakub Jelinek <jakub@redhat.com>
3312 Richard Biener <rguenther@suse.de>
3314 PR tree-optimization/81503
3315 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3316 folded constant fits in the target type; reorder tests for clarity.
3318 2017-08-29 Martin Liska <mliska@suse.cz>
3320 * passes.def: Include pass_lower_switch.
3321 * stmt.c (dump_case_nodes): Remove and move to
3322 tree-switch-conversion.
3323 (case_values_threshold): Likewise.
3324 (expand_switch_as_decision_tree_p): Likewise.
3325 (emit_case_decision_tree): Likewise.
3326 (expand_case): Likewise.
3327 (balance_case_nodes): Likewise.
3328 (node_has_low_bound): Likewise.
3329 (node_has_high_bound): Likewise.
3330 (node_is_bounded): Likewise.
3331 (emit_case_nodes): Likewise.
3332 (struct simple_case_node): New struct.
3333 (add_case_node): Remove.
3334 (emit_case_dispatch_table): Use vector instead of case_list.
3335 (reset_out_edges_aux): Remove.
3336 (compute_cases_per_edge): Likewise.
3337 (expand_case): Build list of simple_case_node.
3338 (expand_sjlj_dispatch_table): Use it.
3339 * tree-switch-conversion.c (struct case_node): Moved from
3340 stmt.c and adjusted.
3341 (emit_case_nodes): Likewise.
3342 (node_has_low_bound): Likewise.
3343 (node_has_high_bound): Likewise.
3344 (node_is_bounded): Likewise.
3345 (case_values_threshold): Likewise.
3346 (reset_out_edges_aux): Likewise.
3347 (compute_cases_per_edge): Likewise.
3348 (add_case_node): Likewise.
3349 (dump_case_nodes): Likewise.
3350 (balance_case_nodes): Likewise.
3351 (expand_switch_as_decision_tree_p): Likewise.
3352 (emit_jump): Likewise.
3353 (emit_case_decision_tree): Likewise.
3354 (try_switch_expansion): Likewise.
3355 (do_jump_if_equal): Likewise.
3356 (emit_cmp_and_jump_insns): Likewise.
3357 (fix_phi_operands_for_edge): New function.
3358 (record_phi_operand_mapping): Likewise.
3359 (class pass_lower_switch): New pass.
3360 (pass_lower_switch::execute): New function.
3361 (make_pass_lower_switch): Likewise.
3362 (conditional_probability):
3363 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3364 * tree-pass.h: Add make_pass_lower_switch.
3366 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3369 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3372 2017-08-29 Richard Biener <rguenther@suse.de>
3374 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3375 we do not add a DW_AT_inline attribute twice.
3376 (gen_subprogram_die): Remove code setting DW_AT_inline on
3377 DECL_ABSTRACT_P nodes.
3379 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3381 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3382 calls to internal functions.
3383 (gimplify_modify_expr): Likewise.
3384 * tree-call-cdce.c (use_internal_fn): Likewise.
3385 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3386 (convert_to_divmod): Set the nothrow flag.
3387 * tree-if-conv.c (predicate_mem_writes): Likewise.
3388 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3389 (vectorizable_call): Likewise.
3390 (vectorizable_store): Likewise.
3391 (vectorizable_load): Likewise.
3392 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3393 (vect_recog_mask_conversion_pattern): Likewise.
3395 2017-08-29 Martin Liska <mliska@suse.cz>
3398 * gcc.c (driver_handle_option): Add new argument.
3399 * opts-common.c (handle_option): Pass
3400 target_option_override_hook.
3401 * opts-global.c (lang_handle_option): Add new option.
3402 (set_default_handlers): Add new argument.
3403 (decode_options): Likewise.
3404 * opts.c (target_handle_option): Likewise.
3405 (common_handle_option): Call target_option_override_hook.
3406 * opts.h (struct cl_option_handler_func): Add hook for
3407 target option override.
3408 (struct cl_option_handlers): Likewise.
3409 (set_default_handlers): Add new argument.
3410 (decode_options): Likewise.
3411 (common_handle_option): Likewise.
3412 (target_handle_option): Likewise.
3413 * toplev.c (toplev::main): Pass targetm.target_option.override
3416 2017-08-29 Richard Biener <rguenther@suse.de>
3417 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3419 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3420 life to the active subtree.
3422 2017-08-28 Jeff Law <law@redhat.com>
3424 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3425 derive_equivalences.
3426 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3428 (edge_info::derive_equivalences): New private member function
3430 * tree-ssa-dom.c (class edge_info): Changed from a struct
3431 to a class. Add ctor/dtor, methods and data members.
3432 (edge_info::edge_info): Renamed from allocate_edge_info.
3433 Initialize additional members.
3434 (edge_info::~edge_info): New.
3435 (free_dom_edge_info): Delete the edge info.
3436 (record_edge_info): Use new class & associated member functions.
3437 Tighten forms for testing for edge equivalences.
3438 (record_temporary_equivalences): Iterate over the simple
3439 equivalences rather than assuming there's only one per edge.
3440 (cprop_into_successor_phis): Iterate over the simple
3441 equivalences rather than assuming there's only one per edge.
3442 (optimize_stmt): Use operand_equal_p rather than pointer
3443 equality for mini-DSE code.
3445 2017-08-28 Nathan Sidwell <nathan@acm.org>
3447 * gcc.c (execute): Fold SIGPIPE handling into switch
3448 statement. Adjust internal error message.
3450 2017-08-28 Richard Biener <rguenther@suse.de>
3453 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3454 Do nothing for removed DIEs.
3456 2017-08-28 Richard Biener <rguenther@suse.de>
3458 PR tree-optimization/81977
3459 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3462 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3465 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3466 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3467 using targetm.gen_mem_thread_fence.
3469 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3472 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3473 predicate to register_operand. Reorder operands.
3474 (*btr<mode>): Ditto.
3475 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3476 (*btr<mode>_mask): Ditto.
3478 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3480 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3481 * config/rs6000/xmmintrin.h: New file.
3482 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3484 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3487 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3488 parameter and_insn and return it.
3489 (recombine_lvx_pattern): Insert a copy to ensure availability of
3490 the base register of the copied masking operation at the point of
3491 the instruction replacement.
3492 (recombine_stvx_pattern): Likewise.
3494 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3496 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3497 undocumented switches.
3498 (-mpower9-dform-vector): Likewise.
3499 (-mpower9-dform): Likewise.
3500 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3501 comments to delete references to -mpower9-dform* switches.
3502 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3503 Delete reference to -mpower9-dform* switches, test for
3504 -mpower9-vector instead.
3505 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3506 (OTHER_P9_VECTOR_MASKS): Likewise.
3507 (POWERPC_MASKS): Likewise.
3508 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3509 tests against -mpower9-dform* to -mpower9-vector. Delete code
3510 that checked for -mpower9-dform* consistancy with other options.
3511 Add test for -mpower9-misc to enable other power9 switches.
3512 (rs6000_init_hard_regno_mode_ok): Likewise.
3513 (rs6000_option_override_internal): Likewise.
3514 (rs6000_emit_prologue): Likewise.
3515 (rs6000_emit_epilogue): Likewise.
3516 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3517 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3518 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3520 (emit_fusion_p9_store): Likewise.
3521 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3522 resetting these macros if the assembler does not support ISA 3.0
3524 (TARGET_P9_DFORM_VECTOR): Likewise.
3525 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3526 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3528 2017-08-25 Alan Modra <amodra@gmail.com>
3530 PR rtl-optimization/81747
3531 * cse.c (cse_extended_basic_block): Don't attempt to record
3532 equivalences for degenerate conditional branches that branch
3533 to their fall-through.
3535 2017-08-24 Martin Sebor <msebor@redhat.com>
3538 * gimple-fold.c (size_must_be_zero_p): New function.
3539 (gimple_fold_builtin_memory_op): Call it.
3541 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3543 * config/rs6000/mm_malloc.h: New file.
3545 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3547 PR tree-optimization/81913
3548 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3549 analysis when either IVs in condition can wrap.
3551 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3553 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3554 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3556 2017-08-24 Richard Biener <rguenther@suse.de>
3559 * targhooks.c (default_target_can_inline_p): Properly
3560 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3561 is present and always compare.
3562 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3563 infer -mfpmath=sse from TARGET_SSE_P.
3564 (ix86_can_inline_p): Properly use target_option_default_node when
3565 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3567 2017-08-24 Richard Biener <rguenther@suse.de>
3570 * dwarf2out.c (output_die): Handle flag_generate_offload like
3572 (output_comp_unit): Likewise.
3573 (gen_array_type_die): Likewise.
3574 (dwarf2out_early_finish): Likewise.
3575 (note_variable_value_in_expr): Likewise.
3576 (dwarf2out_finish): Likewise. Adjust assert.
3577 * cgraphunit.c (symbol_table::compile): Move setting of
3578 flag_generate_offload earlier ...
3579 (symbol_table::finalize_compilation_unit): ... here, before
3580 early debug finalization.
3582 2017-08-24 Richard Biener <rguenther@suse.de>
3584 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3585 and ipa-fnsummary.h.
3586 (ix86_can_inline_p): When ix86_fpmath flags do not match
3587 check whether the callee uses FP math at all.
3589 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3592 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3593 instead of TYPE_PRECISION.
3595 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3596 Alan Hayward <alan.hayward@arm.com>
3597 David Sherwood <david.sherwood@arm.com>
3599 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3601 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3603 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3605 (ix86_frame::stack_realign_allocate): New field.
3606 (struct machine_frame_state): Modify comments.
3607 (machine_frame_state::sp_realigned_fp_end): New field.
3608 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3610 (sp_valid_at): Add assertion to assure no attempt to access invalid
3611 offset of a realigned stack.
3612 (fp_valid_at): Likewise.
3613 (choose_baseaddr): Modify comments.
3614 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3615 ix86_expand_prologue.
3616 (ix86_expand_prologue): Modify stack realignment and allocation.
3617 (ix86_expand_epilogue): Modify comments.
3618 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3619 avx2_runtime, avx512f, and avx512f_runtime.
3621 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3623 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3624 (mstackrealign): Do not init to -1.
3625 * config/i386/i386.c (ix86_option_override_internal):
3626 Check opts_set, not opts when setting default value of
3627 opts->x_ix86_force_align_arg_pointer.
3629 2017-08-23 Richard Biener <rguenther@suse.de>
3631 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3632 lang_hooks.decl_printable_name.
3633 * print-rtl-function.c (print_rtx_function): Likewise.
3634 * tree-pretty-print.c (dump_function_header): Likewise.
3636 2017-08-23 Richard Biener <rguenther@suse.de>
3639 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3640 -g0 at compile-time.
3642 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3645 * doc/sourcebuild.texi (Other hardware attributes):
3648 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3651 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3652 Fix single-use check.
3654 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3656 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3658 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3660 * doc/install.texi: Modify to add more details on running selected
3663 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3665 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3666 is combined with -mabi=ms.
3667 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3670 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3672 PR tree-optimization/81488
3673 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3674 and cached_basis fields.
3675 (MAX_SPREAD): New constant.
3676 (alloc_cand_and_find_basis): Initialize new fields.
3677 (clear_visited): New function.
3678 (create_phi_basis_1): Rename from create_phi_basis, set visited
3679 and cached_basis fields.
3680 (create_phi_basis): New wrapper function.
3681 (phi_add_costs_1): Rename from phi_add_costs, add spread
3682 parameter, set visited field, short-circuit when limits reached.
3683 (phi_add_costs): New wrapper function.
3684 (record_phi_increments_1): Rename from record_phi_increments, set
3686 (record_phi_increments): New wrapper function.
3687 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3688 (phi_incr_cost): New wrapper function.
3689 (all_phi_incrs_profitable_1): Rename from
3690 all_phi_incrs_profitable, set visited field.
3691 (all_phi_incrs_profitable): New wrapper function.
3693 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3694 Alan Hayward <alan.hayward@arm.com>
3695 David Sherwood <david.sherwood@arm.com>
3697 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3698 that takes the outer and inner modes.
3699 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3700 comparison as the canonical test for a paradoxical subreg.
3701 * combine.c (simplify_set): Use paradoxical_subreg_p.
3702 (make_extraction): Likewise.
3703 (force_to_mode): Likewise.
3704 (rtx_equal_for_field_assignment_p): Likewise.
3705 (gen_lowpart_for_combine): Likewise.
3706 (simplify_comparison): Likewise.
3707 * cse.c (equiv_constant): Likewise.
3708 * expmed.c (store_bit_field_1): Likewise.
3709 * final.c (alter_subreg): Likewise.
3710 * fwprop.c (propagate_rtx): Likewise.
3711 (forward_propagate_subreg): Likewise.
3712 * ira-conflicts.c (ira_build_conflicts): Likewise.
3713 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3714 * lra-constraints.c (curr_insn_transform): Likewise.
3715 (split_reg): Likewise.
3716 * lra-eliminations.c (move_plus_up): Likewise.
3717 (lra_eliminate_regs_1): Likewise.
3718 * recog.c (general_operand): Likewise.
3719 * ree.c (combine_reaching_defs): Likewise.
3720 * reload.c (push_reload): Likewise.
3721 (find_reloads): Likewise.
3722 * reload1.c (elimination_effects): Likewise.
3723 (compute_reload_subreg_offset): Likewise.
3724 (choose_reload_regs): Likewise.
3725 * rtlanal.c (subreg_lsb_1): Likewise.
3726 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3727 (simplify_subreg): Likewise.
3728 * var-tracking.c (track_loc_p): Likewise.
3729 * emit-rtl.c (byte_lowpart_offset): Likewise.
3730 (paradoxical_subreg_p): Delete out-of-line definition.
3732 2017-08-22 Jeff Law <law@redhat.com>
3734 PR tree-optimization/81741
3735 PR tree-optimization/71947
3736 * tree-ssa-dom.c: Include tree-inline.h.
3737 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3738 equivalences if one is more expensive to compute than the other.
3739 * tree-ssa-scopedtables.h (class const_or_copies): Make
3740 record_const_or_copy_raw method private.
3741 (class avail_exprs_stack): New method simplify_binary_operation.
3742 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3743 avail_exprs_stack::simplify_binary_operation as needed.
3744 (avail_exprs_stack::simplify_binary_operation): New function.
3746 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3748 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3749 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3750 (DOT_SYMBOLS): Likewise.
3751 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3752 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3753 (RS6000_ABI_NAME): Likewise.
3754 (TARGET_CMODEL): Likewise.
3755 (TOC_SECTION_ASM_OP): Likewise.
3756 (SET_CMODEL): New macro.
3757 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3759 2017-08-22 Richard Biener <rguenther@suse.de>
3761 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3762 to allow for free-lang-data replacements similar to verify_type_variant.
3764 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3765 Alan Hayward <alan.hayward@arm.com>
3766 David Sherwood <david.sherwood@arm.com>
3768 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3769 VOIDmode for the LABEL_REF.
3771 2017-08-22 Richard Biener <rguenther@suse.de>
3773 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3775 2017-08-22 Richard Biener <rguenther@suse.de>
3777 * common.opt (feliminate-dwarf2-dups): Ignore.
3778 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3779 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3780 same_die_p_wrap, compute_section_prefix,
3781 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3782 (comdat_symbol_id, comdat_symbol_number): Likewise.
3783 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3785 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3786 (output_die): Mark unreachable path unreachable.
3787 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3788 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3789 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3790 (dwarf2out_early_finish): Likewise.
3792 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
3794 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
3796 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
3799 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
3800 not VAR_P. Filter attribute warnings with OPT_Wattributes.
3801 (avr_attribute_table) <io, io_low, address>: Initialize
3802 .decl_required with true.
3804 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
3806 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
3807 undocumented debugging options.
3808 (-mvsx-scalar-double): Likewise.
3809 (-mallow-df-permute): Likewise.
3810 (-mvectorize-builtins): Likewise.
3811 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3812 (rs6000_builtin_vectorized_function): Likewise.
3813 (rs6000_builtin_md_vectorized_function): Likewise.
3814 (rs6000_opt_vars): Likewise.
3816 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
3819 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
3820 (*btrq_imm): Rename from *btrq.
3821 (*btcq_imm): Rename from *btcq.
3822 (btsc): New code attribute.
3823 (*<btsc><mode>): New insn pattern.
3824 (*btr<mode>): Ditto.
3825 (*<btsc><mode>_mask): New insn_and_split pattern.
3826 (*btr<mode>_mask): Ditto.
3828 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3829 Alan Hayward <alan.hayward@arm.com>
3830 David Sherwood <david.sherwood@arm.com>
3832 * function.c (pad_below): Simplify padding calculation.
3834 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3835 Alan Hayward <alan.hayward@arm.com>
3836 David Sherwood <david.sherwood@arm.com>
3838 * target.def (function_prologue): Remove frame size argument.
3839 (function_epilogue): Likewise.
3840 * doc/tm.texi: Regenerate.
3841 * final.c (final_start_function): Update call to function_prologue.
3842 (final_end_function): Update call to function_epilogue.
3843 (default_function_pro_epilogue): Remove frame size argument.
3844 * output.h (default_function_pro_epilogue): Likewise.
3845 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3846 (arm_output_function_prologue): Likewise.
3847 * config/frv/frv.c (frv_function_prologue): Likewise.
3848 (frv_function_epilogue): Likewise.
3849 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3850 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3851 (ia64_output_function_epilogue): Likewise.
3852 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3853 (m32r_output_function_epilogue): Likewise.
3854 * config/microblaze/microblaze.c (microblaze_function_prologue)
3855 (microblaze_function_epilogue): Likewise.
3856 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3857 (mips_output_function_epilogue): Likewise.
3858 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3859 (mmix_target_asm_function_epilogue): Likewise.
3860 * config/msp430/msp430.c (msp430_start_function): Likewise.
3861 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3862 (nds32_asm_function_epilogue): Likewise.
3863 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3864 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3865 (pa_output_function_epilogue): Likewise.
3866 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3867 (rs6000_output_function_epilogue): Likewise.
3868 * config/rl78/rl78.c (rl78_start_function): Likewise.
3869 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3870 (rs6000_output_function_epilogue): Likewise.
3871 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3872 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3873 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3874 (sparc_asm_function_epilogue): Likewise.
3876 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3878 * tree.h (type_has_mode_precision_p): New function.
3879 * convert.c (convert_to_integer_1): Use it.
3880 * expr.c (expand_expr_real_2): Likewise.
3881 (expand_expr_real_1): Likewise.
3882 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3883 * match.pd: Likewise.
3884 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3885 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3886 * tree-tailcall.c (process_assignment): Likewise.
3887 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3888 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3889 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3890 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3891 (vectorizable_assignment): Likewise.
3892 (vectorizable_shift): Likewise.
3893 (vectorizable_operation): Likewise.
3894 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3896 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3898 * match.pd: Add pow (C, x) simplification.
3900 2017-08-21 Richard Biener <rguenther@suse.de>
3902 PR tree-optimization/81900
3903 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3904 for blocks with abnormal predecessors.
3905 (compute_antic): Do not set visited flag prematurely.
3907 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3910 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3912 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3914 * stor-layout.h (vector_type_mode): Move to...
3915 * tree.h (vector_type_mode): ...here.
3916 * stor-layout.c (vector_type_mode): Move to...
3917 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3919 2017-08-21 Richard Biener <rguenther@suse.de>
3921 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3922 register_external_die hooks.
3923 (debug_false_tree_charstarstar_uhwistar): Declare.
3924 (debug_nothing_tree_charstar_uhwi): Likewise.
3925 * debug.c (do_nothing_debug_hooks): Adjust.
3926 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3927 (debug_nothing_tree_charstar_uhwi): Likewise.
3928 * dbxout.c (dbx_debug_hooks): Adjust.
3929 (xcoff_debug_hooks): Likewise.
3930 * sdbout.c (sdb_debug_hooks): Likewise.
3931 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3932 * dwarf2out.c (macinfo_label_base): New global.
3933 (dwarf2out_register_external_die): New function for the
3934 register_external_die hook.
3935 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3936 (dwarf2_debug_hooks): Use them.
3937 (dwarf2_lineno_debug_hooks): Adjust.
3938 (struct die_struct): Add with_offset flag.
3939 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3940 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3941 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3942 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3943 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3944 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3945 defining section names for the early LTO debug variants.
3946 (reset_indirect_string): New helper.
3947 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3948 (print_dw_val): Add support for offsetted symbol references.
3949 (get_ultimate_context): Split out from is_cxx.
3950 (is_cxx): Use get_ultimate_context.
3951 (is_fortran): Add decl overload.
3952 (compute_comp_unit_symbol): Split out worker from
3953 compute_section_prefix.
3954 (compute_section_prefix): Call compute_comp_unit_symbol and
3955 set comdat_type_p here.
3956 (output_die): Skip DIE symbol output for the LTO added one.
3957 Handle DIE symbol references with offset.
3958 (output_comp_unit): Guard section name mangling properly.
3959 For LTO debug sections emit a symbol at the section beginning
3960 which we use to refer to its DIEs.
3961 (add_abstract_origin_attribute): For DIEs registered via
3962 dwarf2out_register_external_die directly refer to the early
3963 DIE rather than indirectly through the shadow one we created.
3964 Remove obsolete call to dwarf2out_abstract_function for
3965 non-function/block origins.
3966 (gen_array_type_die): When generating early LTO debug do
3967 not emit DW_AT_string_length.
3968 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3969 late when in LTO. As suggested place a gcc_unreachable for
3970 the DECL_ABSTRACT_P case.
3971 (gen_subprogram_die): Avoid another specification DIE
3972 for early built declarations/definitions for the late LTO case.
3973 (gen_variable_die): Add type references for late duplicated VLA dies
3975 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3976 we have the abstract instance already.
3977 (process_scope_var): Adjust decl DIE contexts in LTO which
3978 first puts them in limbo.
3979 (gen_decl_die): Do not generate type DIEs late apart from
3980 types for VLAs or for decls we do not yet have a DIE. Do not
3981 call dwarf2out_abstract_function late.
3982 (dwarf2out_early_global_decl): Make sure to create DIEs
3983 for abstract instances of a decl first.
3984 (dwarf2out_late_global_decl): Adjust comment.
3985 (output_macinfo_op): With multiple macro sections use
3986 macinfo_label_base to distinguish labels.
3987 (output_macinfo): Likewise. Update macinfo_label_base.
3988 Pass in the line info label.
3989 (note_variable_value_in_expr): When generating LTO resolve
3990 all variable values here by generating DIEs as needed.
3991 (init_sections_and_labels): Add early LTO debug flag parameter
3992 and generate different sections and names if set. Add generation
3993 counter for the labels so we can have multiple of them.
3994 (reset_dies): Helper to allow DIEs to be output multiple times.
3995 (dwarf2out_finish): When outputting DIEs to the fat part of an
3996 LTO object first reset DIEs.
3997 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3998 (modified_type_die): Check for decl_ultimate_origin being self
4000 (gen_type_die_with_usage): Likewise.
4001 (gen_typedef_die): Allow decl_ultimate_origin being self.
4002 (set_decl_abstract_flags): Remove.
4003 (set_block_abstract_flags): Likewise.
4004 (dwarf2out_abstract_function): Treat the early generated DIEs
4005 as the abstract copy and only add DW_AT_inline and
4006 DW_AT_artificial here and call set_decl_origin_self.
4007 If the DIE has an abstract origin don't do anything.
4008 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4009 if we have none yet (Go fails to build one, PR78628).
4010 (variably_modified_type_p): Prevent endless recursion for Ada
4011 cyclic pointer types.
4012 * lto-streamer-in.c: Include debug.h.
4013 (dref_queue): New global.
4014 (lto_read_tree_1): Stream in DIE references.
4015 (lto_input_tree): Register DIE references.
4016 (input_function): Stream DECL_DEBUG_ARGS.
4017 * lto-streamer-out.c: Include debug.h.
4018 (lto_write_tree_1): Output DIE references.
4019 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4020 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4021 (output_function): Stream DECL_DEBUG_ARGS.
4022 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4023 Stream DECL_ABSTRACT_ORIGIN.
4024 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4025 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4026 DECL_CONTEXT for file-scope decls.
4027 * lto-streamer.h (struct dref_entry): Declare.
4028 (dref_queue): Likewise.
4029 * cfgexpand.c (pass_expand::execute): Do not call the
4030 outlining_inline_function hook here.
4031 * lto-wrapper.c (debug_obj): New global.
4032 (tool_cleanup): Unlink it if required.
4033 (debug_objcopy): New function.
4034 (run_gcc): Handle early debug sections in the IL files by
4035 extracting them to separate files, partially linkin them and
4036 feeding the result back as result to the linker.
4037 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4038 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4039 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4040 sections into a separate segment.
4041 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4043 (darwin_asm_dwarf_section): Likewise.
4044 (darwin_asm_output_dwarf_offset): Likewise.
4045 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4047 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4048 Alan Hayward <alan.hayward@arm.com>
4049 David Sherwood <david.sherwood@arm.com>
4051 * read-md.h (md_reader::record_potential_iterator_use): Replace
4052 pointer argument with an rtx and an index.
4053 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4054 (apply_mode_iterator): Likewise.
4055 (apply_code_iterator): Likewise.
4056 (apply_int_iterator): Likewise.
4057 (apply_subst_iterator): Likewise.
4058 (record_iterator_use): Likewise.
4059 (record_attribute_use): Likewise.
4060 (md_reader::record_potential_iterator_use): Likewise. Update calls
4061 to record_iterator_use and apply_iterator.
4062 (iterator_use): Replace ptr with x and index.
4063 (attribute_use): Likewise.
4064 (apply_attribute_uses): Update calls to apply_iterator.
4065 (apply_iterators): Likewise. Update initialization of iterator_use.
4066 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4067 and record_potential_iterator_use.
4068 (rtx_reader::read_rtx_operand): Likewise.
4070 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4071 Alan Hayward <alan.hayward@arm.com>
4072 David Sherwood <david.sherwood@arm.com>
4074 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4077 2017-08-21 Richard Biener <rguenther@suse.de>
4080 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4081 at struct end conservatively when comparing common bases.
4083 2017-08-21 Richard Biener <rguenther@suse.de>
4085 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4086 (mem_ref_in_stmt): Remove.
4087 (determine_max_movement): Use ref index to get at the reference.
4088 (invariantness_dom_walker::before_dom_children): Deal with
4089 lim data already initialized.
4090 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4092 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4094 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4095 (TARGET_ISA_ROUND): Ditto.
4096 (TARGET_ROUND): Ditto.
4097 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4098 * config/i386/i386.md: Ditto.
4099 * config/i386/sse.md: Ditto.
4100 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4101 with OPTION_MASK_ISA_SSE4_1.
4103 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4106 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4107 __builtin_ia32_lzcnt_u16.
4109 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4112 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4113 (rs6000_set_current_function): Rewrite function to use it.
4115 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4118 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4119 and TYPE_WARN_IF_NOT_ALIGN.
4120 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4121 (handle_warn_if_not_align): New.
4122 (place_union_field): Call handle_warn_if_not_align.
4123 (place_field): Call handle_warn_if_not_align.
4124 Copy TYPE_WARN_IF_NOT_ALIGN.
4125 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4126 (layout_type): Likewise.
4127 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4129 (tree_decl_common): Add warn_if_not_align.
4130 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4131 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4132 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4133 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4134 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4135 * doc/extend.texi: Document warn_if_not_aligned attribute.
4136 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4138 2017-08-17 Martin Liska <mliska@suse.cz>
4141 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4142 (get_data_dependence): Use it as pointer type.
4143 (distribute_loop): Likewise.
4145 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4147 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4148 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4149 (p8_vmrgow_<mode>_direct): New define_insn.
4150 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4151 handle endianness for vmrgew and vmrgow permute patterns.
4153 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4155 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4156 * config/rs6000/rs6000-cpus.def: Remove comment.
4157 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4158 (POWERPC_MASKS): Likewise.
4159 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4160 use of TARGET_VSX_TIMODE.
4161 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4162 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4163 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4164 (rs6000_option_override_internal): Remove dead code.
4165 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4166 (rs6000_legitimize_reload_address): Likewise.
4167 (rs6000_legitimate_address_p): Likewise.
4168 (rs6000_opt_masks): Delete "vsx-timode".
4169 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4170 from function comment.
4171 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4172 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4173 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4175 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4176 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4177 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4180 (define_peephole2): Likewise.
4182 2017-08-17 Martin Sebor <msebor@redhat.com>
4185 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4186 past the end of an array.
4187 (test_pp_format): Add test cases.
4189 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4191 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4192 missing ECF_NOTHROW flags.
4194 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4197 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4198 operands residing in integer registers.
4199 (*vsx_le_perm_load_<mode>): Likewise.
4200 (*vsx_le_perm_store_<mode>): Likewise.
4201 (define_peephole2): Add peepholes to optimize the above.
4203 2017-08-17 Marek Polacek <polacek@redhat.com>
4206 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4207 to mimic what shorten_compare did. Change the return type to bool.
4208 (fold_cond_expr_with_comparison): Update call to
4209 operand_equal_for_comparison_p.
4210 (fold_ternary_loc): Likewise.
4212 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4214 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4216 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4218 2017-08-17 Richard Biener <rguenther@suse.de>
4220 * tree-ssa-structalias.c (solve_graph): When propagating
4221 to successors update the graphs succ edges and avoid duplicate work.
4223 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4226 * config/i386/i386.c (ix86_option_override_internal): Save target
4227 specific options after ix86_stack_protector_guard_reg was changed.
4229 2017-08-17 Richard Biener <rguenther@suse.de>
4231 PR tree-optimization/81827
4232 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4233 (new_var_info): Initialize it conservatively.
4234 (get_call_vi): Mark register vars.
4235 (new_scalar_tmp_constraint_exp): Likewise.
4236 (handle_rhs_call): Likewise.
4237 (handle_const_call): Likewise.
4238 (create_function_info_for): Likewise.
4239 (solve_constraints): Sort varinfos to separate register from
4240 non-register vars to pack points-to solution bitmaps during
4243 2017-08-17 Marek Polacek <polacek@redhat.com>
4245 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4247 2017-08-17 Richard Biener <rguenther@suse.de>
4249 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4250 to true when overflow is undefined and we saturated the result.
4252 2017-08-17 Alan Modra <amodra@gmail.com>
4255 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4256 Don't use store multiple if only one reg needs saving.
4257 (interesting_frame_related_regno): New function.
4258 (rs6000_frame_related): Don't emit frame info for regs that
4260 (rs6000_emit_epilogue): Likewise.
4262 2017-08-16 Nathan Sidwell <nathan@acm.org>
4264 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4265 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4266 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4267 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4268 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4269 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4270 (verify_type): Adjust for TYPE_BINFO move.
4271 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4272 process TYPE_BINFO directly.
4273 (hash_tree): Likewise.
4274 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4276 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4279 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4281 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4283 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4286 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4287 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4288 Add ix86_binary_operator_ok to insn constraint.
4289 (*iordi_1_bts): Ditto.
4290 (*xordi_1_btc): Ditto.
4291 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4292 Update corresponding peephole2 pattern.
4296 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4298 PR tree-optimization/81832
4299 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4300 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4302 2017-08-16 Marek Polacek <polacek@redhat.com>
4305 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4306 regarding size_zero_node.
4308 2017-08-16 Martin Liska <mliska@suse.cz>
4311 * config.gcc: Respect previously set extra_objs in case
4314 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4316 PR tree-optimization/81835
4317 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4318 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4319 not depend on the phi.
4321 2017-08-16 Alan Modra <amodra@gmail.com>
4323 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4326 2017-08-16 Alan Modra <amodra@gmail.com>
4328 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4329 (save_reg_p): ..into this. Update all callers.
4330 (first_reg_to_save): Simplify.
4332 2017-08-16 Alan Modra <amodra@gmail.com>
4334 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4337 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4341 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4342 constant count if that count is less than 32.
4344 2017-08-15 Nathan Sidwell <nathan@acm.org>
4346 * gcc.c (execute): Emit friendlier message if inferior is killed
4347 by an external cause.
4349 2017-08-15 Richard Biener <rguenther@suse.de>
4351 PR tree-optimization/81790
4352 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4353 CONSTRUCTORs from simplifying and VN.
4355 2017-08-14 Martin Sebor <msebor@redhat.com>
4357 * builtin-attrs.def: Add comments.
4359 2017-08-14 Martin Sebor <msebor@redhat.com>
4362 * doc/extend.texi (attribute nonstring): Document new attribute.
4364 2017-08-14 Martin Sebor <msebor@redhat.com>
4367 * tree-diagnostic.c (default_tree_printer): Handle %G.
4368 * gimple-pretty-print.h (percent_G_format): Declare new function.
4369 * gimple-pretty-print.c (percent_G_format): Define.
4370 * tree-pretty-print.c (percent_K_format): Add argument.
4372 2017-08-14 Martin Sebor <msebor@redhat.com>
4374 PR translation/79998
4375 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4376 Remove a stray space.
4378 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4381 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4382 (*iordi_1_bts): Ditto.
4383 (*xordi_1_btc): Ditto.
4385 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4388 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4389 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4391 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4392 quoted strings, and make more translator-friendly.
4393 (darwin_rs6000_override_options): Likewise.
4394 (rs6000_option_override_internal): Likewise.
4395 (rs6000_return_in_memory): Fix overlong line.
4396 (init_cmulative_args): Use quoted strings, and make more
4397 translator-friendly.
4398 (rs6000_pass_by_reference): Fix overlong line.
4399 (def_builtin): Use quoted strings.
4400 (altivec_expand_predicate_builtin): Use quoted strings, and make
4401 more translator-friendly.
4402 (htm_expand_builtin): Use quoted strings.
4403 (cpu_expand_builtin): Use quoted strings, and make more
4404 translator-friendly.
4405 (altivec_expand_builtin): Likewise.
4406 (paired_expand_predicate_builtin): Likewise.
4407 (rs6000_invalid_builtin): Likewise.
4408 (builtin_function_type): Use quoted strings.
4409 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4410 more translator-friendly.
4411 (rs6000_trampoline_init): Likewise.
4412 (rs6000_handle_altivec_attribute): Likewise.
4413 (rs6000_inner_target_options): Use quoted strings.
4414 (rs6000_disable_incompatible_switches): Likewise.
4415 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4416 strings, and make more translator-friendly.
4417 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4419 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4421 PR tree-optimization/81799
4422 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4423 cond_expr to simple gimple operand.
4425 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4428 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4430 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4434 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4435 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4436 TARGET_GASISR_PROLOGUES.
4437 * config/avr/avr.c (avr_option_override): Same.
4438 (avr_pass_pre_proep::execute): Same.
4440 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4443 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4444 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4446 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4448 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4449 define_insn_and_split. Split to a memory load from 0 in
4450 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4451 using PTR mode iterator.
4452 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4453 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4454 (*add_tp_<mode>): Redefine as define_insn_and_split.
4455 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4456 address space. Merge with *add_tp_x32 using PTR mode iterator.
4457 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4458 Split to an add with a memory load from 0 in
4459 DEFAULT_TLS_SEG_REG address space.
4461 2017-08-12 Andrew Pinski <apinski@cavium.com>
4463 * config/aarch64/aarch64-option-extensions.def (rdma):
4464 Fix feature string to what Linux prints out in /proc/cpuinfo.
4466 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4469 * dwarf2out.c (modified_type_die): For C typedef types that have
4470 an ultimate origin, process the ultimate origin instead of the
4472 (gen_typedef_die): Assert that input DECLs have no ultimate
4474 (gen_type_die_with_usage): For typedef variants that have an
4475 ultimate origin, just call gen_decl_die on the original DECL.
4476 (process_scope_var): Avoid creating DIEs for local typedefs and
4477 concrete static variables.
4479 2017-08-12 Alan Modra <amodra@gmail.com>
4483 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4484 match gnu-user.h startfile.
4485 (ENDFILE_LINUX_SPEC): Similarly.
4487 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4490 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4492 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4494 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4495 * config/aarch64/aarch64.md (mov<mode>): Change.
4496 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4497 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4498 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4500 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4502 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4503 for storage order barriers.
4505 2017-08-11 Martin Liska <mliska@suse.cz>
4508 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4509 variables of void type.
4511 2017-08-11 Martin Liska <mliska@suse.cz>
4513 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4514 TARGET_SUPPORTS_ALIASES.
4515 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4516 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4517 (optimize_weakref): Likewise.
4518 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4519 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4520 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4522 2017-08-11 Martin Liska <mliska@suse.cz>
4525 * config/i386/i386.c (make_resolver_func): Do complete
4526 refactoring of the function.
4528 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4531 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4532 * config/i386/i386.c (ix86_stack_protect_guard): Use
4533 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4534 * doc/invoke.texi (x86 Options): Document
4535 -mstack-protector-guard-symbol= option.
4537 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4539 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4540 * config/i386/i386.c (ix86_split_stack_guard): New function.
4541 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4542 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4543 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4544 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4545 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4546 (split_stack_space_check): Call ix86_split_stack_guard.
4548 2017-08-10 Martin Sebor <msebor@redhat.com>
4550 * print-tree.c (print_node): Print location using the established
4552 Replace spaces with colons.
4553 (debug_raw, debug): Ditto.
4555 2017-08-10 Martin Sebor <msebor@redhat.com>
4558 * pretty-print.c (pp_format): Correct the handling of %s precision.
4560 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4563 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4565 (ix86_finalize_stack_frame_flags): This. Also clear
4566 frame_pointer_needed if -fno-omit-frame-pointer is used without
4568 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4569 with ix86_finalize_stack_frame_flags.
4570 (ix86_expand_epilogue): Likewise.
4571 (ix86_expand_split_stack_prologue): Likewise.
4572 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4574 2017-08-10 Martin Liska <mliska@suse.cz>
4577 * c-attribs.c (handle_target_attribute):
4578 Report warning for an empty string argument of target attribute.
4580 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4583 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4585 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4586 or DECL_NONLOCAL labels.
4587 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4588 or DECL_NONLOCAL labels here.
4590 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4592 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4593 to indicate when early gimple folding has been disabled.
4594 (rs6000_gimple_fold_builtin): Add debug content.
4595 (rs6000_invalid_builtin): Fix whitespace.
4596 (rs6000_expand_builtin): Fix whitespace.
4597 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4599 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4602 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4603 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4605 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4607 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4608 (qdf24xx): Likewise.
4609 * config/aarch64/aarch64.md: Include falkor.md.
4610 * config/aarch64/falkor.md: New.
4612 2017-08-09 Marek Polacek <polacek@redhat.com>
4615 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4616 * diagnostic.c (emit_diagnostic): Add a comment.
4617 (emit_diagnostic_valist): New function.
4619 2017-08-09 Marek Polacek <polacek@redhat.com>
4622 * input.c (make_location): New overload.
4623 * input.h (make_location): Declare.
4625 2017-08-08 Alan Modra <amodra@gmail.com>
4626 H.J. Lu <hongjiu.lu@intel.com>
4629 * gcc.c (NO_PIE_SPEC): Delete.
4630 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4632 (LINK_PIE_SPEC): ..to here.
4633 (LINK_COMMAND_SPEC): Support -no-pie.
4634 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4635 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4637 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4638 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4639 (ENDFILE_CRTEND_SPEC): Similarly.
4641 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4644 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4645 (mstack-protector-guard-offset=): Ditto.
4646 * config/i386/i386.c (ix86_option_override): Handle
4647 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4649 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4650 ix86_stack_protect_guard_offset variables.
4651 (TARGET_STACK_PROTECT_GUARD): Always define.
4652 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4653 and -mstack-protector-guard-offset= options.
4655 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4657 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4658 boundary case for the last candidate.
4660 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4662 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4663 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4665 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4668 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4669 * config/aarch64/aarch64-builtins.c
4670 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4671 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4672 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4674 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4675 Andrew Pinski <pinskia@gmail.com>
4678 * internal-fn.def (XORSIGN): New.
4679 * optabs.def (xorsign_optab): New.
4680 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4681 (convert_expand_mult_copysign): New.
4682 (pass_optimize_widening_mul::execute): Call
4683 convert_expand_mult_copysign.
4685 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4687 PR tree-optimization/81354
4688 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4689 Insert on edges rather than explicitly creating landing pads.
4690 (analyze_candidates_and_replace): Commit edge inserts.
4692 2017-08-08 Richard Biener <rguenther@suse.de>
4695 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4696 (expand_simple_operations): Also look through ADDR_EXPRs with
4697 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4699 2017-08-08 Richard Biener <rguenther@suse.de>
4701 PR tree-optimization/81723
4702 * tree-vect-slp.c (struct bst_traits): New hash traits.
4703 (bst_fail): New global.
4704 (vect_build_slp_tree_2): New worker, split out from ...
4705 (vect_build_slp_tree): ... this now wrapping it with using
4706 bst_fail set to cache SLP tree build fails. Properly handle
4708 (vect_analyze_slp_instance): Allocate and free bst_fail.
4710 2017-08-08 Martin Liska <mliska@suse.cz>
4713 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4714 LABEL_DECLs that can be from a different function.
4716 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4718 PR tree-optimization/81744
4719 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4720 loop's number of iterations.
4722 2017-08-08 Martin Liska <mliska@suse.cz>
4724 * asan.c: Include header files.
4725 * attribs.c (build_decl_attribute_variant): New function moved
4727 (build_type_attribute_qual_variant): Likewise.
4728 (cmp_attrib_identifiers): Likewise.
4729 (simple_cst_list_equal): Likewise.
4730 (omp_declare_simd_clauses_equal): Likewise.
4731 (attribute_value_equal): Likewise.
4732 (comp_type_attributes): Likewise.
4733 (build_type_attribute_variant): Likewise.
4734 (lookup_ident_attribute): Likewise.
4735 (remove_attribute): Likewise.
4736 (merge_attributes): Likewise.
4737 (merge_type_attributes): Likewise.
4738 (merge_decl_attributes): Likewise.
4739 (merge_dllimport_decl_attributes): Likewise.
4740 (handle_dll_attribute): Likewise.
4741 (attribute_list_equal): Likewise.
4742 (attribute_list_contained): Likewise.
4743 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4744 (lookup_attribute_by_prefix): Likewise.
4745 * bb-reorder.c: Include header files.
4746 * builtins.c: Likewise.
4747 * calls.c: Likewise.
4748 * cfgexpand.c: Likewise.
4749 * cgraph.c: Likewise.
4750 * cgraphunit.c: Likewise.
4751 * convert.c: Likewise.
4752 * dwarf2out.c: Likewise.
4753 * final.c: Likewise.
4754 * fold-const.c: Likewise.
4755 * function.c: Likewise.
4756 * gimple-expr.c: Likewise.
4757 * gimple-fold.c: Likewise.
4758 * gimple-pretty-print.c: Likewise.
4759 * gimple.c: Likewise.
4760 * gimplify.c: Likewise.
4761 * hsa-common.c: Likewise.
4762 * hsa-gen.c: Likewise.
4763 * internal-fn.c: Likewise.
4764 * ipa-chkp.c: Likewise.
4765 * ipa-cp.c: Likewise.
4766 * ipa-devirt.c: Likewise.
4767 * ipa-fnsummary.c: Likewise.
4768 * ipa-inline.c: Likewise.
4769 * ipa-visibility.c: Likewise.
4771 * lto-cgraph.c: Likewise.
4772 * omp-expand.c: Likewise.
4773 * omp-general.c: Likewise.
4774 * omp-low.c: Likewise.
4775 * omp-offload.c: Likewise.
4776 * omp-simd-clone.c: Likewise.
4777 * opts-global.c: Likewise.
4778 * passes.c: Likewise.
4779 * predict.c: Likewise.
4780 * sancov.c: Likewise.
4781 * sanopt.c: Likewise.
4782 * symtab.c: Likewise.
4783 * toplev.c: Likewise.
4784 * trans-mem.c: Likewise.
4785 * tree-chkp.c: Likewise.
4786 * tree-eh.c: Likewise.
4787 * tree-into-ssa.c: Likewise.
4788 * tree-object-size.c: Likewise.
4789 * tree-parloops.c: Likewise.
4790 * tree-profile.c: Likewise.
4791 * tree-ssa-ccp.c: Likewise.
4792 * tree-ssa-live.c: Likewise.
4793 * tree-ssa-loop.c: Likewise.
4794 * tree-ssa-sccvn.c: Likewise.
4795 * tree-ssa-structalias.c: Likewise.
4796 * tree-ssa.c: Likewise.
4797 * tree-streamer-in.c: Likewise.
4798 * tree-vectorizer.c: Likewise.
4799 * tree-vrp.c: Likewise.
4801 * ubsan.c: Likewise.
4802 * varasm.c: Likewise.
4803 * varpool.c: Likewise.
4804 * tree.c: Remove functions moved to attribs.[ch].
4806 * config/aarch64/aarch64.c: Add attrs.h header file.
4807 * config/alpha/alpha.c: Likewise.
4808 * config/arc/arc.c: Likewise.
4809 * config/arm/arm.c: Likewise.
4810 * config/avr/avr.c: Likewise.
4811 * config/bfin/bfin.c: Likewise.
4812 * config/c6x/c6x.c: Likewise.
4813 * config/cr16/cr16.c: Likewise.
4814 * config/cris/cris.c: Likewise.
4815 * config/darwin.c: Likewise.
4816 * config/epiphany/epiphany.c: Likewise.
4817 * config/fr30/fr30.c: Likewise.
4818 * config/frv/frv.c: Likewise.
4819 * config/ft32/ft32.c: Likewise.
4820 * config/h8300/h8300.c: Likewise.
4821 * config/i386/winnt.c: Likewise.
4822 * config/ia64/ia64.c: Likewise.
4823 * config/iq2000/iq2000.c: Likewise.
4824 * config/lm32/lm32.c: Likewise.
4825 * config/m32c/m32c.c: Likewise.
4826 * config/m32r/m32r.c: Likewise.
4827 * config/m68k/m68k.c: Likewise.
4828 * config/mcore/mcore.c: Likewise.
4829 * config/microblaze/microblaze.c: Likewise.
4830 * config/mips/mips.c: Likewise.
4831 * config/mmix/mmix.c: Likewise.
4832 * config/mn10300/mn10300.c: Likewise.
4833 * config/moxie/moxie.c: Likewise.
4834 * config/msp430/msp430.c: Likewise.
4835 * config/nds32/nds32-isr.c: Likewise.
4836 * config/nds32/nds32.c: Likewise.
4837 * config/nios2/nios2.c: Likewise.
4838 * config/nvptx/nvptx.c: Likewise.
4839 * config/pa/pa.c: Likewise.
4840 * config/pdp11/pdp11.c: Likewise.
4841 * config/powerpcspe/powerpcspe.c: Likewise.
4842 * config/riscv/riscv.c: Likewise.
4843 * config/rl78/rl78.c: Likewise.
4844 * config/rx/rx.c: Likewise.
4845 * config/s390/s390.c: Likewise.
4846 * config/sh/sh.c: Likewise.
4847 * config/sol2.c: Likewise.
4848 * config/sparc/sparc.c: Likewise.
4849 * config/spu/spu.c: Likewise.
4850 * config/stormy16/stormy16.c: Likewise.
4851 * config/tilegx/tilegx.c: Likewise.
4852 * config/tilepro/tilepro.c: Likewise.
4853 * config/v850/v850.c: Likewise.
4854 * config/vax/vax.c: Likewise.
4855 * config/visium/visium.c: Likewise.
4856 * config/xtensa/xtensa.c: Likewise.
4858 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4861 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4862 constraints since the -mupper-regs-* switches have been
4864 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4865 into a vector from a double word element that was extracted from
4866 another vector, and eliminate extra XXPERMDI instructions.
4867 (vsx_concat_<mode>_2): Likewise.
4868 (vsx_concat_<mode>_3): Likewise.
4869 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4870 concat to allow optimizing inserts from previous extracts.
4872 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4874 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4875 memory reference to a SSP offset in TLS address space.
4876 (ix86_print_operand) <case '@'>: Remove.
4877 (ix86_print_operand_punct_valid_p): Remove '@' code.
4878 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4880 (stack_tls_protect_set_<mode>): Remove.
4881 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4882 (stack_tls_protect_test_<mode>): Remove.
4883 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4885 2017-08-07 Olivier Hainque <hainque@adacore.com>
4888 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4890 2017-08-07 Douglas Rupp <rupp@adacore.com>
4892 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4893 variable was referenced as multidir in command.
4895 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4898 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4901 2017-08-07 Martin Liska <mliska@suse.cz>
4903 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4904 * config/rl78/rl78.c: Add include of attribs.h.
4905 * config/sh/sh.c: Likewise.
4906 * config/v850/v850.c: Likewise.
4908 2017-08-07 Tom de Vries <tom@codesourcery.com>
4911 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4913 2017-08-07 Martin Liska <mliska@suse.cz>
4915 * config/mips/mips.c: Include attribs.h.
4917 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4920 * doc/invoke.texi: Document change in behvaior for -Ofast for
4923 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4925 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4927 (aarch64_pop_regs): Likewise.
4928 (aarch64_gen_load_pair): Likewise.
4929 (aarch64_save_callee_saves): Likewise.
4930 (aarch64_restore_callee_saves): Likewise.
4932 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4934 * config/i386/i386.c: Revert the last change.
4936 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4939 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4941 (ix86_finalize_stack_frame_flags): This. Also clear
4942 frame_pointer_needed if -fno-omit-frame-pointer is used without
4944 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4945 with ix86_finalize_stack_frame_flags.
4946 (ix86_expand_epilogue): Likewise.
4947 (ix86_expand_split_stack_prologue): Likewise.
4949 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4952 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4953 to P_AES for Westmere.
4955 2017-08-07 Jonathan Yong <10walls@gmail.com>
4957 * config/i386/mingw.opt (fset-stack-executable): Removed.
4958 * config/i386/cygming.opt (fset-stack-executable): Moved
4960 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4962 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4964 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4966 2017-08-07 Marek Polacek <polacek@redhat.com>
4969 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4971 2017-08-07 Martin Liska <mliska@suse.cz>
4973 * attribs.h (canonicalize_attr_name): New function.
4974 (cmp_attribs): Move from c-format.c and adjusted.
4975 (is_attribute_p): Moved from tree.h.
4976 * tree-inline.c: Add new includes.
4977 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4978 (private_is_attribute_p): Remove.
4979 (private_lookup_attribute): Likewise.
4980 (private_lookup_attribute_by_prefix): Simplify.
4981 (remove_attribute): Use is_attribute_p.
4982 * tree.h: Remove removed declarations.
4984 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4987 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4988 instead of computing it in the function. Formatting fix.
4989 (expand_case): Don't rely on default_edge being the first edge,
4990 clear it if removing it, pass default_edge to
4991 emit_case_dispatch_table.
4992 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4995 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4997 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4998 insn in the function, emit NOP after the insn.
5000 2017-08-06 Tom de Vries <tom@codesourcery.com>
5002 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5005 2017-08-06 Tom de Vries <tom@codesourcery.com>
5007 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5010 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5012 PR tree-optimization/57371
5013 * match.pd: New pattern.
5015 2017-08-04 Marek Polacek <polacek@redhat.com>
5018 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5019 perform the computation in offset_int.
5021 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5023 PR tree-optimization/81136
5024 * tree-vectorizer.h: Include tree-hash-traits.h.
5025 (vec_base_alignments): New typedef.
5026 (vec_info): Add a base_alignments field.
5027 (vect_record_base_alignments): Declare.
5028 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5030 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5031 (create_data_ref): Add an is_conditional_in_stmt argument.
5032 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5033 the is_conditional_in_stmt field.
5034 (data_ref_loc): Add an is_conditional_in_stmt field.
5035 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5036 (find_data_references_in_stmt): Update call to create_data_ref.
5037 (graphite_find_data_references_in_stmt): Likewise.
5038 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5039 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5040 (vect_record_base_alignment): New function.
5041 (vect_record_base_alignments): Likewise.
5042 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5043 for nested statements even if we fail to compute a misalignment.
5044 Use pooled base alignments for unconditional references.
5045 (vect_find_same_alignment_drs): Compare base addresses instead
5047 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5048 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5050 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5052 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5053 Add an explicit name for the enum. Use auto_vec for slp_instances
5055 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5057 (_bb_vec_info): Add a constructor and destructor.
5058 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5059 (destroy_loop_vec_info): Delete.
5060 (vect_destroy_datarefs): Likewise.
5061 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5062 (vec_info::vec_info): New function.
5063 (vec_info::~vec_info): Likewise.
5064 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5065 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5066 destroy_loop_vec_info.
5067 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5068 (_loop_vec_info::_loop_vec_info): ...this.
5069 (destroy_loop_vec_info): Replace with...
5070 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5071 the stmt_vec_infos. Leave handling of vec_info information to its
5072 destructor. Remove explicit vector releases.
5073 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5074 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5075 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5076 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5077 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5078 (destroy_bb_vec_info): Replace with...
5079 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5080 information to its destructor.
5081 (vect_slp_analyze_bb_1): Use new and delete instead of
5082 new_bb_vec_info and destroy_bb_vec_info.
5083 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5086 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5088 * tree-data-ref.h (subscript): Add access_fn field.
5089 (data_dependence_relation): Add could_be_independent_p.
5090 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5091 (same_access_functions): Move to tree-data-ref.c.
5092 * tree-data-ref.c (ref_contains_union_access_p): New function.
5093 (access_fn_component_p): Likewise.
5094 (access_fn_components_comparable_p): Likewise.
5095 (dr_analyze_indices): Add a reference to access_fn_component_p.
5096 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5098 (constant_access_functions): Likewise.
5099 (add_other_self_distances): Likewise.
5100 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5101 (initialize_data_dependence_relation): Use XCNEW and remove
5102 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5103 of access functions that have the same type. Allow the
5104 subsequence to end with different bases in some circumstances.
5105 Record the chosen access functions in SUB_ACCESS_FN.
5106 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5107 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5108 (subscript_dependence_tester_1): Likewise dra and drb.
5109 (build_classic_dist_vector): Update calls accordingly.
5110 (subscript_dependence_tester): Likewise.
5111 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5112 DDR_COULD_BE_INDEPENDENT_P.
5113 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5114 comp_alias_ddrs instead of may_alias_ddrs.
5115 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5117 (vect_analyze_data_ref_dependence): Use it if
5118 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5119 distance vectors if that fails.
5120 (dependence_distance_ge_vf): New function.
5121 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5122 LOOP_VINFO_MAY_ALIAS_DDRS.
5124 2017-08-04 Richard Biener <rguenther@suse.de>
5127 * fold-const.c (fold_binary_loc): Properly restrict
5128 minus_var0 && minus_var1 case when associating undefined overflow
5131 2017-08-04 Tom de Vries <tom@codesourcery.com>
5133 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5135 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5137 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5138 Don't start diagnostic messages with a capital letter.
5139 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5141 (rs6000_invalid_builtin): Likewise.
5142 (rs6000_trampoline_init): Likewise.
5144 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5147 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5148 after setting changeable df flags.
5150 2017-08-03 Richard Biener <rguenther@suse.de>
5152 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5153 up if the use is in USE - X.
5155 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5157 * toplev.c (dumpfile.h): New include.
5158 (internal_error_reentered): New static function. Use it...
5159 (internal_error_function): ...here to handle reentered internal_error.
5161 2017-08-03 Richard Biener <rguenther@suse.de>
5164 * fold-const.c (split_tree): Add minus_var and minus_con
5165 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5166 here but always use minus_*.
5167 (associate_trees): Assert we never associate with MINUS_EXPR
5168 and NULL first operand. Do not recurse for PLUS_EXPR operands
5169 when associating as MINUS_EXPR either.
5170 (fold_binary_loc): Track minus_var and minus_con.
5172 2017-08-03 Tom de Vries <tom@codesourcery.com>
5175 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5176 ACCEL_COMPILER, apply finish_options on
5177 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5179 2017-08-03 Tom de Vries <tom@codesourcery.com>
5182 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5183 function_entry_patch_area_size > 0.
5185 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5188 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5189 instead of 10??LU, perform unit multiplication in wide_int,
5190 don't change alloc_object_size_limit if the limit is larger
5193 PR tree-optimization/81655
5194 PR tree-optimization/81588
5195 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5196 the case when ranges[i].low and high are 1 for unsigned type with
5200 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5201 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5203 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5205 * tree-vrp.h: Add include guard.
5207 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5210 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5211 (ud2): New insn pattern.
5212 * config/i386/i386.c (ix86_expand_epilogue):
5213 For naked functions, generate ud2 instead of trap insn.
5215 2017-08-02 Marek Polacek <polacek@redhat.com>
5218 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5220 2017-08-02 Tom de Vries <tom@codesourcery.com>
5221 Cesar Philippidis <cesar@codesourcery.com>
5223 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5224 Add missing edge probabilities.
5226 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5228 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5231 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5234 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5235 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5236 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5237 prologue_seq sequences - if any.
5239 2017-08-02 Richard Biener <rguenther@suse.de>
5241 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5242 via vectors if supported, integer extracts via punning if supported
5243 or otherwise vector extracts.
5245 2017-08-02 Richard Biener <rguenther@suse.de>
5247 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5249 (bitmap_insert_into_set): ... this.
5251 2017-08-02 Richard Biener <rguenther@suse.de>
5253 PR tree-optimization/81633
5255 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5257 PR tree-optimization/71752
5258 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5260 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5262 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5263 (machine_function::call_ms2sysv_pad_out): Remove field.
5264 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5265 (ix86_compute_frame_layout): Likewise.
5267 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5270 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5271 attribute with interrupt attribute.
5273 2017-08-01 Andrew Pinski <apinski@cavium.com>
5275 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5276 BIT_INSERT_EXPR's operand 1
5277 to see if the types precision matches.
5279 2017-08-01 Martin Liska <mliska@suse.cz>
5282 * builtins.c (expand_builtin_memcpy_args): Remove.
5283 (expand_builtin_memcpy): Call newly added function
5284 expand_builtin_memory_copy_args.
5285 (expand_builtin_memcpy_with_bounds): Likewise.
5286 (expand_builtin_mempcpy): Remove last argument.
5287 (expand_builtin_mempcpy_with_bounds): Likewise.
5288 (expand_builtin_memory_copy_args): New function created from
5289 expand_builtin_mempcpy_args with small modifications.
5290 (expand_builtin_mempcpy_args): Remove.
5291 (expand_builtin_stpcpy): Remove unused argument.
5292 (expand_builtin): Likewise.
5293 (expand_builtin_with_bounds): Likewise.
5295 2017-08-01 Martin Liska <mliska@suse.cz>
5298 Make mempcpy more optimal (PR middle-end/70140).
5300 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5303 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5304 __builtin_vec_cmpne verify both arguments are compatible vectors
5305 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5306 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5307 move computation of aligned to after checking the argument types.
5311 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5312 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5315 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5316 Doug Rupp <rupp@adacore.com>
5317 Olivier Hainque <hainque@adacore.com>
5319 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5320 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5322 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5323 for TARGET_OS_CPP_BUILTIN.
5324 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5325 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5327 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5328 passing required abi options to the assembler for EABI configurations.
5329 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5330 of .text.hot and .text.unlikely sections for kernel modules when
5331 using ARM style exceptions.
5332 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5333 options. Add EXTRA_CC1_SPEC.
5334 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5336 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5338 (ARM_TARGET2_DWARF_FORMAT): Define.
5339 * config/arm/t-vxworks: Adjust multilib control to removal of the
5340 Diab command line options.
5342 2017-08-01 Martin Liska <mliska@suse.cz>
5344 PR gcov-profile/81561
5345 * gcov.c (unblock): Make unblocking safe as we need to preserve
5346 index correspondence of blocks and block_lists.
5348 2017-08-01 Richard Biener <rguenther@suse.de>
5350 PR tree-optimization/81181
5351 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5352 (compute_antic): ... end of iteration here.
5354 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5356 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5357 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5358 (ftree-slp-vectorize): Likewise.
5359 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5360 can no longer be set independent of flag_tree_loop_vectorize.
5361 * omp-general.c (emp_max_vf): Likewise.
5362 * opts.c (enable_fdo_optimizations): Remove references to
5363 flag_tree_vectorize, these are now implicit.
5364 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5365 and leave it for the options machinery.
5367 2017-08-01 Martin Liska <mliska@suse.cz>
5370 * builtins.c (expand_builtin_memcpy_args): Remove.
5371 (expand_builtin_memcpy): Call newly added function
5372 expand_builtin_memory_copy_args.
5373 (expand_builtin_memcpy_with_bounds): Likewise.
5374 (expand_builtin_mempcpy): Remove last argument.
5375 (expand_builtin_mempcpy_with_bounds): Likewise.
5376 (expand_builtin_memory_copy_args): New function created from
5377 expand_builtin_mempcpy_args with small modifications.
5378 (expand_builtin_mempcpy_args): Remove.
5379 (expand_builtin_stpcpy): Remove unused argument.
5380 (expand_builtin): Likewise.
5381 (expand_builtin_with_bounds): Likewise.
5383 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5386 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5387 print "ds:" only for immediates in generic address space.
5389 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5392 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5393 (ix86_function_ok_for_sibcall): Return false for naked functions.
5395 2017-08-01 Richard Biener <rguenther@suse.de>
5397 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5398 (compute_antic): Seed worklist with exit block predecessors.
5399 * cfganal.c (dfs_find_deadend): For a cycle return the source
5400 of the edge closing it.
5402 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5404 * config/aarch64/aarch64.c
5405 (aarch64_can_const_movi_rtx_p): Move 0 check.
5407 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5409 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5410 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5412 * match.pd: Ditto in address comparison pattern.
5414 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5416 PR tree-optimization/81627
5417 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5418 closed ssa form for store-store chain.
5420 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5422 PR tree-optimization/81620
5423 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5424 for store-store chain.
5426 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5428 PR tree-optimization/81588
5429 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5430 ranges[i].in_p, invert comparison code ccode. For >/>=,
5431 swap rhs1 and rhs2 and comparison code unconditionally,
5432 for </<= don't do that. Don't swap rhs1/rhs2 again if
5433 ranges[i].in_p, instead invert comparison code ccode if
5434 opcode or oe->rank is BIT_IOR_EXPR.
5437 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5438 a direct optab to conversion optab.
5439 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5440 with GET_MODE_INNER as last argument instead of optab_handler.
5441 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5442 vector extraction if possible and optab is available.
5443 * expr.c (store_constructor): Use convert_optab_handler instead
5444 of optab_handler. Use vector initialization from smaller
5445 vectors if possible and optab is available.
5446 * tree-vect-stmts.c (vectorizable_load): Likewise.
5447 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5449 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5450 of vec_init from half-sized vectors with the same element mode.
5451 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5452 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5453 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5454 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5455 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5456 after mode in gen_vec_extract* calls.
5457 (vec_extract<mode>): Renamed to ...
5458 (vec_extract<mode><ssescalarmodelower>): ... this.
5459 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5460 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5461 element mode after mode in gen_vec_init* calls.
5462 (VEC_INIT_HALF_MODE): New mode iterator.
5463 (vec_init<mode>): Renamed to ...
5464 (vec_init<mode><ssescalarmodelower>): ... this.
5465 (vec_init<mode><ssehalfvecmodelower>): New expander.
5466 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5467 (vec_extractv2sfsf): ... this.
5468 (vec_initv2sf): Renamed to ...
5469 (vec_initv2sfsf): ... this.
5470 (vec_extractv2si): Renamed to ...
5471 (vec_extractv2sisi): ... this.
5472 (vec_initv2si): Renamed to ...
5473 (vec_initv2sisi): ... this.
5474 (vec_extractv4hi): Renamed to ...
5475 (vec_extractv4hihi): ... this.
5476 (vec_initv4hi): Renamed to ...
5477 (vec_initv4hihi): ... this.
5478 (vec_extractv8qi): Renamed to ...
5479 (vec_extractv8qiqi): ... this.
5480 (vec_initv8qi): Renamed to ...
5481 (vec_initv8qiqi): ... this.
5482 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5483 (vec_init<mode>): Renamed to ...
5484 (vec_init<mode><VEC_base_l>): ... this.
5485 (vec_extract<mode>): Renamed to ...
5486 (vec_extract<mode><VEC_base_l>): ... this.
5487 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5488 (vec_initv2sfsf): ... this.
5489 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5490 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5491 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5492 element mode after mode in gen_vec_init* calls.
5493 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5494 (vec_init<mode><Vel>): ... this.
5495 (vec_extract<mode>): Renamed to ...
5496 (vec_extract<mode><Vel>): ... this.
5497 * config/aarch64/iterators.md (Vel): New mode attribute.
5498 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5499 Add element mode after mode in gen_vec_extract* calls.
5500 * config/s390/vector.md (non_vec_l): New mode attribute.
5501 (vec_extract<mode>): Renamed to ...
5502 (vec_extract<mode><non_vec_l>): ... this.
5503 (vec_init<mode>): Renamed to ...
5504 (vec_init<mode><non_vec_l>): ... this.
5505 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5506 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5508 * config/arm/iterators.md (V_elem_l): New mode attribute.
5509 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5510 (vec_extract<mode><V_elem_l>): ... this.
5511 (vec_extractv2di): Renamed to ...
5512 (vec_extractv2didi): ... this.
5513 (vec_init<mode>): Renamed to ...
5514 (vec_init<mode><V_elem_l>): ... this.
5515 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5516 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5517 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5518 Add element mode after gen_vec_extract* calls.
5519 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5520 (vec_init<mode><unitmode>): ... this.
5521 (vec_extract<mode>): Renamed to ...
5522 (vec_extract<mode><unitmode>): ... this.
5523 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5524 (vec_init<mode><unitmode>): ... this.
5525 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5526 (vec_initv2sfsf): ... this.
5527 (vec_extractv2sf): Renamed to ...
5528 (vec_extractv2sfsf): ... this.
5529 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5530 Add element mode after gen_vec_extract* calls.
5531 * config/mips/mips.md (unitmode): New mode iterator.
5532 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5533 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5534 * config/spu/spu.md (inner_l): New mode attribute.
5535 (vec_init<mode>): Renamed to ...
5536 (vec_init<mode><inner_l>): ... this.
5537 (vec_extract<mode>): Renamed to ...
5538 (vec_extract<mode><inner_l>): ... this.
5539 * config/sparc/sparc.md (veltmode): New mode iterator.
5540 (vec_init<VMALL:mode>): Renamed to ...
5541 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5542 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5543 (vec_initv2sisi): ... this.
5544 (vec_initv2sf): Renamed to ...
5545 (vec_initv2sfsf): ... this.
5546 (vec_extractv2sf): Renamed to ...
5547 (vec_extractv2sfsf): ... this.
5548 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5549 (vec_init<mode>): Renamed to ...
5550 (vec_init<mode><VEC_base_l>): ... this.
5551 (vec_extract<mode>): Renamed to ...
5552 (vec_extract<mode><VEC_base_l>): ... this.
5553 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5554 (vec_initv2sfsf): ... this.
5555 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5556 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5557 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5558 gen_vec_init* calls.
5560 2017-08-01 Richard Biener <rguenther@suse.de>
5562 PR tree-optimization/81297
5563 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5564 TREE_OVERFLOW from INTEGER_CSTs.
5566 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5568 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5570 2017-07-31 Carl Love <cel@us.ibm.com>
5572 * config/rs6000/rs6000-c: Add support for built-in functions
5573 vector signed char vec_xl_be (signed long long, signed char *);
5574 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5575 vector signed int vec_xl_be (signed long long, signed int *);
5576 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5577 vector signed long long vec_xl_be (signed long long, signed long long *);
5578 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5579 vector signed short vec_xl_be (signed long long, signed short *);
5580 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5581 vector double vec_xl_be (signed long long, double *);
5582 vector float vec_xl_be (signed long long, float *);
5583 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5584 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5585 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5587 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5588 (altivec_expand_builtin): Add switch statement to call
5589 altivec_expand_xl_be for each builtin.
5590 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5591 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5592 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5593 __builtin_vsx_le_be_v16qi.
5594 * doc/extend.texi: Update the built-in documentation file for the
5595 new built-in functions.
5597 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5600 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5602 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5604 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5606 * config.gcc: Add z14.
5607 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5608 CPU model numbers for z13s and z14.
5609 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5611 * config/s390/s390-opts.h (enum processor_type): Rename
5612 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5613 * config/s390/s390.c (processor_table): Add field for CPU name to
5614 be passed to Binutils.
5615 (s390_asm_output_machine_for_arch): Use the new field in
5616 processor_table for Binutils.
5617 (s390_expand_builtin): Replace arch12 with z14.
5618 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5619 (s390_get_sched_attrmask): Likewise.
5620 (s390_get_unit_mask): Likewise.
5621 * config/s390/s390.opt: Add z14 to processor_type enum.
5623 2017-07-31 Martin Jambor <mjambor@suse.cz>
5626 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5627 regardless of optimization level.
5629 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5630 Martin Liska <mliska@suse.cz>
5632 * predict.def: Remove old comment and adjust probability.
5633 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5636 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5639 * config/i386/i386.c (ix86_function_naked): New function.
5640 (ix86_can_use_return_insn_p): Return false for naked functions.
5641 (ix86_expand_prologue): Skip prologue for naked functions.
5642 (ix86_expand_epilogue): Skip epilogue for naked functions
5643 and emit trap instruction.
5644 (ix86_warn_func_return): New function.
5645 (ix86_attribute_table): Add "naked" attribute specification.
5646 (TARGET_WARN_FUNC_RETURN): Define.
5647 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5649 2017-07-31 Martin Liska <mliska@suse.cz>
5651 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5652 (dump_gimple_bb_header): Always dump BB info.
5653 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5655 2017-07-31 Martin Liska <mliska@suse.cz>
5658 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5659 also with current_function_decl non-null equality.
5661 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5664 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5665 change type to the element type, instead add eltype variable and
5666 use it where we are interested in the element type.
5668 PR tree-optimization/81603
5669 * ipa-polymorphic-call.c
5670 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5671 offset arithmetic in offset_int, bail out if the resulting bit offset
5672 doesn't fit into shwi.
5674 2017-07-31 Martin Liska <mliska@suse.cz>
5676 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5677 (gimplify_save_expr): Fix comment.
5679 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5682 * config/i386/i386.c (ix86_function_arg): Update arguments for
5684 (ix86_compute_frame_layout): Set the initial stack offset to
5685 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5686 INCOMING_FRAME_SP_OFFSET.
5687 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5688 stack before exception handler returns.
5689 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5690 the 'ERROR_CODE' for exception handler.
5692 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5694 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5695 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5696 (ASM_OUTPUT_REG_POP): Ditto.
5697 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5698 instead of asm_fprintf to output pure string.
5700 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5702 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5703 to imported_module_or_decl hook.
5704 (debug_nothing_tree_tree_tree_bool): Remove.
5705 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5706 * debug.c (do_nothing_debug_hooks): Use
5707 debug_nothing_tree_tree_tree_bool_bool instead of
5708 debug_nothing_tree_tree_tree_bool.
5709 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5710 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5711 * sdbout.c (sdb_debug_hooks): Likewise.
5712 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5713 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5715 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5716 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5717 attribute, don't add anything.
5719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5721 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5722 (fold_build2_stat_loc): Likewise.
5723 (fold_build3_stat_loc): Likewise.
5724 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5725 (fold_build1_loc): Remove macro.
5726 (fold_build2_loc): Likewise.
5727 (fold_build3_loc): Likewise.
5729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5731 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5732 (gimple_build_debug_bind_source_stat): Likewise.
5733 * gimple.h (gimple_build_debug_bind): Remove macro.
5734 (gimple_build_debug_bind_source): Likewise.
5736 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5738 * bitmap.c (bitmap_alloc): Adjust.
5739 (bitmap_gc_alloc): Likewise.
5740 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5742 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5744 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5745 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5746 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5747 (bitmap_gc_alloc_stat): Likewise.
5748 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5750 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5752 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5753 * rtl.h (shallow_copy_rtx): Remove macro.
5755 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5757 * emit-rtl.c (gen_raw_REG): Adjust.
5758 * gengenrtl.c (gendef): Likewise.
5759 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5760 * rtl.h (rtx_alloc): Remove macro.
5762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5764 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5765 (build_tree_list_stat): Likewise.
5766 * tree.h (build_tree_list): Remove macro.
5767 (build_tree_list_vec): Likewise.
5769 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5771 * tree.c (make_vector_stat): Remove _stat from name.
5772 (build_vector_stat): Likewise.
5773 * tree.h (make_vector_stat): Remove macro.
5774 (build_vector_stat): Likewise.
5776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5778 * tree.h (build_var_debug_value): Remove prototype.
5780 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5782 * tree.c (tree_cons_stat): Remove _stat from name.
5783 * tree.h (tree_cons): Remove macro.
5785 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5787 * tree.c (build_vl_exp_stat): Remove _stat from name.
5788 * tree.h (build_vl_exp): Remove macro.
5790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5792 * tree.c (build_decl_stat): Remove _stat from name.
5793 * tree.h (build_decl): Remove macro.
5795 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5797 * gimple.c (gimple_build_with_ops_stat): Adjust.
5798 (gimple_alloc_stat): Remove _stat from name.
5799 * gimple.h (gimple_alloc): Remove macro.
5801 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5803 * tree.c (make_tree_vec_stat): Remove _stat from name.
5804 (grow_tree_vec_stat): Likewise.
5805 * tree.h (make_tree_vec_stat): Adjust prototype.
5806 (grow_tree_vec_stat): Likewise.
5807 (make_tree_vec): Remove macro.
5808 (grow_tree_vec): Likewise.
5810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5812 * fold-const.c (fold_build1_stat_loc): Adjust.
5813 (fold_build2_stat_loc): Likewise.
5814 (fold_build3_stat_loc): Likewise.
5815 * tree.c (build0_stat): Remove _stat from name.
5816 (build1_stat): Likewise.
5817 (build2_stat): Likewise.
5818 (build3_stat): Likewise.
5819 (build4_stat): Likewise.
5820 (build5_stat): Likewise.
5821 * tree.h (build1_loc): Remove macro, and rename _stat function
5823 (build2_loc): Likewise.
5824 (build3_loc): Likewise.
5825 (build4_loc): Likewise.
5826 (build5_loc): Likewise.
5828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5830 * tree.c (make_int_cst_stat): Remove _stat from name.
5831 * tree.h (make_int_cst_stat): Adjust prototype.
5832 (make_int_cst): Remove macro.
5834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5836 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5837 * tree.h (make_tree_binfo_stat): Adjust prototype.
5838 (make_tree_binfo): Remove.
5840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5842 * tree.c (copy_node_stat): Rename to copy_node.
5843 (build_distinct_type_copy): Adjust.
5844 * tree.h (copy_node_stat): Adjust prototype.
5845 (copy_node): Remove macro.
5847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5849 * tree.c (make_node_stat): rename to make_node.
5850 (build_tree_list_stat): Adjust.
5851 (build0_stat): Likewise.
5852 (build2_stat): Likewise.
5853 (build3_stat): Likewise.
5854 (build4_stat): Likewise.
5855 (build5_stat): Likewise.
5856 (build_decl_stat): Likewise.
5857 * tree.h (make_node_stat): Adjust prototype.
5858 (make_node): remove macro.
5860 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5862 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5863 (PPC_FEATURE2_SCV): Likewise.
5864 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5866 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5868 * config/aarch64/aarch64.c
5869 (aarch64_internal_mov_immediate): Add new special pattern.
5870 * config/aarch64/aarch64.md (*movdi_aarch64):
5871 Add reg/32bit const mov case.
5873 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5874 Richard Sandiford <richard.sandiford@linaro.org>
5876 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5877 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5878 Add integer and movi cases.
5879 (movi-split-hf-df-sf split, fp16): New.
5880 (enabled): Added TARGET_FP_F16INST.
5881 * config/aarch64/iterators.md (GPF_HF): New.
5882 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5884 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5886 * config/aarch64/aarch64.c
5887 (aarch64_simd_container_mode): Add prototype.
5888 (aarch64_expand_mov_immediate): Add HI support.
5889 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5890 (aarch64_can_const_movi_rtx_p): New.
5891 (aarch64_preferred_reload_class):
5892 Remove restrictions of using FP registers for certain SIMD operations.
5893 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5894 (aarch64_valid_floating_const): Add integer move validation.
5895 (aarch64_simd_imm_scalar_p): Remove.
5896 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5897 (aarch64_legitimate_constant_p): Expand list of supported cases.
5898 * config/aarch64/aarch64-protos.h
5899 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5900 (aarch64_reinterpret_float_as_int): New.
5901 (aarch64_simd_imm_scalar_p): Remove.
5902 * config/aarch64/constraints.md (Uvi): New.
5903 (Dd): Split into Ds and new Dd.
5904 * config/aarch64/aarch64.md (*movsi_aarch64):
5906 (*movdi_aarch64): Add SIMD mov case.
5908 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5910 * tree-predcom.c: (struct chain): Handle store-store chain in which
5911 stores for elimination only store loop invariant values.
5912 (execute_pred_commoning_chain): Ditto.
5913 (prepare_initializers_chain_store_elim): Ditto.
5914 (prepare_finalizers): Ditto.
5915 (is_inv_store_elimination_chain): New function.
5916 (initialize_root_vars_store_elim_1): New function.
5918 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5920 * tree-predcom.c: Revise general description of the pass.
5921 (enum chain_type): New enum type for store elimination.
5922 (struct chain): New field supporting store elimination.
5923 (struct component): Ditto.
5924 (dump_chain): Dump store-stores chain.
5925 (release_chain): Release resources.
5926 (split_data_refs_to_components): Compute and create component
5927 contains only stores for elimination.
5928 (get_chain_last_ref_at): New function.
5929 (make_invariant_chain): Initialization.
5930 (make_rooted_chain): Specify chain type in parameter and record it.
5931 (add_looparound_copies): Skip for store-stores chain.
5932 (determine_roots_comp): Compute type of chain and pass it to
5934 (initialize_root_vars_store_elim_2): New function.
5935 (finalize_eliminated_stores): New function.
5936 (remove_stmt): Handle store for elimination.
5937 (execute_pred_commoning_chain): Execute predictive commoning on
5939 (determine_unroll_factor): Skip unroll for store-stores chain.
5940 (prepare_initializers_chain_store_elim): New function.
5941 (prepare_initializers_chain): Hanlde store-store chain.
5942 (prepare_finalizers_chain, prepare_finalizers): New function.
5943 (tree_predictive_commoning_loop): Return integer value indicating
5944 if loop is unrolled or lcssa form is corrupted.
5945 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5947 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5949 * tree-predcom.c (initialize_root): Delete.
5950 (execute_pred_commoning_chain): Initialize root vars and replace
5951 reference of non-combined chain directly, rather than call above
5954 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5956 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5957 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5959 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5961 * tree-predcom.c (struct chain): New field init_seq.
5962 (release_chain): Release init_seq.
5963 (prepare_initializers_chain): Record intialization stmts in above
5965 (insert_init_seqs): New function.
5966 (tree_predictive_commoning_loop): Call insert_init_seqs.
5968 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5970 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5972 2017-07-28 Richard Biener <rguenther@suse.de>
5974 * match.pd: Remove superfluous :c.
5975 * genmatch.c (simplify::id): Add member.
5976 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5978 (current_id): New global.
5979 (dt_node::parent): Move from ...
5980 (dt_operand::parent): ... here. Add for_id member.
5981 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5982 (decision_tree::find_node): Relax order requirement when
5983 merging DT_TRUE nodes to ones inbetween the current simplify
5984 and the one we try to merge with. Add diagnostic whenever
5985 we need to enforce pattern order by not merging.
5986 (decision_tree::insert): Set current_id.
5987 (decision_tree::print_node): Dump parent node and for_id.
5988 (parser::last_id): Add member.
5989 (parser::push_simplify): Assign unique id.
5990 (parser::parser): Initialize last_id.
5992 2017-07-28 Martin Liska <mliska@suse.cz>
5995 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5996 gimple_build_debug_bind.
5998 2017-07-28 Richard Biener <rguenther@suse.de>
6000 PR tree-optimization/81502
6001 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6003 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6005 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6006 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6007 for BIT_FIELD_REF args.
6008 * fold-const.c (make_bit_field_ref): Likewise.
6009 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6011 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6014 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6015 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6016 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6017 Or it into SANITIZER_UNDEFINED.
6018 * ubsan.c: Include gimple-fold.h and varasm.h.
6019 (ubsan_expand_ptr_ifn): New function.
6020 (instrument_pointer_overflow): New function.
6021 (maybe_instrument_pointer_overflow): New function.
6022 (instrument_object_size): Formatting fix.
6023 (pass_ubsan::execute): Call instrument_pointer_overflow
6024 and maybe_instrument_pointer_overflow.
6025 * internal-fn.c (expand_UBSAN_PTR): New function.
6026 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6027 * sanitizer.def (__ubsan_handle_pointer_overflow,
6028 __ubsan_handle_pointer_overflow_abort): New builtins.
6029 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6030 * internal-fn.def (UBSAN_PTR): New internal function.
6031 * opts.c (sanitizer_opts): Add pointer-overflow.
6032 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6033 * fold-const.c (build_range_check): Compute pointer range check in
6034 integral type if pointer arithmetics would be needed. Formatting
6037 2017-07-28 Martin Liska <mliska@suse.cz>
6040 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6041 parameters that are of a variable-length.
6043 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6045 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6047 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6048 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6049 (CRT_CALL_STATIC_FUNCTION): Likewise.
6050 (ASM_DEFAULT_SPEC): New define.
6051 (ASM_SPEC32): Likewise.
6052 (ASM_SPEC64): Likewise.
6053 (ASM_SPEC_COMMON): Likewise.
6054 (ASM_SPEC): Likewise.
6055 (INVALID_64BIT): Likewise.
6056 (LINK_OS_DEFAULT_SPEC): Likewise.
6057 (LINK_OS_SPEC32): Likewise.
6058 (LINK_OS_SPEC64): Likewise.
6059 (POWERPC_LINUX): Likewise.
6060 (PTRDIFF_TYPE): Likewise.
6061 (RESTORE_FP_PREFIX): Likewise.
6062 (RESTORE_FP_SUFFIX): Likewise.
6063 (SAVE_FP_PREFIX): Likewise.
6064 (SAVE_FP_SUFFIX): Likewise.
6065 (SIZE_TYPE): Likewise.
6066 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6067 (TARGET_64BIT): Likewise.
6068 (TARGET_64BIT): Likewise.
6069 (TARGET_AIX): Likewise.
6070 (WCHAR_TYPE_SIZE): Likewise.
6071 (WCHAR_TYPE): Undefine.
6072 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6073 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6074 (CPP_OS_RTEMS_SPEC): Delete.
6075 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6076 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6078 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6080 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6082 PR tree-optimization/81578
6083 * tree-parloops.c (build_new_reduction): Bail out if
6084 reduction_code isn't one of the standard OpenMP reductions.
6085 Move the details printing after that decision.
6087 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6089 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6090 related to reload_in_progress.
6091 (splat_input_operand): Likewise.
6092 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6094 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6096 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6097 (TARGET_INSTANTIATE_DECLS): Likewise.
6098 (legitimate_indexed_address_p): Delete reload_in_progress code.
6099 (rs6000_debug_legitimate_address_p): Likewise.
6100 (rs6000_eliminate_indexed_memrefs): Likewise.
6101 (rs6000_emit_le_vsx_store): Likewise.
6102 (rs6000_emit_move_si_sf_subreg): Likewise.
6103 (rs6000_emit_move): Likewise.
6104 (register_to_reg_type): Likewise.
6105 (rs6000_pre_atomic_barrier): Likewise.
6106 (rs6000_machopic_legitimize_pic_address): Likewise.
6107 (rs6000_allocate_stack_temp): Likewise.
6108 (rs6000_address_for_fpconvert): Likewise.
6109 (rs6000_address_for_altivec): Likewise.
6110 (rs6000_secondary_memory_needed_rtx): Delete function.
6111 (rs6000_check_sdmode): Likewise.
6112 (rs6000_alloc_sdmode_stack_slot): Likewise.
6113 (rs6000_instantiate_decls): Likewise.
6114 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6115 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6116 Delete reload_in_progress.
6117 (*vec_reload_and_plus_<mptrsize>): Likewise.
6118 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6119 (vsx_div_v2di): Likewise.
6120 (vsx_udiv_v2di): Likewise.
6122 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6124 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6125 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6126 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6127 (rs6000_debug_reg_global): Delete print of LRA status.
6128 (rs6000_option_override_internal): Delete dead LRA related code.
6129 (rs6000_lra_p): Delete function.
6130 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6132 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6134 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6135 * config/riscv/rtems.h: New file.
6137 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6138 Sudakshina Das <sudi.das@arm.com>
6140 * config/aarch64/aarch64.md
6141 (define_split for and<mode>3nr_compare): Move
6142 non aarch64_logical_operand to a register.
6143 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6144 register immediate operand to a register.
6145 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6147 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6150 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6152 2017-07-27 Richard Biener <rguenther@suse.de>
6154 PR tree-optimization/81573
6155 PR tree-optimization/81494
6156 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6157 multi defuse cycle case.
6159 2017-07-27 Richard Biener <rguenther@suse.de>
6161 PR tree-optimization/81571
6162 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6165 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6167 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6168 earlier and only if MASK_FPU is set. Adjust formatting.
6170 2017-07-27 Martin Liska <mliska@suse.cz>
6172 * opt-functions.awk: Add validation of value of Init.
6173 * optc-gen.awk: Pass new argument.
6175 2017-07-27 Martin Liska <mliska@suse.cz>
6177 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6178 Fix wrong condition.
6180 2017-07-27 Martin Liska <mliska@suse.cz>
6182 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6183 BBs and edges seen by autoFDO.
6185 2017-07-27 Richard Biener <rguenther@suse.de>
6187 PR tree-optimization/81502
6188 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6189 with incompatible but same sized type.
6190 (execute_update_addresses_taken): Likewise.
6192 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6194 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6195 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6197 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6200 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6201 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6202 Change s_operand to memory_operand.
6204 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6206 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6207 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6208 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6209 Emit instructions rather than returning an expression. Handle TFmode
6210 and KFmode by casting to TImode.
6211 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6212 (rs6000_emit_le_vsx_store): Likewise.
6213 * config/rs6000/vsx.md (VSX_TI): New iterator.
6214 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6215 (*vsx_le_undo_permute_<mode>): Likewise.
6216 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6217 emit the split sequence.
6218 (*vsx_le_perm_store_<mode>): Likewise.
6220 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6222 PR tree-optimization/81555
6223 PR tree-optimization/81556
6224 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6225 if true, force CHANGED for the recursive invocation.
6226 (reassociate_bb): Remember original length of ops array, pass
6227 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6229 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6230 attributes for noipa attribute. For naked attribute use
6231 lookup_attribute first before lookup_attribute_spec.
6232 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6234 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6235 for functions with noipa attribute.
6236 (cgraph_externally_visible_p): Return true for functions with noipa
6238 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6239 for functions with noipa attribute.
6240 * doc/extend.texi: Document noipa function attribute.
6241 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6242 also for functions with noipa attribute.
6243 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6245 2017-07-26 Andrew Pinski <apinski@cavium.com>
6247 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6248 vec_unalign_load_cost and vec_unalign_store_cost.
6250 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6252 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6253 -mvsx-small-integer option.
6254 (ISA_3_0_MASKS_IEEE): Likewise.
6255 (OTHER_VSX_VECTOR_MASKS): Likewise.
6256 (POWERPC_MASKS): Likewise.
6257 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6258 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6259 code, only testing for DImode being allowed in non-VSX floating
6261 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6262 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6264 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6265 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6266 TARGET_P8_VECTOR test.
6267 (rs6000_secondary_reload_simple_move): Likewise.
6268 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6269 since TARGET_P9_VECTOR was already tested.
6270 (rs6000_opt_masks): Remove -mvsx-small-integer.
6271 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6272 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6274 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6275 test for TARGET_VEXTRACTUB was used, and that uses
6277 (p9 extract splitter): Likewise.
6278 (vsx_extract_<mode>_di_p9): Likewise.
6279 (vsx_extract_<mode>_store_p9): Likewise.
6280 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6281 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6282 the elimination of TARGET_VSX_SMALL_INTEGER.
6283 (vsx_extract_<mode>_p8): Likewise.
6284 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6285 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6286 (vsx_set_<mode>_p9): Likewise.
6287 (vsx_set_v4sf_p9): Likewise.
6288 (vsx_set_v4sf_p9_zero): Likewise.
6289 (vsx_insert_extract_v4sf_p9): Likewise.
6290 (vsx_insert_extract_v4sf_p9_2): Likewise.
6291 * config/rs6000/rs6000.md (sign extend splitter): Change
6292 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6293 (floatsi<mode>2_lfiwax_mem): Likewise.
6294 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6295 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6296 since a test for TARGET_P9_VECTOR was used.
6297 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6298 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6299 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6300 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6301 TARGET_P8_VECTOR test.
6302 (fix_trunc<mode>si2_stfiwx): Likewise.
6303 (fix_trunc<mode>si2_internal): Likewise.
6304 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6305 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6307 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6308 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6309 TARGET_P8_VECTOR test.
6310 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6311 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6312 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6314 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6315 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6316 since a test for TARGET_P9_VECTOR was used.
6317 (splitter for loading small constants): Likewise.
6319 2017-07-26 Andrew Pinski <apinski@cavium.com>
6321 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6324 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6327 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6328 (fp_valid_at): Likewise.
6330 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6332 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6333 (qdf24xx_addrcost_table): Likewise.
6334 (cortexa57_tunings): Update to use generic_branch_cost.
6335 (cortexa72_tunings): Likewise.
6336 (cortexa73_tunings): Likewise.
6337 (qdf24xx_tunings): Likewise.
6339 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6341 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6342 (thunderx2t99_branch_cost): Likewise.
6343 (cortexa35_tunings): Update to use generic_branch_cost.
6344 (cortexa53_tunings): Likewise.
6345 (cortexa57_tunings): Likewise.
6346 (cortexa72_tunings): Likewise.
6347 (cortexa73_tunings): Likewise.
6348 (thunderx2t99_tunings): Likewise.
6350 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6352 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6353 (sparc_option_override): Honour MASK_FSMULD.
6354 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6355 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6356 * config/sparc/sparc.opt (mfsmuld): New option.
6357 * doc/invoke.texi (mfsmuld): Document option.
6359 2017-07-26 Marek Polacek <polacek@redhat.com>
6362 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6364 2017-07-26 Richard Biener <rguenther@suse.de>
6366 * gimple-match-head.c (do_valueize): Return OP if valueize
6368 (get_def): New helper to get at the def stmt of a SSA name
6370 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6371 do_valueize to get at the def stmt.
6372 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6374 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6377 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6379 2017-07-26 Martin Liska <mliska@suse.cz>
6382 * function.c (expand_function_start): Make expansion of
6383 nonlocal_goto_save_area after parm_birth_insn.
6385 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6387 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6388 from all CPU target flags enable members.
6390 2017-07-26 Richard Biener <rguenther@suse.de>
6392 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6393 (decision_tree::gen): Make 'type' const.
6394 (write_predicate): Likewise.
6396 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6398 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6399 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6400 (rs6000_option_override_internal): Likewise.
6401 (rs6000_expand_vector_set): Likewise.
6402 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6403 (TARGET_UPPER_REGS_SF): Likewise.
6404 (TARGET_UPPER_REGS_DI): Likewise.
6405 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6406 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6407 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6408 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6409 (Splitters for DI constants in Altivec registers): Likewise.
6410 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6411 (vsx_set_v4sf_p9): Likewise.
6412 (vsx_set_v4sf_p9_zero): Likewise.
6413 (vsx_insert_extract_v4sf_p9): Likewise.
6414 (vsx_insert_extract_v4sf_p9_2): Likewise.
6416 2017-07-25 Carl Love <cel@us.ibm.com>
6418 * doc/extend.texi: Update the built-in documentation file for the
6419 existing built-in functions
6420 vector signed char vec_cnttz (vector signed char);
6421 vector unsigned char vec_cnttz (vector unsigned char);
6422 vector signed short vec_cnttz (vector signed short);
6423 vector unsigned short vec_cnttz (vector unsigned short);
6424 vector signed int vec_cnttz (vector signed int);
6425 vector unsigned int vec_cnttz (vector unsigned int);
6426 vector signed long long vec_cnttz (vector signed long long);
6427 vector unsigned long long vec_cnttz (vector unsigned long long);
6429 2017-07-25 Andrew Pinski <apinski@cavium.com>
6431 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6432 accesses where the use is for the first operand of a BIT_INSERT.
6434 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6437 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6438 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6440 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6442 * config/i386/gstabs.h: Delete.
6443 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6445 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6447 * config/i386/i386.c (ix86_decompose_address): Do not check for
6448 register RTX when looking at index_reg or base_reg.
6449 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6451 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6453 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6454 to update EH info here.
6456 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6458 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6460 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6462 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6464 2017-07-25 Torsten Duwe <duwe@suse.de>
6466 * common.opt: Introduce -fpatchable-function-entry
6467 command line option, and its variables function_entry_patch_area_size
6468 and function_entry_patch_area_start.
6469 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6470 including a two-value parser.
6471 * target.def (print_patchable_function_entry): New target hook.
6472 * targhooks.h (default_print_patchable_function_entry): New function.
6473 * targhooks.c (default_print_patchable_function_entry): Likewise.
6474 * toplev.c (process_options): Switch off IPA-RA if
6475 patchable function entries are being generated.
6476 * varasm.c (assemble_start_function): Look at the
6477 patchable-function-entry command line switch and current
6478 function attributes and maybe generate NOP instructions by
6479 calling the print_patchable_function_entry hook.
6480 * doc/extend.texi: Document patchable_function_entry attribute.
6481 * doc/invoke.texi: Document -fpatchable_function_entry
6482 command line option.
6483 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6485 * doc/tm.texi: Re-generate.
6487 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6490 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6491 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6493 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6495 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6497 2017-07-25 Richard Biener <rguenther@suse.de>
6499 PR tree-optimization/81455
6500 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6501 not walk in cycles when looking for guards.
6503 2017-07-25 Richard Biener <rguenther@suse.de>
6505 PR tree-optimization/81529
6506 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6507 when optimizing backedge uses.
6509 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6511 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6513 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6514 to dl_section_ref. On AIX, append an expression to subtract
6515 the size of the section length to dl_section_ref.
6517 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6519 * configure.ac: If any of the config.* scripts fail, exit 1.
6520 * configure: Regenerate.
6522 2017-07-25 Richard Biener <rguenther@suse.de>
6525 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6526 of immediate uses, be more verbose on errors.
6528 2017-07-25 Richard Biener <rguenther@suse.de>
6530 PR tree-optimization/81510
6531 * tree-vect-loop.c (vect_is_simple_reduction): When the
6532 reduction stmt is not inside the loop bail out.
6534 2017-07-25 Richard Biener <rguenther@suse.de>
6536 PR tree-optimization/81303
6537 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6538 profitability check against LOOP_VINFO_NITERSM1.
6540 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6542 * domwalk.c (cmp_bb_postorder): Simplify.
6543 (sort_bbs_postorder): New function. Use it...
6544 (dom_walker::walk): ...here to optimize common cases.
6546 2017-07-25 Martin Liska <mliska@suse.cz>
6549 * ipa-visibility.c (function_and_variable_visibility): Make the
6550 redirection just on target that supports aliasing.
6551 Fix GNU coding style.
6553 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6556 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6558 * config/arm/rtems.h: Likewise.
6559 * config/bfin/rtems.h: Likewise.
6560 * config/i386/rtemself.h: Likewise.
6561 * config/lm32/rtems.h: Likewise.
6562 * config/m32c/rtems.h: Likewise.
6563 * config/m68k/rtemself.h: Likewise.
6564 * config/microblaze/rtems.h: Likewise.
6565 * config/mips/rtems.h: Likewise.
6566 * config/moxie/rtems.h: Likewise.
6567 * config/nios2/rtems.h: Likewise.
6568 * config/powerpcspe/rtems.h: Likewise.
6569 * config/rs6000/rtems.h: Likewise.
6570 * config/rtems.h: Likewise.
6571 * config/sh/rtems.h: Likewise.
6572 * config/sh/rtemself.h: Likewise.
6573 * config/sparc/rtemself.h: Likewise.
6575 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6578 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6579 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6580 * tree-ssa-structalias.c (alias_get_name): Same.
6582 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6585 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6586 instructions if no du chain is found.
6588 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6590 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6592 2017-07-25 Richard Biener <rguenther@suse.de>
6595 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6598 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6600 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6602 (ISA_2_7_MASKS_SERVER): Likewise.
6603 (ISA_3_0_MASKS_IEEE): Likewise.
6604 (OTHER_P8_VECTOR_MASKS): Likewise.
6605 (OTHER_VSX_VECTOR_MASKS): Likewise.
6606 (POWERPC_MASKS): Likewise.
6607 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6608 duplicate list of options.
6609 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6610 explicit -mupper-regs options.
6611 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6612 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6613 alias for -mupper-regs-df.
6614 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6615 (rs6000_init_hard_regno_mode_ok): Likewise.
6616 (rs6000_option_override_internal): Likewise.
6617 (rs6000_opt_masks): Likewise.
6618 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6619 options in terms of whether -mvsx or -mpower8-vector was used.
6620 (TARGET_UPPER_REGS_DI): Likewise.
6621 (TARGET_UPPER_REGS_SF): Likewise.
6622 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6623 -mupper-regs-* options.
6625 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6627 * passes.c (emergency_dump_function): Print some empty lines and a
6628 header before the RTL dump.
6630 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6632 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6634 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6637 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6638 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6640 2017-07-24 Carl Love <cel@us.ibm.com>
6642 * config/rs6000/rs6000-c.c: Add support for built-in functions
6643 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6644 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6645 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6646 vec_extract_fp_from_shortl): Add defines for the two builtins.
6647 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6648 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6650 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6651 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6652 * doc/extend.texi: Update the built-in documentation file for the
6653 new built-in function.
6655 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6658 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6660 * doc/generic.texi: Likewise.
6661 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6662 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6664 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6666 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6667 (aarch64_mls_elt_merge<mode>): Likewise.
6669 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6671 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6672 having __cxa_atexit.
6674 2017-07-23 Michael Collison <michael.collison@arm.com>
6676 * config/arm/arm.c (arm_option_override): Deprecate
6677 use of -mstructure-size-boundary.
6678 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6679 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6681 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6684 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6685 Reduce cost estimate for direct moves.
6687 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6690 * config/i386/i386.c (ix86_option_override_internal): Disable
6691 BMI, BMI2 and TBM instructions for -m16.
6693 2017-07-21 Carl Love <cel@us.ibm.com>
6695 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6696 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6697 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6698 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6699 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6700 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6701 VMULOSW): New enum "unspec" values.
6702 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6703 altivec_vmulosw): New patterns.
6704 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6705 VMULOSW): Add definitions.
6707 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6709 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6710 (qdf24xx): Likewise.
6711 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6712 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6713 (AARCH64_FL_V8_1): Renumber.
6714 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6715 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6716 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6717 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6718 rdma to feature modifiers list.
6720 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6723 * ipa-visibility (function_and_variable_visibility): Convert
6724 recursive PLT call to direct call if appropriate.
6726 2017-07-21 Andrew Pinski <apinski@cavium.com>
6728 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6729 operand 1 to see if the types precision matches.
6730 * fold-const.c (operand_equal_p): Likewise.
6732 2017-07-21 Richard Biener <rguenther@suse.de>
6734 PR tree-optimization/81303
6735 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6736 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6737 (vect_peeling_hash_get_lowest_cost): Adjust.
6738 (vect_enhance_data_refs_alignment): Likewise. Use
6739 vect_get_peeling_costs_all_drs to compute the penalty for no
6740 peeling to match up costs.
6742 2017-07-21 Richard Biener <rguenther@suse.de>
6744 PR tree-optimization/81500
6745 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6746 we didn't identify a reduction path.
6748 2017-07-21 Tom de Vries <tom@codesourcery.com>
6749 Cesar Philippidis <cesar@codesourcery.com>
6751 PR gcov-profile/81442
6752 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6755 2017-07-21 Tom de Vries <tom@codesourcery.com>
6758 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6760 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6761 nvptx_override_options_after_change.
6763 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6765 * dwarf2out.c (output_file_names): Avoid double testing for
6768 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6770 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6772 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6774 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6776 (try_crossjump_to_edge): Do not punt on partitioned functions.
6778 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6780 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6781 Put all BBs reachable only via paths crossing cold region to cold
6783 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6785 2016-07-21 Richard Biener <rguenther@suse.de>
6787 PR tree-optimization/81303
6788 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6789 into account prologue and epilogue iterations when raising
6790 min_profitable_iters to sth at least covering one vector iteration.
6792 2017-07-21 Tamar Christina <tamar.christina@arm.com>
6794 * config/arm/arm.c (arm_test_cpu_arch_dat):
6797 2017-07-20 Nathan Sidwell <nathan@acm.org>
6799 Remove TYPE_METHODS.
6800 * tree.h (TYPE_METHODS): Delete.
6801 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
6802 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
6803 (dbxout_type_methods): Scan TYPE_FIELDS.
6804 (dbxout_type): Don't check TYPE_METHODS here.
6805 * function.c (use_register_for_decl): Always ignore register for
6806 class types when not optimizing.
6807 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
6808 * tree.c (free_lang_data_in_type): Stitch out member functions and
6809 templates from TYPE_FIELDS.
6810 (build_distinct_type_copy, verify_type_variant,
6811 verify_type): Member fns are on TYPE_FIELDS.
6812 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
6813 * tree-pretty-print.c (dump_generic_node): Likewise.
6815 2017-07-20 Jakub Jelinek <jakub@redhat.com>
6818 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
6819 V2TImode and V4TImode.
6820 (ix86_expand_vector_extract): Likewise.
6821 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
6822 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
6823 (ssescalarmode): Handle V4TImode and V2TImode.
6824 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6825 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6826 (VEXTRACTI128_MODE): New mode iterator.
6827 (splitter for *vec_extractv?ti first element): New.
6828 (VEC_INIT_MODE): New mode iterator.
6829 (vec_init<mode>): Consolidate 3 expanders into one using
6830 VEC_INIT_MODE mode iterator.
6832 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6834 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6835 non_spilled_static_chain_regno_p.
6837 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6839 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6841 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6843 * bb-reorder.c (connect_traces): Allow copying of blocks within
6846 2017-07-20 Richard Biener <rguenther@suse.de>
6848 * gimple.h (gimple_phi_result): Add gphi * overload.
6849 (gimple_phi_result_ptr): Likewise.
6850 (gimple_phi_arg): Likewise. Adjust index assert to only
6851 allow actual argument accesses rather than all slots available
6853 (gimple_phi_arg_def): Add gphi * overload.
6854 * tree-phinodes.c (make_phi_node): Initialize only actual
6856 (resize_phi_node): Clear memory not covered by old node,
6857 do not initialize excess argument slots.
6858 (reserve_phi_args_for_new_edge): Initialize new argument slot
6861 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6863 PR tree-optimization/81388
6865 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6867 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6868 by removing computation of may_be_zero.
6870 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6871 Tom de Vries <tom@codesourcery.com>
6874 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6875 when gimple level profile disagrees with what RTL expander did.
6877 2017-07-20 Richard Biener <rguenther@suse.de>
6879 PR tree-optimization/61171
6880 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6881 (vect_analyze_stmt): Add slp instance parameter.
6882 (vectorizable_reduction): Likewise.
6883 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6884 (vect_is_simple_reduction): Deal with chains not detected
6885 as SLP reduction chain, specifically not properly associated
6886 chains containing a mix of plus/minus.
6887 (get_reduction_op): Remove.
6888 (get_initial_defs_for_reduction): Simplify, pass in whether
6889 this is a reduction chain, pass in the SLP node for the PHIs.
6890 (vect_create_epilog_for_reduction): Get the SLP instance as
6892 (vectorizable_reduction): Get the SLP instance as arg.
6893 During analysis remember the SLP node with the PHIs in the
6894 instance. Simplify getting at the vectorized reduction PHIs.
6895 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6896 through SLP instance.
6897 (vect_slp_analyze_operations): Likewise.
6898 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6899 (vect_transform_stmt): Likewise.
6901 2017-07-20 Tom de Vries <tom@codesourcery.com>
6903 PR tree-optimization/81489
6904 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6905 read of phi arg location to before loop that modifies phi.
6907 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6909 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6912 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6915 * except.c (execute): Fix ordering issue.
6917 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6919 PR rtl-optimization/81423
6920 * combine.c (make_compound_operation_int): Don't try to optimize
6921 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6923 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6925 PR rtl-optimization/81423
6926 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6927 with a constant that is -1 in the truncated to mode.
6929 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6931 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6932 (determine_unlikely_bbs): ... here.
6933 * predict.h (propagate_unlikely_bbs_forward): Declare.
6934 * cfgexpand.c (pass_expand::execute): Use it.
6935 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6937 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6938 propagate_unlikely_bbs_forward.
6940 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6943 * except.c (maybe_add_nop_after_section_switch): New function.
6946 2017-07-19 Tom de Vries <tom@codesourcery.com>
6948 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6950 2017-07-19 Tom de Vries <tom@codesourcery.com>
6952 * gimple.h (gimple_phi_arg): Make assert more strict.
6954 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6956 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6957 * config/rs6000/mmintrin.h: New file.
6958 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6960 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6962 PR tree-optimization/81346
6963 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6965 2017-07-19 Tom de Vries <tom@codesourcery.com>
6967 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6969 2017-07-19 Tom de Vries <tom@codesourcery.com>
6971 * config/nvptx/nvptx-modes.def: Add V2DImode.
6972 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6973 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6974 (nvptx_output_mov_insn): Handle lack of mov.b128.
6975 (nvptx_print_operand): Handle 'H' and 'L' codes.
6976 (nvptx_vector_mode_supported): Allow V2DImode.
6977 (nvptx_preferred_simd_mode): New function.
6978 (nvptx_data_alignment): New function.
6979 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6980 nvptx_preferred_simd_mode.
6981 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6983 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6985 2017-07-19 Tom de Vries <tom@codesourcery.com>
6987 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6988 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6989 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6990 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6991 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6992 (mov<VECIM>_insn): New define_insn.
6993 (define_expand "mov<VECIM>): New define_expand.
6995 2017-07-19 Tom de Vries <tom@codesourcery.com>
6997 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6999 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7001 PR tree-optimization/81346
7002 * fold-const.h (fold_div_compare, range_check_type): Declare.
7003 * fold-const.c (range_check_type): New function.
7004 (build_range_check): Use range_check_type.
7005 (fold_div_compare): No longer static, rewritten into
7006 a match.pd helper function.
7007 (fold_comparison): Don't call fold_div_compare here.
7008 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7011 2017-07-19 Nathan Sidwell <nathan@acm.org>
7013 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7014 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7015 * tree.c (find_decls_types_r, verify_type): Use
7016 TYPE_{MIN,MAX}_VALUE_RAW.
7017 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7018 (hash_tree): Likewise.
7019 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7021 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7024 2017-07-18 Tom de Vries <tom@codesourcery.com>
7027 * omp-expand.c (expand_omp_for_static_chunk): Handle
7028 equal-argument loop exit phi.
7030 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7033 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7034 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7035 operand 2 predicate.
7036 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7037 operand 2 predicate.
7038 (ror,rol -> rorx splitters): Use const_int_operand as
7039 operand 2 predicate.
7041 2017-06-18 Richard Biener <rguenther@suse.de>
7043 PR tree-optimization/81410
7044 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7045 the gap in the ! slp_perm SLP case after each group.
7047 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7050 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7053 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7056 * predict.c (set_even_probabilities): Cleanup; do not affect
7057 probabilities that are already known.
7058 (combine_predictions_for_bb): Call even when count is set.
7060 2017-07-18 Nathan Sidwell <nathan@acm.org>
7062 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7065 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7068 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7069 optimization for loop niter analysis.
7071 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7074 * config/avr/avr.c (avr_optimize_casesi): Don't use
7075 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7077 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7079 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7080 body_cost_vec from _vect_peel_extended_info.
7081 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7082 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7085 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7087 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7089 2017-07-18 Richard Biener <rguenther@suse.de>
7091 PR tree-optimization/80620
7092 PR tree-optimization/81403
7093 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7094 info when re-using a VN table entry.
7096 2017-07-18 Richard Biener <rguenther@suse.de>
7098 PR tree-optimization/81418
7099 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7100 vectype_in. Verify that with lane-reducing reduction operations
7101 we have a single def-use cycle.
7103 2017-07-17 Carl Love <cel@us.ibm.com>
7105 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7107 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7108 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7109 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7110 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7111 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7112 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7113 VMULOSW): New enum "unspec" values.
7114 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7115 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7116 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7117 altivec_vmulosw): New patterns.
7118 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7119 VMULOSW): Add definitions.
7120 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7122 * config/alpha/alpha.c: Include predict.h.
7124 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7126 * tree-vrp.c (compare_assert_loc): Fix comparison function
7127 to return predictable results.
7129 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7131 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7134 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7135 * doc/invoke.texi (mexpand-adddi): Update text.
7137 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7139 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7140 that also clobbers the CC register. The old expand code is moved
7142 (*arc_clzsi2): ... here.
7143 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7144 the CC register. The old expand code is moved to ...
7145 (arc_ctzsi2): ... here.
7147 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7149 * config/arc/arc.opt (mindexed-loads): Use initial value
7150 TARGET_INDEXED_LOADS_DEFAULT.
7151 (mauto-modify-reg): Use initial value
7152 TARGET_AUTO_MODIFY_REG_DEFAULT.
7153 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7154 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7155 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7156 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7158 2017-07-17 Martin Liska <mliska@suse.cz>
7161 * opts.c (finish_options): Do not allow -fgnu-tm
7162 w/ -fsanitize={kernel-,}address. Say sorry.
7164 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7167 * tree-loop-distribution.c (classify_partition): Only assert on
7168 numer of iterations.
7169 (merge_dep_scc_partitions): Delete prameter. Update function call.
7170 (distribute_loop): Remove code handling loop with unknown niters.
7171 (pass_loop_distribution::execute): Skip loop with unknown niters.
7173 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7176 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7177 function sort_partitions_by_post_order.
7179 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7181 PR tree-optimization/81374
7182 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7183 the max index of basic blocks, rather than number of basic blocks.
7185 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7187 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7189 (arc_legitimate_pic_operand_p): Likewise.
7190 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7192 (arc_needs_pcl_p): Likewise.
7193 (arc_legitimate_pc_offset_p): Likewise.
7194 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7195 function is also used in constrains.md.
7196 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7197 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7198 PLUS. Only return true/false in known cases, otherwise assert.
7199 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7200 is already called in arc_legitimate_constant_p.
7201 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7203 (LEGITIMATE_PIC_OPERAND_P): Use
7204 arc_raw_symbolic_reference_mentioned_p function.
7205 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7210 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7211 Andrew Burgess <andrew.burgess@embecosm.com>
7213 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7214 (arc_return_address_register): New function.
7215 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7216 (arc_handle_fndecl_attribute): Add naked attribute.
7217 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7218 (TARGET_WARN_FUNC_RETURN): Likewise.
7219 (arc_allocate_stack_slots_for_args): New function.
7220 (arc_warn_func_return): Likewise.
7221 (machine_function): Change type fn_type.
7222 (arc_compute_function_type): Consider new naked function type,
7223 change function return type.
7224 (arc_must_save_register): Adapt to handle new
7225 arc_compute_function_type's return type.
7226 (arc_expand_prologue): Likewise.
7227 (arc_expand_epilogue): Likewise.
7228 (arc_return_address_regs): Delete.
7229 (arc_return_address_register): New function.
7230 (arc_epilogue_uses): Use above function.
7231 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7232 (arc_function_type): Change encoding, add naked type.
7233 (ARC_INTERRUPT_P): Change to handle the new encoding.
7234 (ARC_FAST_INTERRUPT_P): Likewise.
7235 (ARC_NORMAL_P): Define.
7236 (ARC_NAKED_P): Likewise.
7237 (arc_compute_function_type): Delete prototype.
7238 * config/arc/arc.md (in_ret_delay_slot): Use
7239 arc_return_address_register function.
7240 (simple_return): Likewise.
7241 (p_return_i): Likewise.
7243 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7245 PR tree-optimization/81428
7246 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7247 can't be built for those types.
7249 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7251 Remove stuff dead since r239246.
7253 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7254 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7255 (avr_inform_devices): Remove dead stuff.
7257 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7259 * config/arm/arm_neon.h: Fix softp typo.
7261 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7263 PR tree-optimization/81365
7264 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7265 aggregate moves onto bb predecessor edges, make sure there are no
7266 loads that could alias the lhs in between the start of bb and the
7269 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7272 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7273 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7274 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7276 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7278 PR tree-optimization/81396
7279 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7280 (init_symbolic_number): Initialize it to 1.
7281 (perform_symbolic_merge): Add n_ops from both operands into the new
7283 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7284 without base_addr as useless if they need more than one operation.
7285 (bswap_replace): Handle !bswap case for NULL base_addr.
7287 2017-07-17 Tom de Vries <tom@codesourcery.com>
7290 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7293 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7295 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7296 conditional builtin define __FIX_LEON3FT_B2BST.
7298 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7300 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7301 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7304 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7306 PR rtl-optimization/81424
7307 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7308 to remove potential trapping from operands if -fnon-call-exceptions.
7310 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7312 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7313 profile_proability for scalling.
7314 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7316 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7318 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7320 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7322 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7323 fixpoint arithmetics.
7325 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7327 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7328 fixpoint arithmetics.
7330 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7332 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7333 fixpoint arithmetics.
7335 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7337 * profile-count.h (profile_probability::from_reg_br_prob_note,
7338 profile_probability::to_reg_br_prob_note): New functions.
7339 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7340 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7341 * predict.c (probability_reliable_p): Update.
7342 (edge_probability_reliable_p): Update.
7343 (br_prob_note_reliable_p): Update.
7344 (invert_br_probabilities): Update.
7345 (add_reg_br_prob_note): New function.
7346 (combine_predictions_for_insn): Update.
7347 * asan.c (asan_clear_shadow): Update.
7348 * cfgbuild.c (compute_outgoing_frequencies): Update.
7349 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7350 (update_br_prob_note): Update.
7351 (rtl_verify_edges): Update.
7352 (purge_dead_edges): Update.
7353 (fixup_reorder_chain): Update.
7354 * emit-rtl.c (try_split): Update.
7355 * ifcvt.c (cond_exec_process_insns): Update.
7356 (cond_exec_process_if_block): Update.
7357 (dead_or_predicable): Update.
7358 * internal-fn.c (expand_addsub_overflow): Update.
7359 (expand_neg_overflow): Update.
7360 (expand_mul_overflow): Update.
7361 * loop-doloop.c (doloop_modify): Update.
7362 * loop-unroll.c (compare_and_jump_seq): Update.
7363 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7364 * predict.h: Update.
7365 * reorg.c (mostly_true_jump): Update.
7367 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7368 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7369 * config/arc/arc.c: (emit_unlikely_jump): Update.
7370 * config/arm/arm.c: (emit_unlikely_jump): Update.
7371 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7372 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7373 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7374 (ix86_print_operand): Update.
7375 (ix86_split_fp_branch): Update.
7376 (predict_jump): Update.
7377 * config/ia64/ia64.c (ia64_print_operand): Update.
7378 * config/mmix/mmix.c (mmix_print_operand): Update.
7379 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7380 (rs6000_expand_split_stack_prologue): Update.
7381 * config/rs6000/rs6000.c: Update.
7382 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7383 (s390_expand_vec_movstr): Update.
7384 (s390_expand_cs_tdsi): Update.
7385 (s390_expand_split_stack_prologue): Update.
7386 * config/sh/sh.c (sh_print_operand): Update.
7387 (expand_cbranchsi4): Update.
7388 (expand_cbranchdi4): Update.
7389 * config/sparc/sparc.c (output_v9branch): Update.
7390 * config/spu/spu.c (get_branch_target): Update.
7391 (ea_load_store_inline): Update.
7392 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7393 * config/tilepro/tilepro.c: Update.
7395 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7397 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7398 (gimplify_save_expr): Likewise.
7400 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7402 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7404 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7406 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7408 * timevar.def (TV_IPA_FNSUMMARY): Define.
7410 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7412 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7413 to back store errata sensitive sequence from being generated.
7414 (sqrtdf2_fix): Likewise.
7416 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7418 * tree-ssa-threadupdate.c (compute_path_counts,
7419 update_joiner_offpath_counts): Use profile_probability.
7421 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7424 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7426 * config/arm/arm-c.c (arm_cpu_builtins): Define
7427 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7429 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7431 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7432 array entries to represent __ieee128 versions of the
7433 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7434 scalar_extract_sig, and scalar_insert_exp built-in functions.
7435 (altivec_resolve_overloaded_builtin): Add special case handling
7436 for the __builtin_scalar_insert_exp function, as represented by
7437 the P9V_BUILTIN_VEC_VSIEDP constant.
7438 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7439 exponent support for __ieee128 argument.
7440 (VSESQP): Add scalar extract signature support for __ieee128
7442 (VSTDCNQP): Add scalar test negative support for __ieee128
7444 (VSIEQP): Add scalar insert exponent support for __int128 argument
7445 with __ieee128 result.
7446 (VSIEQPF): Add scalar insert exponent support for __ieee128
7447 argument with __ieee128 result.
7448 (VSTDCQP): Add scalar test data class support for __ieee128
7450 (VSTDCNQP): Add overload support for scalar test negative with
7452 (VSTDCQP): Add overload support for scalar test data class
7454 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7456 (UNSPEC_VSX_SIEXPQP): New constant.
7457 (xsxexpqp): New insn for VSX scalar extract exponent quad
7459 (xsxsigqp): New insn for VSX scalar extract significand quad
7461 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7462 precision with floating point argument.
7463 (xststdcqp): New expand for VSX scalar test data class quad
7465 (xststdcnegqp): New expand for VSX scalar test negative quad
7467 (xststdcqp): New insn to match expansions for VSX scalar test data
7468 class quad precision and VSX scalar test negative quad precision.
7469 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7470 special case operand checking to enforce that second operand of
7471 VSX scalar test data class with quad precision argument is a 7-bit
7473 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7474 prototypes and descriptions of __ieee128 versions of
7475 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7476 scalar_test_data_class, and scalar_test_neg built-in functions.
7478 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7480 PR tree-optimization/81162
7481 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7482 replace a negate with an add.
7484 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7486 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7488 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7490 * config/arm/arm-c.c (arm_cpu_builtins): Define
7491 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7493 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7495 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7496 (armv8-r): Set ARM Cortex-R52 as default CPU.
7497 * config/arm/arm-tables.opt: Regenerate.
7498 * config/arm/arm-tune.md: Regenerate.
7499 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7501 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7502 extension for -mcpu=cortex-r52.
7504 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7506 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7507 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7508 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7509 (fp-armv8): Define it as FP_ARMv8 only.
7510 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7511 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7513 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7514 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7515 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7516 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7517 than TARGET_FPU_ARMV8.
7518 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7519 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7520 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7522 * config/arm/neon.md (neon_vrint): Likewise.
7523 (neon_vcvt): Likewise.
7524 (neon_<fmaxmin_op><mode>): Likewise.
7525 (<fmaxmin><mode>3): Likewise.
7526 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7527 * config/arm/predicates.md (arm_cond_move_operator): Check against
7528 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7530 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7532 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7535 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7537 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7538 loop in comment with memset.
7540 2017-07-14 Martin Liska <mliska@suse.cz>
7542 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7543 * dwarf2out.c (is_java): Remove the function.
7544 (output_pubname): Remove usage of the function.
7545 (lower_bound_default): Remove usage of DW_LANG_Java.
7546 (gen_compile_unit_die): Likewise.
7547 * gcc.c: Remove compiler defaults for .java and .zip files.
7548 * gimple-expr.c (remove_suffix): Change as there's no longer
7549 extension than 4-letter one.
7550 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7551 (gimplify_save_expr): Likewise.
7552 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7553 as it's possible even for other languages than Java.
7554 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7555 * lto-opts.c (lto_write_options): Remove reference to Java.
7556 * opts.c (strip_off_ending): Update file extension handling.
7557 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7558 * tree-eh.c (lower_resx): Likewise.
7559 * tree.c (free_lang_data_in_type): Remove dead code.
7560 (find_decls_types_r): Likewise.
7561 (build_common_builtin_nodes): Remove Java from a comment.
7562 (verify_type): Remove dead code.
7563 * varasm.c (assemble_external): Remove Java from a comment.
7565 2017-07-14 Martin Liska <mliska@suse.cz>
7567 * opts.c (finish_options): Add quotes.
7568 (common_handle_option): Likewise.
7570 2017-07-14 Martin Liska <mliska@suse.cz>
7572 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7573 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7575 * dwarf2out.c (lower_bound_default): Do not handle
7577 (gen_compile_unit_die): Likewise.
7578 * gcc.c: Remove default extension binding for GNU Pascal.
7579 * stmt.c: Remove Pascal language from a comment.
7580 * xcoffout.c: Likewise.
7582 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7585 * diagnostic-show-locus.c (fixit_cmp): New function.
7586 (layout::layout): Sort m_fixit_hints.
7587 (column_range::column_range): Assert that the values are valid.
7588 (struct char_span): New struct.
7589 (correction::overwrite): New method.
7590 (struct source_line): New struct.
7591 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7592 calls in terms of classes source_line and char_span, and
7593 correction::overwrite.
7594 (selftest::test_overlapped_fixit_printing_2): New function.
7595 (selftest::diagnostic_show_locus_c_tests): Call it.
7597 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7599 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7600 early if there is no lhs.
7602 2017-07-13 Martin Liska <mliska@suse.cz>
7604 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7605 (gen_reference_type_die): Likewise.
7606 * stor-layout.c: Remove Pascal-related comment.
7608 2017-07-13 Martin Liska <mliska@suse.cz>
7610 * opts.c (finish_options): Add quotes to error messages.
7611 (parse_sanitizer_options): Likewise.
7613 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7615 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7617 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7619 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7621 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7623 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7625 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7627 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7630 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7631 provides the hardware capability bits, define the macro
7632 __BUILTIN_CPU_SUPPORTS__.
7633 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7634 if GLIBC does not provide the hardware capability bits. Add a
7635 gcc_unreachable call if the built-in cpu function is neither
7636 __builtin_cpu_is nor __builtin_cpu_supports.
7637 (rs6000_get_function_versions_dispatcher): Change the warning
7638 that an old GLIBC is used which does not export the capability
7639 bits to be an error.
7640 * doc/extend.texi (target_clones attribute): Document the
7641 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7642 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7643 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7644 the macros defined by GCC if the newer GLIBC is available.
7646 2017-07-12 Jeff Law <law@redhat.com>
7648 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7649 remaining includes slightly.
7650 * config/riscv/riscv-builtins.c: Include profile-count.h.
7652 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7655 * config/avr/avr.c (avr_set_current_function): In diagnostic
7656 messages: Quote keywords and (parts of) identifiers.
7657 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7660 2017-07-12 Carl Love <cel@us.ibm.com>
7662 * config/rs6000/rs6000-c.c: Add support for built-in functions
7663 vector bool char vec_revb (vector bool char);
7664 vector bool short vec_revb (vector short char);
7665 vector bool int vec_revb (vector bool int);
7666 vector bool long long vec_revb (vector bool long long);
7667 * doc/extend.texi: Update the built-in documentation file for the
7668 new built-in functions.
7670 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7672 * config/s390/s390.md: Remove movcc splitter.
7674 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7676 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7677 load/store on condition.
7679 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7682 * config/avr/avr.c (avr_encode_section_info)
7683 [progmem && !TREE_READONLY]: Error if progmem object needs
7686 2017-07-11 Michael Collison <michael.collison@arm.com>
7688 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7691 2017-07-11 Carl Love <cel@us.ibm.com>
7693 * config/rs6000/rs6000-c.c: Add support for builtins
7694 vector unsigned int vec_parity_lsbb (vector signed int);
7695 vector unsigned int vec_parity_lsbb (vector unsigned int);
7696 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7697 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7698 vector unsigned long long vec_parity_lsbb (vector signed long long);
7699 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7700 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7701 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7702 * doc/extend.texi: Update the built-in documentation file for the
7703 new built-in functions.
7705 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7707 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7708 (layout::m_primary_loc): New field.
7709 (layout::layout): Initialize new field. Move location filtering
7710 logic from here to...
7711 (layout::maybe_add_location_range): ...this new method. Add
7712 support for filtering to just the lines already specified by other
7714 (layout::will_show_line_p): New method.
7715 (gcc_rich_location::add_location_if_nearby): New method.
7716 (selftest::test_add_location_if_nearby): New test function.
7717 (selftest::diagnostic_show_locus_c_tests): Call it.
7718 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7721 2017-07-11 Tom de Vries <tom@codesourcery.com>
7723 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7724 (bb_first_real_insn): New function.
7725 (nvptx_single): Add extra initialization of broadcasted condition
7728 2017-07-11 Nathan Sidwell <nathan@acm.org>
7730 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7732 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7734 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7735 Binutils doc as TEXI will mess them up.
7736 * doc/invoke.texi (AVR Options): Same here.
7738 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7740 * config/sparc/sparc.opt (mfix-ut700): New option.
7741 (mfix-gr712rc): Likewise.
7742 (sparc_fix_b2bst): New variable.
7743 * doc/invoke.texi (SPARC options): Document them.
7744 (ARM options): Fix warnings.
7745 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7746 instructions to prevent sequences that can trigger the store-store
7747 errata for certain LEON3FT processors.
7748 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7749 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7750 * config/sparc/sparc.md (fix_b2bst): New attribute.
7751 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7753 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7756 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7758 (*rsqrtsf2_sse): Ditto.
7760 (div<mode>3): Macroize insn from divdf3 and divsf3
7761 using MODEF mode iterator.
7763 2017-07-10 Martin Sebor <msebor@redhat.com>
7765 PR tree-optimization/80397
7766 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7767 instead of testing for equality to INTEGER_TYPE.
7769 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7771 * config.gcc: Remove uclibc from arc target spec.
7773 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7775 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7777 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7780 * lto-wrapper.c (remove_option): New function.
7781 (merge_and_complain): Merge PIC/PIE options more realistically.
7783 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7785 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7789 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7790 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7791 * config.in: Regenerate.
7792 * configure: Regenerate.
7793 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
7794 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
7795 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
7796 (TARGET_GASISR_PROLOGUES): ...target mask.
7797 * common/config/avr/avr-common.c
7798 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
7799 Set -mgas-isr-prologues.
7800 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
7801 INSERT_PASS_BEFORE for it.
7802 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
7803 * config/avr/avr.c (avr_option_override)
7804 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
7805 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
7806 (avr_attribute_table) <no_gccisr>: Add new function attribute.
7807 (avr_set_current_function) <is_no_gccisr>: Init machine field.
7808 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
7810 (make_avr_pass_pre_proep): New function.
7811 (emit_push_sfr) <treg>: Add argument to function and use it
7813 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
7814 and set machine->gasisr.yes.
7815 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
7816 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
7817 __gcc_isr.n_pushed to .L__stack_usage.
7818 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
7819 (avr_asm_final_postscan_insn): ...this new static function.
7820 * config/avr/avr.h (machine_function)
7821 <is_no_gccisr, use_L__stack_usage>: New fields.
7822 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
7823 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
7824 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7825 (gasisr, *gasisr): New expander and insn.
7826 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7827 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7828 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7830 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7832 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7835 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7837 Move jump-tables out of .text again.
7840 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7841 (ASM_OUTPUT_ADDR_VEC): New function.
7842 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7843 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7844 INSN_ADDRESSes as asm comment.
7845 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7846 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7847 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7848 * config/avr/avr.md (*tablejump): Adjust comment.
7849 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7850 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7852 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7853 (avr_output_addr_vec): New proto.
7854 (avr_log_t) <insn_addresses>: New field.
7856 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7859 * config/i386/i386.c (ix86_function_arg_advance): Set
7860 outgoing_args_on_stack to true if there are outgoing arguments
7862 (ix86_function_arg): Likewise.
7863 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7864 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7865 * config/i386/i386.h (machine_function): Add
7866 outgoing_args_on_stack.
7868 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7870 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7872 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7874 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7876 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7877 (REAL_H): Remove $(MACHMODE_H).
7878 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7880 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7881 $(MACHMODE_H) and double-int.h.
7882 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7884 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7887 2017-07-07 Andrew Pinski <apinski@cavium.com>
7889 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7890 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7892 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7894 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7895 Add warning if GCC was not configured to link against a GLIBC that
7896 exports the hardware capability bits.
7897 (make_resolver_func): Make resolver function private and not a
7898 COMDAT function. Create the name with clone_function_name instead
7899 of make_unique_name.
7902 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7903 correct operand in doing the split.
7905 2017-07-07 Carl Love <cel@us.ibm.com>
7907 * config/rs6000/rs6000-c: Add support for built-in function
7908 vector unsigned short vec_pack_to_short_fp32 (vector float,
7910 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7911 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7912 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7913 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7914 (convert_4f32_8i16): Add define_expand.
7915 * doc/extend.texi: Update the built-in documentation file for the
7916 new built-in function.
7918 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7920 * config/sparc/m8.md: New file.
7921 * config/sparc/sparc.md: Include m8.md.
7923 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7925 * config/sparc/sparc.opt: New option -mvis4b.
7926 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7927 (sparc_option_override): Handle VIS4B.
7928 (enum sparc_builtins): Define
7929 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7930 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7931 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7932 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7933 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7934 (check_constant_argument): New function.
7935 (sparc_vis_init_builtins): Define builtins
7936 __builtin_vis_dictunpack{8,16,32},
7937 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7938 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7939 __builtin_vis_fpcmpde{8,16,32}shl and
7940 __builtin_vis_fpcmpur{8,16,32}shl.
7941 (sparc_expand_builtin): Check that the constant operands to
7942 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7943 constant and in range.
7944 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7946 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7947 (SPARC_IMM5_P): Likewise.
7948 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7949 (enabled): Handle vis4b.
7950 (UNSPEC_DICTUNPACK): New unspec.
7951 (UNSPEC_FPCMPSHL): Likewise.
7952 (UNSPEC_FPUCMPSHL): Likewise.
7953 (UNSPEC_FPCMPDESHL): Likewise.
7954 (UNSPEC_FPCMPURSHL): Likewise.
7955 (cpu_feature): New CPU feature `vis4b'.
7956 (dictunpack{8,16,32}): New insns.
7957 (FPCSMODE): New mode iterator.
7958 (fpcscond): New code iterator.
7959 (fpcsucond): Likewise.
7960 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7961 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7962 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7963 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7964 * config/sparc/constraints.md: Define constraints `q' for unsigned
7965 2-bit integer constants and `t' for unsigned 5-bit integer
7967 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7969 (imm5_operand_dictunpack16): Likewise.
7970 (imm5_operand_dictunpack32): Likewise.
7971 (imm2_operand): Likewise.
7972 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7973 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7974 ditunpack* and fpcmp*shl builtins.
7976 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7978 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7979 * config.in: Add HAVE_AS_SPARC6 define.
7980 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7982 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7984 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7985 (CPP_CPU_SPEC): Handle m8.
7986 (ASM_CPU_SPEC): Likewise.
7987 * config/sparc/sparc-opts.h (enum processor_type): Add
7989 * config/sparc/sparc.c (m8_costs): New struct.
7990 (sparc_option_override): Handle TARGET_CPU_m8.
7991 (sparc32_initialize_trampoline): Likewise.
7992 (sparc64_initialize_trampoline): Likewise.
7993 (sparc_issue_rate): Likewise.
7994 (sparc_register_move_cost): Likewise.
7995 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7996 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7997 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7998 (CPP_CPU_SPEC): Handle M8.
7999 (ASM_CPU_SPEC): Likewise.
8000 (AS_M8_FLAG): Define.
8001 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8002 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8003 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8005 * configure: Regenerate.
8006 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8009 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8011 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8013 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8014 ("*movdi_insn_sp32"): Do not set v3pipe.
8015 ("*movsi_insn"): Likewise.
8016 ("*movdi_insn_sp64"): Likewise.
8017 ("*movsf_insn"): Likewise.
8018 ("*movdf_insn_sp32"): Likewise.
8019 ("*movdf_insn_sp64"): Likewise.
8020 ("*zero_extendsidi2_insn_sp64"): Likewise.
8021 ("*sign_extendsidi2_insn"): Likewise.
8022 ("*mov<VM32:mode>_insn"): Likewise.
8023 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8024 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8025 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8026 ("<vlop:code><VL:mode>3"): Likewise.
8027 ("*not_<vlop:code><VL:mode>3"): Likewise.
8028 ("*nand<VL:mode>_vis"): Likewise.
8029 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8030 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8031 ("one_cmpl<VL:mode>2"): Likewise.
8032 ("faligndata<VM64:mode>_vis"): Likewise.
8033 ("alignaddrsi_vis"): Likewise.
8034 ("alignaddrdi_vis"): Likweise.
8035 ("alignaddrlsi_vis"): Likewise.
8036 ("alignaddrldi_vis"): Likewise.
8037 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8038 ("bmaskdi_vis"): Likewise.
8039 ("bmasksi_vis"): Likewise.
8040 ("bshuffle<VM64:mode>_vis"): Likewise.
8041 ("cmask8<P:mode>_vis"): Likewise.
8042 ("cmask16<P:mode>_vis"): Likewise.
8043 ("cmask32<P:mode>_vis"): Likewise.
8044 ("pdistn<P:mode>_vis"): Likewise.
8045 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8047 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8049 * config/sparc/sparc.md ("subtype"): New insn attribute.
8050 ("*wrgsr_sp64"): Set insn subtype.
8051 ("*rdgsr_sp64"): Likewise.
8052 ("alignaddrsi_vis"): Likewise.
8053 ("alignaddrdi_vis"): Likewise.
8054 ("alignaddrlsi_vis"): Likewise.
8055 ("alignaddrldi_vis"): Likewise.
8056 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8057 ("fexpand_vis"): Likewise.
8058 ("fpmerge_vis"): Likewise.
8059 ("faligndata<VM64:mode>_vis"): Likewise.
8060 ("bshuffle<VM64:mode>_vis"): Likewise.
8061 ("cmask8<P:mode>_vis"): Likewise.
8062 ("cmask16<P:mode>_vis"): Likewise.
8063 ("cmask32<P:mode>_vis"): Likewise.
8064 ("fchksm16_vis"): Likewise.
8065 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8066 ("fmean16_vis"): Likewise.
8067 ("fp<plusminus_insn>64_vis"): Likewise.
8068 ("<plusminus_insn>v8qi3"): Likewise.
8069 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8070 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8071 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8072 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8073 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8074 ("*movqi_insn"): Likewise.
8075 ("*movhi_insn"): Likewise.
8076 ("*movsi_insn"): Likewise.
8077 ("movsi_pic_gotdata_op"): Likewise.
8078 ("*movdi_insn_sp32"): Likewise.
8079 ("*movdi_insn_sp64"): Likewise.
8080 ("movdi_pic_gotdata_op"): Likewise.
8081 ("*movsf_insn"): Likewise.
8082 ("*movdf_insn_sp32"): Likewise.
8083 ("*movdf_insn_sp64"): Likewise.
8084 ("*zero_extendhisi2_insn"): Likewise.
8085 ("*zero_extendqihi2_insn"): Likewise.
8086 ("*zero_extendqisi2_insn"): Likewise.
8087 ("*zero_extendqidi2_insn"): Likewise.
8088 ("*zero_extendhidi2_insn"): Likewise.
8089 ("*zero_extendsidi2_insn_sp64"): Likewise.
8090 ("ldfsr"): Likewise.
8091 ("prefetch_64"): Likewise.
8092 ("prefetch_32"): Likewise.
8093 ("tie_ld32"): Likewise.
8094 ("tie_ld64"): Likewise.
8095 ("*tldo_ldub_sp32"): Likewise.
8096 ("*tldo_ldub1_sp32"): Likewise.
8097 ("*tldo_ldub2_sp32"): Likewise.
8098 ("*tldo_ldub_sp64"): Likewise.
8099 ("*tldo_ldub1_sp64"): Likewise.
8100 ("*tldo_ldub2_sp64"): Likewise.
8101 ("*tldo_ldub3_sp64"): Likewise.
8102 ("*tldo_lduh_sp32"): Likewise.
8103 ("*tldo_lduh1_sp32"): Likewise.
8104 ("*tldo_lduh_sp64"): Likewise.
8105 ("*tldo_lduh1_sp64"): Likewise.
8106 ("*tldo_lduh2_sp64"): Likewise.
8107 ("*tldo_lduw_sp32"): Likewise.
8108 ("*tldo_lduw_sp64"): Likewise.
8109 ("*tldo_lduw1_sp64"): Likewise.
8110 ("*tldo_ldx_sp64"): Likewise.
8111 ("*mov<VM32:mode>_insn"): Likewise.
8112 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8113 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8115 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8117 * config/sparc/sparc.md ("type"): New insn type viscmp.
8118 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8120 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8121 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8122 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8123 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8125 ("n7_vis_logical_11cycle"): Likewise.
8126 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8127 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8128 * config/sparc/niagara.md ("niag_vis"): Likewise.
8129 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8130 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8132 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8134 * config/sparc/sparc.md: New instruction type `bmask'.
8135 (bmaskdi_vis): Use the `bmask' type.
8136 (bmasksi_vis): Likewise.
8137 * config/sparc/ultra3.md (us3_array): Likewise.
8138 * config/sparc/niagara7.md (n7_array): Likewise.
8139 * config/sparc/niagara4.md (n4_array): Likewise.
8140 * config/sparc/niagara2.md (niag2_vis): Likewise.
8141 (niag3_vis): Likewise.
8142 * config/sparc/niagara.md (niag_vis): Likewise.
8144 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8146 * ipa-comdats.c: Remove optimize check from gate.
8147 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8148 for functions not optimized.
8149 (ipa_fn_summary_read): Skip optimize check.
8150 (ipa_fn_summary_write): Likewise.
8151 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8153 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8155 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8156 (check_callers): Give up on caller which is not optimized.
8157 (inline_small_functions): Likewise.
8158 (ipa_inline): Do not give up when not optimizing.
8159 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8160 away unoptimizes cdtors.
8161 (whole_program_function_and_variable_visibility): Do
8162 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8163 * ipa.c (process_references): Do not check optimize.
8164 (symbol_table::remove_unreachable_nodes): Update optimize check.
8165 (set_writeonly_bit): Update optimize check.
8166 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8167 (pass_ipa_single_use::gate): Remove.
8169 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8171 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8172 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8173 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8174 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8175 permute_load, permute_store, adjust_extract, adjust_splat,
8176 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8177 replace_swap_with_copy, dump_swap_insn_table,
8178 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8179 recombine_lvx_pattern, recombine_stvx_pattern,
8180 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8181 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8182 to file rs6000-p8swap.c.
8183 * config/rs6000/rs6000-p8swap.c: New file.
8184 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8185 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8186 and rs6000*-*-* targets.
8188 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8190 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8192 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8194 * lto-wrapper.c (merge_and_complain): Do not merge
8195 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8196 fsigned_zeros, ftrapping_math, fwrapv.
8197 (append_compiler_options): Do not track these options.
8198 (append_linker_options): Likewie
8200 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8202 * cgraphunit.c (cgraph_node::finalize_function): When
8203 !flag_toplevel_reorde set no_reorder flag.
8204 (varpool_node::finalize_decl): Likewise.
8205 (symbol_table::compile): Drop no toplevel reorder path.
8207 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8209 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8210 edges; zero probability is not better than uninitialized.
8212 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8214 * asan.h (asan_sanitize_allocas_p): Declare.
8215 * asan.c (asan_sanitize_allocas_p): New function.
8216 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8217 (handle_builtin_alloca): Likewise.
8218 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8219 if !asan_sanitize_allocas_p.
8220 * params.def (asan-instrument-allocas): Add new option.
8221 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8222 * opts.c (common_handle_option): Disable allocas sanitization for
8225 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8227 * asan.c: Include gimple-fold.h.
8228 (get_last_alloca_addr): New function.
8229 (handle_builtin_stackrestore): Likewise.
8230 (handle_builtin_alloca): Likewise.
8231 (asan_emit_allocas_unpoison): Likewise.
8232 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8233 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8234 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8235 (instrument_builtin_call): Pass gimple iterator to
8236 get_mem_refs_of_builtin_call.
8237 (last_alloca_addr): New global.
8238 * asan.h (asan_emit_allocas_unpoison): Declare.
8239 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8240 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8241 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8242 if function calls alloca.
8243 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8244 * gimple-fold.h (replace_call_with_value): Declare.
8245 * internal-fn.c: Include asan.h.
8246 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8247 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8249 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8251 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8252 (C_SELFTEST_FLAGS): New.
8253 (CPP_SELFTEST_FLAGS): New.
8254 (SELFTEST_DEPS): New, from deps of s-selftest.
8255 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8256 (CPP_SELFTEST_DEPS): New.
8257 (selftest): Add dependency on s-selftest-c++.
8258 (s-selftest): Rename to...
8259 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8260 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8261 than SELFTEST_FLAGS.
8262 (selftest-gdb): Rename to...
8263 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8265 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8266 (selftest-valgrind): Rename to...
8267 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8269 (selftest-valgrind): Reintroduce as an alias for
8270 selftest-c-valgrind.
8271 (s-selftest-c++): New.
8272 (selftest-c++-gdb): New.
8273 (selftest-c++-valgrind): New.
8275 2017-07-06 Olivier Hainque <hainque@adacore.com>
8277 * gcc.c (process_command): When deciding if undefined variables
8278 should be ignored when processing specs, accept "gcc -v" as well.
8280 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8282 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8283 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8285 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8287 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8288 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8289 * config/arm/arm-tables.opt: Regenerate.
8290 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8292 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8294 2017-07-06 Carl Love <cel@us.ibm.com>
8296 * ChangeLog: Clean up from mid air collision
8298 2017-07-06 Carl Love <cel@us.ibm.com>
8300 * config/rs6000/rs6000-c.c: Add support for built-in functions
8301 vector signed int vec_subc (vector signed int, vector signed int);
8302 vector signed __int128 vec_subc (vector signed __int128,
8303 vector signed __int128);
8304 vector unsigned __int128 vec_subc (vector unsigned __int128,
8305 vector unsigned __int128);
8306 vector signed int vec_sube (vector signed int, vector signed int,
8308 vector unsigned int vec_sube (vector unsigned int,
8309 vector unsigned int,
8310 vector unsigned int);
8311 vector signed __int128 vec_sube (vector signed __int128,
8312 vector signed __int128,
8313 vector signed__int128);
8314 vector unsigned __int128 vec_sube (vector unsigned __int128,
8315 vector unsigned __int128,
8316 vector unsigned __int128);
8317 vector signed int vec_subec (vector signed int, vector signed int,
8319 vector unsigned int vec_subec (vector unsigned int,
8320 vector unsigned int,
8321 vector unsigned int);
8322 vector signed __int128 vec_subec (vector signed __int128,
8323 vector signed __int128,
8324 vector signed__int128);
8325 vector unsigned __int128 vec_subec (vector unsigned __int128,
8326 vector unsigned __int128,
8327 vector unsigned __int128);
8328 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8329 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8330 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8331 BU_ALTIVEC_OVERLOAD_X definitions.
8332 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8333 * doc/extend.texi: Update the built-in documentation file for the new
8336 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8339 * diagnostic-show-locus.c (layout::layout): Use start and finish
8340 spelling location for the start and finish of each range.
8341 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8342 Add unused aspect param.
8343 * input.c (expand_location_1): Add "aspect" param, and use it
8344 to access the correct part of the location.
8345 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8347 (expand_location_to_spelling_point): Likewise.
8348 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8349 param, and pass it to expand_location_1.
8351 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8353 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8354 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8355 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8356 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8357 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8358 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8359 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8360 _mm_maskz_getmant_ss): New intrinsics.
8361 (__builtin_ia32_getexpss128_mask): Changed to ...
8362 __builtin_ia32_getexpss128_round ... this.
8363 (__builtin_ia32_getexpsd128_mask): Changed to ...
8364 __builtin_ia32_getexpsd128_round ... this.
8365 * config/i386/i386-builtin-types.def
8366 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8367 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8368 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8369 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8370 __builtin_ia32_getmantss_mask_round): New builtins.
8371 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8372 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8373 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8374 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8375 * config/i386/sse.md
8376 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8377 avx512f_sgetexp<mode><mask_scalar_name>
8378 <round_saeonly_scalar_name> ... this.
8379 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8380 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8381 vgetexp<ssescalarmodesuffix>
8382 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8383 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8384 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8385 avx512f_vgetmant<mode><mask_scalar_name>
8386 <round_saeonly_scalar_name> ... this.
8387 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8388 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8389 vgetmant<ssescalarmodesuffix>
8390 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8391 %0<mask_scalar_operand4>, %1, %2
8392 <round_saeonly_scalar_mask_op4>, %3} ... this.
8393 * config/i386/subst.md (mask_scalar_operand4,
8394 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8395 round_saeonly_scalar_nimm_predicate): New subst attributes.
8397 2017-07-06 Julia Koval <julia.koval@intel.com>
8399 * config/i386/i386.c (ix86_erase_embedded_rounding):
8400 Remove code for old rounding pattern.
8402 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8404 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8406 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8408 * doc/sourcebuild.texi (Test Directives, Variants of
8409 dg-require-support): Add documentation for dg-require-stack-check.
8411 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8413 * config/i386/subst.md (mask_scalar, round_scalar,
8414 round_saeonly_scalar): New meta-templates.
8415 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8416 round_scalar_mask_operand3, round_scalar_mask_op3,
8417 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8418 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8419 round_saeonly_scalar_constraint,
8420 round_saeonly_scalar_prefix): New subst attribute.
8421 * config/i386/sse.md
8422 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8423 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8424 <round_scalar_name> ... this.
8425 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8426 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8427 <round_scalar_name> ... this.
8428 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8429 <sse>_vm<code><mode>3<mask_scalar_name>
8430 <round_saeonly_scalar_name> ... this.
8431 (v<plusminus_mnemonic><ssescalarmodesuffix>
8432 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8433 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8434 v<plusminus_mnemonic><ssescalarmodesuffix>
8435 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8436 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8437 (v<multdiv_mnemonic><ssescalarmodesuffix>
8438 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8439 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8440 v<multdiv_mnemonic><ssescalarmodesuffix>
8441 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8442 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8443 (v<maxmin_float><ssescalarmodesuffix>
8444 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8445 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8446 v<maxmin_float><ssescalarmodesuffix>
8447 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8448 %0<mask_scalar_operand3>, %1, %<iptr>2
8449 <round_saeonly_scalar_mask_op3>} ... this.
8451 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8453 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8454 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8456 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8457 Alan Hayward <alan.hayward@arm.com>
8458 David Sherwood <david.sherwood@arm.com>
8460 * combine.c (simplify_if_then_else): Remove "enum" before
8462 * compare-elim.c (can_eliminate_compare): Likewise.
8463 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8465 (aarch64_lookup_simd_builtin_type): Likewise.
8466 (aarch64_simd_builtin_type): Likewise.
8467 (aarch64_init_simd_builtin_types): Likewise.
8468 (aarch64_simd_expand_args): Likewise.
8469 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8471 (aarch64_reverse_mask): Likewise.
8472 (aarch64_simd_emit_reg_reg_move): Likewise.
8473 (aarch64_gen_adjusted_ldpstp): Likewise.
8474 (aarch64_ccmp_mode_to_code): Likewise.
8475 (aarch64_operands_ok_for_ldpstp): Likewise.
8476 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8477 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8479 (aarch64_min_divisions_for_recip_mul): Likewise.
8480 (aarch64_reassociation_width): Likewise.
8481 (aarch64_get_condition_code_1): Likewise.
8482 (aarch64_simd_emit_reg_reg_move): Likewise.
8483 (aarch64_simd_attr_length_rglist): Likewise.
8484 (aarch64_reverse_mask): Likewise.
8485 (aarch64_operands_ok_for_ldpstp): Likewise.
8486 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8487 (aarch64_gen_adjusted_ldpstp): Likewise.
8488 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8490 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8491 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8492 (arm_lookup_simd_builtin_type): Likewise.
8493 (arm_simd_builtin_type): Likewise.
8494 (arm_init_simd_builtin_types): Likewise.
8495 (arm_expand_builtin_args): Likewise.
8496 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8497 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8498 (ft32_setup_incoming_varargs): Likewise.
8499 (ft32_function_arg): Likewise.
8500 (ft32_function_arg_advance): Likewise.
8501 (ft32_pass_by_reference): Likewise.
8502 (ft32_arg_partial_bytes): Likewise.
8503 (ft32_valid_pointer_mode): Likewise.
8504 (ft32_addr_space_pointer_mode): Likewise.
8505 (ft32_addr_space_legitimate_address_p): Likewise.
8506 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8508 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8509 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8510 (iamcu_alignment): Likewise.
8511 (canonicalize_vector_int_perm): Likewise.
8512 (ix86_noce_conversion_profitable_p): Likewise.
8513 (ix86_mpx_bound_mode): Likewise.
8514 (ix86_operands_ok_for_move_multiple): Likewise.
8515 * config/microblaze/microblaze-protos.h
8516 (microblaze_expand_conditional_branch_reg): Likewise.
8517 * config/microblaze/microblaze.c
8518 (microblaze_expand_conditional_branch_reg): Likewise.
8519 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8521 (rs6000_reassociation_width): Likewise.
8522 (rs6000_invalid_binary_op): Likewise.
8523 (fusion_p9_p): Likewise.
8524 (emit_fusion_p9_load): Likewise.
8525 (emit_fusion_p9_store): Likewise.
8526 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8528 (riscv_hard_regno_mode_ok_p): Likewise.
8529 (riscv_address_insns): Likewise.
8530 (riscv_split_symbol): Likewise.
8531 (riscv_legitimize_move): Likewise.
8532 (riscv_function_value): Likewise.
8533 (riscv_hard_regno_nregs): Likewise.
8534 (riscv_expand_builtin): Likewise.
8535 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8536 (riscv_build_integer): Likewise.
8537 (riscv_split_integer): Likewise.
8538 (riscv_legitimate_constant_p): Likewise.
8539 (riscv_cannot_force_const_mem): Likewise.
8540 (riscv_regno_mode_ok_for_base_p): Likewise.
8541 (riscv_valid_base_register_p): Likewise.
8542 (riscv_valid_offset_p): Likewise.
8543 (riscv_valid_lo_sum_p): Likewise.
8544 (riscv_classify_address): Likewise.
8545 (riscv_legitimate_address_p): Likewise.
8546 (riscv_address_insns): Likewise.
8547 (riscv_load_store_insns): Likewise.
8548 (riscv_force_binary): Likewise.
8549 (riscv_split_symbol): Likewise.
8550 (riscv_force_address): Likewise.
8551 (riscv_legitimize_address): Likewise.
8552 (riscv_move_integer): Likewise.
8553 (riscv_legitimize_const_move): Likewise.
8554 (riscv_legitimize_move): Likewise.
8555 (riscv_address_cost): Likewise.
8556 (riscv_subword): Likewise.
8557 (riscv_output_move): Likewise.
8558 (riscv_canonicalize_int_order_test): Likewise.
8559 (riscv_emit_int_order_test): Likewise.
8560 (riscv_function_arg_boundary): Likewise.
8561 (riscv_pass_mode_in_fpr_p): Likewise.
8562 (riscv_pass_fpr_single): Likewise.
8563 (riscv_pass_fpr_pair): Likewise.
8564 (riscv_get_arg_info): Likewise.
8565 (riscv_function_arg): Likewise.
8566 (riscv_function_arg_advance): Likewise.
8567 (riscv_arg_partial_bytes): Likewise.
8568 (riscv_function_value): Likewise.
8569 (riscv_pass_by_reference): Likewise.
8570 (riscv_setup_incoming_varargs): Likewise.
8571 (riscv_print_operand): Likewise.
8572 (riscv_elf_select_rtx_section): Likewise.
8573 (riscv_save_restore_reg): Likewise.
8574 (riscv_for_each_saved_reg): Likewise.
8575 (riscv_register_move_cost): Likewise.
8576 (riscv_hard_regno_mode_ok_p): Likewise.
8577 (riscv_hard_regno_nregs): Likewise.
8578 (riscv_class_max_nregs): Likewise.
8579 (riscv_memory_move_cost): Likewise.
8580 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8581 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8582 (rl78_addr_space_address_mode): Likewise.
8583 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8585 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8586 (rs6000_reassociation_width): Likewise.
8587 (rs6000_invalid_binary_op): Likewise.
8588 (fusion_p9_p): Likewise.
8589 (emit_fusion_p9_load): Likewise.
8590 (emit_fusion_p9_store): Likewise.
8591 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8592 (ok_for_simple_move_operands): Likewise.
8593 (ok_for_simple_move_strict_operands): Likewise.
8594 (ok_for_simple_arith_logic_operands): Likewise.
8595 (visium_legitimize_reload_address): Likewise.
8596 (visium_select_cc_mode): Likewise.
8597 (output_cbranch): Likewise.
8598 (visium_split_double_move): Likewise.
8599 (visium_expand_copysign): Likewise.
8600 (visium_expand_int_cstore): Likewise.
8601 (visium_expand_fp_cstore): Likewise.
8602 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8603 (visium_function_arg): Likewise.
8604 (visium_function_arg_advance): Likewise.
8605 (visium_libcall_value): Likewise.
8606 (visium_setup_incoming_varargs): Likewise.
8607 (visium_legitimate_constant_p): Likewise.
8608 (visium_legitimate_address_p): Likewise.
8609 (visium_legitimize_address): Likewise.
8610 (visium_secondary_reload): Likewise.
8611 (visium_register_move_cost): Likewise.
8612 (visium_memory_move_cost): Likewise.
8613 (prepare_move_operands): Likewise.
8614 (ok_for_simple_move_operands): Likewise.
8615 (ok_for_simple_move_strict_operands): Likewise.
8616 (ok_for_simple_arith_logic_operands): Likewise.
8617 (visium_function_value_1): Likewise.
8618 (rtx_ok_for_offset_p): Likewise.
8619 (visium_legitimize_reload_address): Likewise.
8620 (visium_split_double_move): Likewise.
8621 (visium_expand_copysign): Likewise.
8622 (visium_expand_int_cstore): Likewise.
8623 (visium_expand_fp_cstore): Likewise.
8624 (visium_split_cstore): Likewise.
8625 (visium_select_cc_mode): Likewise.
8626 (visium_split_cbranch): Likewise.
8627 (output_cbranch): Likewise.
8628 (visium_print_operand_address): Likewise.
8629 * expmed.c (flip_storage_order): Likewise.
8630 * expmed.h (emit_cstore): Likewise.
8631 (flip_storage_order): Likewise.
8632 * genrecog.c (validate_pattern): Likewise.
8633 * hsa-gen.c (gen_hsa_addr): Likewise.
8634 * internal-fn.c (expand_arith_overflow): Likewise.
8635 * ira-color.c (allocno_copy_cost_saving): Likewise.
8636 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8637 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8638 (process_invariant_for_inheritance): Likewise.
8639 * lra-eliminations.c (move_plus_up): Likewise.
8640 * omp-low.c (lower_oacc_reductions): Likewise.
8641 * simplify-rtx.c (simplify_subreg): Likewise.
8642 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8643 (TARGET_CHKP_BOUND_MODE): Likewise..
8644 * targhooks.c (default_chkp_bound_mode): Likewise.
8645 (default_setup_incoming_vararg_bounds): Likewise.
8646 * targhooks.h (default_chkp_bound_mode): Likewise.
8647 (default_setup_incoming_vararg_bounds): Likewise.
8648 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8649 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8650 (have_whole_vector_shift): Likewise.
8651 * tree-vect-stmts.c (vectorizable_load): Likewise.
8652 * doc/tm.texi: Regenerate.
8654 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8656 Graceful degrade if Binutils PR21472 is not available.
8659 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8660 .rodata in flash test fails.
8661 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8662 * confgure: Regenerate.
8663 * config.in: Regenerate.
8664 * config/avr/avr.c (avr_asm_named_section)
8665 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8666 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8667 (avr_asm_init_sections): Same.
8669 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8671 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8672 (fma<VH:mode>4_intrinsic): Likewise.
8673 (*fmsub<VCVTF:mode>4): Likewise.
8674 (*fmsub<VH:mode>4_intrinsic): Likewise.
8676 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8679 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8680 Don't depend on "optimize > 0".
8681 (out_movhi_r_mr, out_movqi_mr_r): Same.
8682 (out_movhi_mr_r, out_movqi_r_mr): Same.
8683 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8684 io_address_operand on "optimize > 0".
8686 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8688 * tree-loop-distribution.c: Add general explanantion on the pass.
8689 (generate_loops_for_partition): Mark distributed loop.
8690 (pg_add_dependence_edges): New parameter. Handle alias data
8691 dependence specially and record it in the parameter if asked.
8692 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8693 (init_partition_graph_vertices, add_partition_graph_edge): New.
8694 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8695 (free_partition_graph_vdata, build_partition_graph): New.
8696 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8697 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8698 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8699 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8700 (version_for_distribution_p, finalize_partitions): New.
8701 (distribute_loop): Handle alias data dependence specially. Factor
8702 out loop fusion code as functions and call these functions.
8704 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8706 * tree-loop-distribution.c (classify_partition): New parameter and
8707 better handle reduction statement.
8708 (rdg_build_partitions): Revise comment.
8709 (distribute_loop): Compute statements in all partitions and pass it
8710 to classify_partition.
8712 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8714 * tree-loop-distribution.c (enum partition_type): New.
8715 (struct partition): New field type.
8716 (partition_merge_into): Add parameter. Update partition type.
8717 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8718 (build_rdg_partition_for_vertex): Compute partition type.
8719 (rdg_build_partitions): Dump partition type.
8720 (distribute_loop): Update calls to partition_merge_into.
8722 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8724 * tree-loop-distribution.c (struct ddr_hasher): New.
8725 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8727 (classify_partition): Call get_data_dependence.
8728 (pg_add_dependence_edges): Ditto.
8729 (distribute_loop): Release data dependence hash table.
8731 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8733 * tree-loop-distribution.c (ref_base_address): Delete.
8734 (similar_memory_accesses): Rename ...
8735 (share_memory_accesses): ... to this. Check if partitions access
8736 the same memory reference.
8737 (distribute_loop): Call share_memory_accesses.
8739 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8741 * tree-loop-distribution.c (struct partition): New field recording
8743 (partition_alloc, partition_free): Init and release data refs.
8744 (partition_merge_into): Merge data refs.
8745 (build_rdg_partition_for_vertex): Collect data refs for partition.
8746 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8748 (distribute_loop): Remve data refs from vertice data of partition
8751 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8753 * tree-loop-distribution.c (params.h): Include header file.
8754 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8755 (datarefs_vec): New global var.
8756 (create_rdg_vertices): Use datarefs_vec directly.
8757 (free_rdg): Don't free data references.
8758 (build_rdg): Update use. Don't free data references.
8759 (distribute_loop): Compute global variable for data references.
8760 Bail out if there are too many data references.
8762 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8764 * tree-loop-distribution.c (loop_nest): New global var.
8765 (build_rdg): Use loop directly, rather than loop nest.
8766 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8768 (distribute_loop): Compute global variable loop nest. Update use.
8770 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8772 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8773 (partition_merge_into): New parameter. Dump reason for fusion.
8774 (distribute_loop): Update use of partition_merge_into.
8776 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8778 * tree-loop-distribution.c (bb_top_order_index): New.
8779 (bb_top_order_index_size, bb_top_order_cmp): New.
8780 (stmts_from_loop): Use topological order.
8781 (pass_loop_distribution::execute): Compute and release topological
8782 order for basic blocks.
8784 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8786 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8789 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8791 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
8792 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
8793 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
8794 * internal-fn.def (LOOP_DIST_ALIAS): New.
8795 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
8796 (fold_loop_internal_call): ... this.
8797 (vect_loop_dist_alias_call): New function.
8798 (set_uid_loop_bbs): Call fold_loop_internal_call.
8799 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
8802 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8805 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
8806 Require dead FLAGS_REG at the beginning of a peephole.
8808 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8811 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
8812 arguments in the call to __builtin_ia32_sbb_u32.
8813 (_subborrow_u64): Swap _X and _Y arguments in the call to
8814 __builtin_ia32_sbb_u64.
8816 2017-07-04 Jakub Jelinek <jakub@redhat.com>
8819 * tree-vrp.c (compare_assert_loc): Turn into a function template
8820 with stable template parameter. Only test if a->e is NULL,
8821 !a->e == !b->e has been verified already. Use e == NULL or
8822 e != NULL instead of e or ! e tests. If stable is true, don't use
8823 iterative_hash_expr, on the other side allow a or b or both NULL
8824 and sort the NULLs last.
8825 (process_assert_insertions): Sort using compare_assert_loc<false>
8826 instead of compare_assert_loc, later sort using
8827 compare_assert_loc<true> before calling process_assert_insertions_for
8828 in a loop. Use break instead of continue once seen NULL pointer.
8830 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8832 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8833 Cortex-R7 and Cortex-R8 processors.
8835 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8837 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8838 uninitialized while src is not.
8840 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8842 * common/config/arm/arm-common.c: Adjust include path for
8844 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8845 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8846 (arm-cpu-data.h): Likewise.
8847 (arm-cpu-cdata.h): Likewise.
8848 * config/arm/arm-cpu.h: Delete.
8849 * config/arm/arm-cpu-cdata.h: Delete.
8850 * config/arm/arm-cpu-data.h: Delete.
8852 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8854 * config/arm/arm-cpus.in (cortex-a55): New.
8855 (cortex-a75): Likewise.
8856 (cortex-a75.cortex-a55): Likewise.
8857 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8859 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8860 * config/arm/arm-cpu-cdata.h: Regenerate.
8861 * config/arm/arm-cpu-data.h: Regenerate.
8862 * config/arm/arm-cpu.h: Regenerate.
8863 * config/arm/arm-tables.opt: Regenerate.
8864 * config/arm/arm-tune.md: Regenerate.
8866 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8868 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8870 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8872 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8875 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8877 PR tree-optimization/81292
8878 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8879 full_string_p, also call adjust_related_strinfos if the adjustment
8880 is simple, otherwise invalidate related strinfos.
8882 2017-07-04 Martin Liska <mliska@suse.cz>
8885 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8886 newly created variable as DECL_IGNORED_P.
8888 2017-07-04 Martin Liska <mliska@suse.cz>
8891 * ipa-inline.c (inline_small_functions):
8892 Use xstrdup_for_dump.
8894 2017-07-04 Tom de Vries <tom@codesourcery.com>
8896 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8898 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8901 * config/darwin.c (darwin_function_switched_text_sections):
8904 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8906 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8908 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8910 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8912 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8914 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8915 min_profitable_iters, and th as inclusive lower bounds.
8916 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8917 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8918 for min_profitable_iters and min_profitable_estimate.
8919 (vect_transform_loop): Treat th as an inclusive lower bound.
8920 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8922 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8925 * config/darwin.c (darwin_function_switched_text_sections):
8926 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8927 in two pieces, and suppress the use of buf.
8929 2017-07-03 Nathan Sidwell <nathan@acm.org>
8931 * hash-table.h (hash_table_mod1): Fix indentation.
8933 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8936 * predict.c (force_edge_cold): Be more careful about propagation
8938 * profile-count.h (profile_probability::guessed,
8939 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8941 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8943 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8945 * doc/invoke.texi (rcpc architecture extension): Document it.
8947 2017-07-03 Richard Biener <rguenther@suse.de>
8949 PR tree-optimization/60510
8950 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8951 the scalar reduction PHI and use it.
8952 (vectorizable_reduction): Properly guard the single_defuse_cycle
8953 path for non-SLP reduction chains where we cannot use it.
8954 Rework reduc_def/index and vector type deduction. Rework
8955 vector operand gathering during reduction op code-gen.
8956 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8957 chains dissolve the chain and leave it to non-SLP reduction
8960 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8962 * tree-data-ref.h (dr_alignment): Declare.
8963 * tree-data-ref.c (dr_alignment): New function.
8964 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8965 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8967 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8969 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8971 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8972 and base_misalignment fields.
8973 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8974 * tree-data-ref.c: Include builtins.h.
8975 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8976 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8977 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8978 * tree-vect-data-refs.c: Include tree-cfg.h.
8979 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8980 fields instead of calculating an alignment here.
8981 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8982 innermost_loop_behavior fields.
8984 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8986 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8988 (DR_STEP_ALIGNMENT): New macro.
8989 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8990 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8991 (create_data_ref): Print it.
8992 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8993 to tell whether the step preserves vector (mis)alignment.
8994 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8995 Move the check for an integer step and generalise to all INTEGER_CST.
8996 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8997 Print the outer step alignment.
8999 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9001 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9002 with offset_alignment.
9003 (DR_ALIGNED_TO): Delete.
9004 (DR_OFFSET_ALIGNMENT): New macro.
9005 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9006 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9007 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9008 (create_data_ref): Likewise.
9009 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9010 (vect_analyze_data_refs): Likewise.
9011 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9012 creating dummy innermost behavior.
9014 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9016 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9017 with a "innermost_loop_behavior *" and refeence tree.
9018 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9019 (create_data_ref): Update call accordingly.
9020 * tree-predcom.c (find_looparound_phi): Likewise.
9022 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9024 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9025 fields with dr_wrt_vec_loop.
9026 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9027 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9028 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9029 (vect_dr_behavior): New function.
9030 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9031 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9032 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9033 track whether the step preserves the misalignment.
9034 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9035 Use vect_dr_behavior.
9036 (vect_setup_realignment): Update call accordingly.
9037 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9038 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9039 call to vect_create_addr_base_for_vector_ref.
9040 (vect_create_cond_for_align_checks): Likewise.
9041 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9042 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9043 (vect_recog_mask_conversion_pattern): Likewise.
9044 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9045 (new_stmt_vec_info): Remove redundant zeroing.
9047 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9049 * common/config/arm/arm-common.c (arm_be8_option): New function.
9050 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9051 (ISA_ARMv6): Add isa_bit_be8.
9052 * config/arm/arm.h (arm_be8_option): Add prototype.
9053 (BE8_SPEC_FUNCTION): New define.
9054 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9055 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9056 (mlittle-endian): Similarly.
9057 (mbe8, mbe32): New options.
9058 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9059 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9061 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9063 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9065 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9067 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9068 (cleanup_tree_cfg_bb): Use it.
9069 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9071 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9073 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9076 * loop-doloop.c (add_test): Update profile.
9078 2017-07-03 Martin Liska <mliska@suse.cz>
9081 * sanopt.c (rewrite_usage_of_param): New function.
9082 (sanitize_rewrite_addressable_params): Likewise.
9083 (pass_sanopt::execute): Call rewrite_usage_of_param.
9085 2017-07-03 Richard Biener <rguenther@suse.de>
9087 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9088 back to using VIEW_CONVERT_EXPR.
9090 2017-07-03 Martin Liska <mliska@suse.cz>
9093 * doc/extend.texi: Document when a resolver function is
9094 generated for target_clones.
9096 2017-07-03 Martin Liska <mliska@suse.cz>
9098 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9099 and shadow memory of auto variables which are subject of
9100 use-after-scope sanitization.
9101 (asan_expand_mark_ifn): Add do set only when is_poison.
9103 2016-07-03 Richard Biener <rguenther@suse.de>
9105 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9107 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9108 (vectorizable_reduction): Perform reduction PHI creation when
9109 visiting a reduction PHI and adjust and simplify code generation
9110 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9111 (vect_transform_loop): Visit reduction PHIs.
9112 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9113 defs into the SLP tree.
9114 (vect_build_slp_tree): Reduction defs terminate the recursion.
9115 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9117 (vect_get_vec_defs_for_stmt_copy): Export.
9118 (vect_get_vec_defs): Likewise.
9119 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9121 (vect_get_vec_defs_for_stmt_copy): Declare.
9122 (vect_get_vec_defs): Likewise.
9124 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9126 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9127 parameter with a "loop" parameter and use it instead of the
9128 loop containing DR_STMT. Don't check simple_iv when doing
9129 BB analysis. Describe the two analysis modes in the comment.
9131 2017-07-03 Tom de Vries <tom@codesourcery.com>
9133 PR tree-optimization/69468
9134 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9135 (find_same_succ_bb): Handle ignore_edge_flags.
9137 2017-07-03 Tom de Vries <tom@codesourcery.com>
9139 PR tree-optimization/81192
9140 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9142 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9144 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9146 2017-07-03 Tom de Vries <tom@codesourcery.com>
9148 PR tree-optimization/81192
9149 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9150 BB_SAME_SUCC (bb) == NULL.
9152 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9154 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9157 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9159 * dumpfile.c: Include profile-count.h
9160 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9162 (insert_cond_bb): Update profile.
9163 * tree-cfg.h (insert_cond_bb): Update prototype.
9164 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9165 * tree-dump.c: Do not include tree-cfg.
9167 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9169 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9171 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9173 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9176 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9178 * tree-complex.c (expand_complex_div_wide): update profile.
9180 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9181 Alan Hayward <alan.hayward@arm.com>
9182 David Sherwood <david.sherwood@arm.com>
9184 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9185 (CORETYPES_H): New define.
9186 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9187 (insn-modes-inline.h, s-modes-inline-h): New rules.
9188 (generated_files): Add insn-modes-inline.h.
9189 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9190 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9191 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9192 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9193 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9194 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9195 (build/gencodes.o, build/genconditions.o): Likewise.
9196 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9197 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9198 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9199 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9200 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9201 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9202 * coretypes.h: Include everything up to real.h for generators.
9203 Include insn-modes.h first. Include wide-int-print.h after
9204 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9205 * machmode.h: Don't include insn-modes.h here.
9206 * function-tests.c: Remove includes of signop.h, machmode.h,
9207 double-int.h and wide-int.h.
9209 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9211 * optc-save-gen.awk: Likewise.
9212 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9213 * godump.c: Remove include of wide-int-print.h.
9214 * pretty-print.h: Likewise.
9215 * wide-int-print.cc: Likewise.
9216 * wide-int.cc: Likewise.
9217 * hash-map-tests.c: Remove include of signop.h.
9218 * hash-set-tests.c: Likewise.
9219 * rtl-tests.c: Likewise.
9220 * mkconfig.sh: Remove include of machmode.h.
9221 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9223 (emit_insn_modes_inline_h): ...this new function. Emit the code
9224 into an insn-modes-inline.h header file, adding appropriate
9225 include guards and end comments.
9226 (emit_insn_modes_c_header): Remove include of machmode.h.
9227 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9229 (main): Handle -i flag and call emit_insn_modes_inline_h when
9232 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9234 * tree-ssa-strlen.c (strinfo): Rename the length field to
9235 nonzero_chars. Add a full_string_p field.
9236 (compare_nonzero_chars, zero_length_string_p): New functions.
9237 (get_addr_stridx): Add an offset_out parameter.
9238 Use compare_nonzero_chars.
9239 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9240 (new_strinfo): Update after above changes to strinfo.
9241 (set_endptr_and_length): Set full_string_p.
9242 (get_string_length): Update after above changes to strinfo.
9243 (unshare_strinfo): Update call to new_strinfo.
9244 (maybe_invalidate): Likewise.
9245 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9246 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9247 as a uhwi instead of an shwi. Update after above changes to
9248 strinfo and new_strinfo.
9249 (zero_length_string): Assert that chainsi contains full strings.
9250 Use zero_length_string_p. Update call to new_strinfo.
9251 (adjust_related_strinfos): Update after above changes to strinfo.
9252 Copy full_string_p from origsi.
9253 (adjust_last_stmt): Use zero_length_string_p.
9254 (handle_builtin_strlen): Update after above changes to strinfo and
9255 new_strinfo. Install the lhs as the string length if the previous
9256 entry didn't describe a full string.
9257 (handle_builtin_strchr): Update after above changes to strinfo
9259 (handle_builtin_strcpy): Likewise.
9260 (handle_builtin_strcat): Likewise.
9261 (handle_builtin_malloc): Likewise.
9262 (handle_pointer_plus): Likewise.
9263 (handle_builtin_memcpy): Likewise. Track nonzero characters
9264 that aren't necessarily followed by a nul terminator.
9265 (handle_char_store): Likewise.
9267 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9269 PR tree-optimization/80769
9270 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9271 for malloc and calloc. Document the new invariant that all related
9272 strinfos have delayed lengths or none do.
9273 (verify_related_strinfos): Move earlier in file.
9274 (set_endptr_and_length): New function, split out from...
9275 (get_string_length): ...here. Also set the lengths of related
9277 (zero_length_string): Assert that chainsi has known (rather than
9279 (adjust_related_strinfos): Likewise.
9281 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9283 PR tree-optimization/81136
9284 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9285 assert that two references with the same misalignment have the same
9286 compile-time misalignment if those compile-time misalignments
9289 2017-07-01 Andi Kleen <ak@linux.intel.com>
9291 * print-tree.c (print_node): Print all attributes.
9293 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9295 * cfg.c (scale_bbs_frequencies): New function.
9296 * cfg.h (scale_bbs_frequencies): Declare it.
9297 * cfgloopanal.c (single_likely_exit): Cleanup.
9298 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9300 (scale_loop_profile): Likewise.
9301 (loop_version): Likewise.
9302 (create_empty_loop_on_edge): Update.
9303 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9304 scale_loop_frequencies, scale_loop_profile, loopify,
9305 loop_version): Update prototypes.
9306 * modulo-sched.c (sms_schedule): Update.
9307 * predict.c (unlikely_executed_edge_p): Also check probability.
9308 (probably_never_executed_edge_p): Fix typo.
9309 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9310 * tree-parloops.c (gen_parallel_loop): Update.
9311 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9312 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9313 * tree-ssa-loop-split.c (split_loop): Update.
9314 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9315 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9316 (vect_loop_versioning): Update.
9317 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9319 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9321 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9323 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9325 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9326 to keep profile consistent.
9328 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9330 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9331 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9332 * profile-count.h (max_safe_multiplier): Make unsigned.
9333 (profile_count::guessed_zero): New.
9335 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9337 * bb-reorder.c (fix_up_crossing_landing_pad,
9338 fix_crossing_conditional_branches): Use make_single_succ_edge
9339 to keep profile consistent.
9341 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9343 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9346 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9349 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9350 the right scopes, make sure cond_jump isn't preserved between multiple
9351 iterations. Search for fallthru edge whenever there are 3+ edges and
9352 use find_fallthru_edge for it.
9354 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9356 Patch by Alexander Monakov <amonakov@ispras.ru>
9357 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9358 probabilities consistently.
9360 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9362 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9363 emit_cmp_and_jump_insns.
9365 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9368 * tree-inline.c (expand_call_inline): Combine profile statuses.
9370 2017-06-30 Andrew Pinski <apinski@cavium.com>
9372 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9373 fold_stmt returned true.
9375 2017-06-30 Nathan Sidwell <nathan@acm.org>
9377 * ggc.h (empty_string): Delete.
9378 * cfgexpand.c (expand_asm_stmt): Use plain "".
9379 * optabs.c (expand_asm_memory_barrier): Likewise.
9380 * stringpool.c (empty_string): Delete.
9381 (digit_vector, digit_string): Delete.
9382 (ggc_alloc_string): Use plain "", don't optimize single digit
9383 strings. Use ggc_alloc_atomic.
9385 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9387 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9388 comparison set and one other set, use the cost of the non-comparison
9391 2017-06-30 Nathan Sidwell <nathan@acm.org>
9393 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9396 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9398 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9399 loops. Remove now unneeded calls to gimple_switch_set_label() that
9400 just set removed labels to NULL_TREE.
9402 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9404 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9405 (set_range_info): ...here. Only call set_range_info_raw if domain
9407 (set_nonzero_bits): Call set_range_info_raw.
9408 * tree-ssanames.h (set_range_info_raw): New.
9410 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9413 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9414 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9415 of nonimmediate_operand and <store_mask_constraint> instead of m
9416 for the input operand. For V8FI iterator, always split if input
9417 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9418 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9419 <store_mask_predicate> instead of register_operand and
9420 <store_mask_constraint> instead of v for the input operand. Make
9421 sure both operands aren't MEMs for if not <mask_applied>.
9423 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9425 * lto-wrapper.c (copy_file) Close both file descriptors before
9428 2017-06-30 Martin Liska <mliska@suse.cz>
9431 * multiple_target.c (create_dispatcher_calls): Make ifunc
9432 also for function that don't have calls or are not referenced.
9434 2017-06-30 Richard Biener <rguenther@suse.de>
9436 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9437 analyze the first scalar stmt. Move vector type computation
9438 for the BB case here from ...
9439 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9440 live operation processing in the SLP case properly.
9442 2017-06-30 Richard Biener <rguenther@suse.de>
9444 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9446 2017-06-30 Martin Liska <mliska@suse.cz>
9449 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9450 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9452 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9454 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9457 2017-06-30 Richard Biener <rguenther@suse.de>
9459 PR tree-optimization/81249
9460 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9461 condition reduction result to original scalar type.
9463 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9465 * profile-count.h (enum profile_quality): Fix typos and whitespace
9468 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9470 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9471 type for branch probabilities.
9473 2017-06-29 Julian Brown <julian@codesourcery.com>
9474 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9476 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9477 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9478 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9479 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9481 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9483 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9484 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9485 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9486 CC usage from generic code to here.
9487 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9488 CC usage into the target macros.
9490 2017-06-29 Maya Rashish <coypu@sdf.org>
9492 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9495 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9497 * arm/arm-builtins.c: Include profile-count.h
9498 * except.c (sjlj_emit_function_enter): Use
9499 profile_probability::unlikely.
9501 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9503 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9504 and tocrel_offset be pointer args rather than implicitly using
9506 (legitimate_constant_pool_address_p, rs6000_emit_move,
9507 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9508 tocrel_offset and use in toc_relative_expr_p call.
9509 (print_operand, print_operand_address): Use static tocrel_base_oac
9510 and tocrel_offset_oac.
9511 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9514 2017-06-29 Maya Rashish <coypu@sdf.org>
9516 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9518 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9520 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9521 objects, take into account only the alignment of 'op0' and 'mode1' if
9524 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9526 * ccmp.c (ccmp_tree_comparison_p): New function.
9527 (ccmp_candidate_p): Update to use above function.
9528 (get_compare_parts): New function.
9529 (expand_ccmp_next): Update to use new functions.
9530 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9532 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9533 take mode as argument.
9534 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9535 * expr.c (expand_expr_real_1): Pass mode as argument.
9537 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9539 * combine.c (combine_instructions): Print insns to dump_file, together
9542 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9544 * asan.c (asan_emit_stack_protection): Update.
9545 (create_cond_insert_point): Update.
9546 * auto-profile.c (afdo_propagate_circuit): Update.
9547 * basic-block.h (struct edge_def): Turn probability to
9548 profile_probability.
9549 (EDGE_FREQUENCY): Update.
9550 * bb-reorder.c (find_traces_1_round): Update.
9551 (better_edge_p): Update.
9552 (sanitize_hot_paths): Update.
9553 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9554 (make_single_succ_edge): Update.
9555 (check_bb_profile): Update.
9556 (dump_edge_info): Update.
9557 (update_bb_profile_for_threading): Update.
9558 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9559 probabilitycount to 0.
9560 * cfgbuild.c (compute_outgoing_frequencies): Update.
9561 * cfgcleanup.c (try_forward_edges): Update.
9562 (outgoing_edges_match): Update.
9563 (try_crossjump_to_edge): Update.
9564 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9565 (expand_gimple_tailcall): Update.
9566 (construct_init_block): Use make_single_succ_edge.
9567 (construct_exit_block): Use make_single_succ_edge.
9568 * cfghooks.c (verify_flow_info): Update.
9569 (redirect_edge_succ_nodup): Update.
9570 (split_edge): Update.
9571 (account_profile_record): Update.
9572 * cfgloopanal.c (single_likely_exit): Update.
9573 * cfgloopmanip.c (scale_loop_profile): Update.
9574 (set_zero_probability): Remove.
9575 (duplicate_loop_to_header_edge): Update.
9576 * cfgloopmanip.h (loop_version): Update prototype.
9577 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9578 (force_nonfallthru_and_redirect): Update.
9579 (update_br_prob_note): Update.
9580 (rtl_verify_edges): Update.
9581 (purge_dead_edges): Update.
9582 (rtl_lv_add_condition_to_bb): Update.
9583 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9584 * cgraphunit.c (init_lowered_empty_function): Update.
9585 (cgraph_node::expand_thunk): Update.
9586 * cilk-common.c: Include profile-count.h
9587 * dojump.c (inv): Remove.
9588 (jumpifnot): Update.
9589 (jumpifnot_1): Update.
9590 (do_jump_1): Update.
9592 (do_jump_by_parts_greater_rtx): Update.
9593 (do_compare_rtx_and_jump): Update.
9594 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9595 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9596 * dwarf2cfi.c: Include profile-count.h
9597 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9598 (sjlj_emit_dispatch_table): Likewise.
9599 * explow.c: Include profile-count.h
9600 * expmed.c (emit_store_flag_force): Update.
9601 (do_cmp_and_jump): Update.
9602 * expr.c (compare_by_pieces_d::generate): Update.
9603 (compare_by_pieces_d::finish_mode): Update.
9604 (emit_block_move_via_loop): Update.
9605 (store_expr_with_bounds): Update.
9606 (store_constructor): Update.
9607 (expand_expr_real_2): Update.
9608 (expand_expr_real_1): Update.
9609 * expr.h (try_casesi, try_tablejump): Update prototypes.
9610 * gimple-pretty-print.c (dump_probability): Update.
9611 (dump_profile): New.
9612 (dump_gimple_label): Update.
9613 (dump_gimple_bb_header): Update.
9614 * graph.c (draw_cfg_node_succ_edges): Update.
9615 * hsa-gen.c (convert_switch_statements): Update.
9616 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9617 (find_if_case_1): Update.
9618 (find_if_case_2): Update.
9619 * internal-fn.c (expand_arith_overflow_result_store): Update.
9620 (expand_addsub_overflow): Update.
9621 (expand_neg_overflow): Update.
9622 (expand_mul_overflow): Update.
9623 (expand_vector_ubsan_overflow): Update.
9624 * ipa-cp.c (good_cloning_opportunity_p): Update.
9625 * ipa-split.c (split_function): Use make_single_succ_edge.
9626 * ipa-utils.c (ipa_merge_profiles): Update.
9627 * loop-doloop.c (add_test): Update.
9628 (doloop_modify): Update.
9629 * loop-unroll.c (compare_and_jump_seq): Update.
9630 (unroll_loop_runtime_iterations): Update.
9631 * lra-constraints.c (lra_inheritance): Update.
9632 * lto-streamer-in.c (input_cfg): Update.
9633 * lto-streamer-out.c (output_cfg): Update.
9634 * mcf.c (adjust_cfg_counts): Update.
9635 * modulo-sched.c (sms_schedule): Update.
9636 * omp-expand.c (expand_omp_for_init_counts): Update.
9637 (extract_omp_for_update_vars): Update.
9638 (expand_omp_ordered_sink): Update.
9639 (expand_omp_for_ordered_loops): Update.
9640 (expand_omp_for_generic): Update.
9641 (expand_omp_for_static_nochunk): Update.
9642 (expand_omp_for_static_chunk): Update.
9643 (expand_cilk_for): Update.
9644 (expand_omp_simd): Update.
9645 (expand_omp_taskloop_for_outer): Update.
9646 (expand_omp_taskloop_for_inner): Update.
9647 * omp-simd-clone.c (simd_clone_adjust): Update.
9648 * optabs.c (expand_doubleword_shift): Update.
9649 (expand_abs): Update.
9650 (emit_cmp_and_jump_insn_1): Update.
9651 (expand_compare_and_swap_loop): Update.
9652 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9653 * predict.c (predictable_edge_p): Update.
9654 (edge_probability_reliable_p): Update.
9655 (set_even_probabilities): Update.
9656 (combine_predictions_for_insn): Update.
9657 (combine_predictions_for_bb): Update.
9658 (propagate_freq): Update.
9659 (estimate_bb_frequencies): Update.
9660 (force_edge_cold): Update.
9661 * profile-count.c (profile_count::dump): Add missing space into dump.
9662 (profile_count::debug): Add newline.
9663 (profile_count::differs_from_p): Explicitly convert to unsigned.
9664 (profile_count::stream_in): Update.
9665 (profile_probability::dump): New member function.
9666 (profile_probability::debug): New member function.
9667 (profile_probability::differs_from_p): New member function.
9668 (profile_probability::differs_lot_from_p): New member function.
9669 (profile_probability::stream_in): New member function.
9670 (profile_probability::stream_out): New member function.
9671 * profile-count.h (profile_count_quality): Rename to ...
9672 (profile_quality): ... this one.
9673 (profile_probability): New.
9674 (profile_count): Update.
9675 * profile.c (compute_branch_probabilities): Update.
9676 * recog.c (peep2_attempt): Update.
9677 * sched-ebb.c (schedule_ebbs): Update.
9678 * sched-rgn.c (find_single_block_region): Update.
9679 (compute_dom_prob_ps): Update.
9680 (schedule_region): Update.
9681 * sel-sched-ir.c (compute_succs_info): Update.
9682 * stmt.c (struct case_node): Update.
9683 (do_jump_if_equal): Update.
9684 (get_outgoing_edge_probs): Update.
9685 (conditional_probability): Update.
9686 (emit_case_dispatch_table): Update.
9687 (expand_case): Update.
9688 (expand_sjlj_dispatch_table): Update.
9689 (emit_case_nodes): Update.
9690 * targhooks.c: Update.
9691 * tracer.c (better_p): Update.
9692 (find_best_successor): Update.
9693 * trans-mem.c (expand_transaction): Update.
9694 * tree-call-cdce.c: Update.
9695 * tree-cfg.c (gimple_split_edge): Upate.
9696 (move_sese_region_to_fn): Upate.
9697 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9698 * tree-eh.c (lower_resx): Upate.
9699 (cleanup_empty_eh_move_lp): Upate.
9700 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9701 * tree-inline.c (copy_edges_for_bb): Update.
9702 (copy_cfg_body): Update.
9703 * tree-parloops.c (gen_parallel_loop): Update.
9704 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9705 (gimple_gen_time_profiler): Update.
9706 * tree-ssa-dce.c (remove_dead_stmt): Update.
9707 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9708 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9709 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9710 (unloop_loops): Update.
9711 (try_peel_loop): Update.
9712 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9713 * tree-ssa-loop-split.c (connect_loops): Update.
9714 (split_loop): Update.
9715 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9716 (hoist_guard): Update.
9717 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9718 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9719 (value_replacement): Update.
9720 * tree-ssa-reassoc.c (branch_fixup): Update.
9721 * tree-ssa-tail-merge.c (replace_block_by): Update.
9722 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9723 (create_edge_and_update_destination_phis): Update.
9724 (compute_path_counts): Update.
9725 (recompute_probabilities): Update.
9726 (update_joiner_offpath_counts): Update.
9727 (freqs_to_counts_path): Update.
9728 (duplicate_thread_path): Update.
9729 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9730 (struct switch_conv_info): Update.
9731 (gen_inbound_check): Update.
9732 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9733 (vect_do_peeling): Update.
9734 (vect_loop_versioning): Update.
9735 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9736 (optimize_mask_stores): Update.
9737 * ubsan.c (ubsan_expand_null_ifn): Update.
9738 * value-prof.c (gimple_divmod_fixed_value): Update.
9739 (gimple_divmod_fixed_value_transform): Update.
9740 (gimple_mod_pow2): Update.
9741 (gimple_mod_pow2_value_transform): Update.
9742 (gimple_mod_subtract): Update.
9743 (gimple_mod_subtract_transform): Update.
9744 (gimple_ic): Update.
9745 (gimple_stringop_fixed_value): Update.
9746 (gimple_stringops_transform): Update.
9747 * value-prof.h: Update.
9749 2017-06-29 Carl Love <cel@us.ibm.com>
9751 * config/rs6000/rs6000-c.c: Add support for built-in functions
9752 vector signed int vec_signed (vector float);
9753 vector signed long long vec_signed (vector double);
9754 vector signed int vec_signed2 (vector double, vector double);
9755 vector signed int vec_signede (vector double);
9756 vector signed int vec_signedo (vector double);
9757 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9758 instruction generator.
9759 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9760 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9761 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9763 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9764 vunsignede_v2df): Add define_expands.
9765 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9766 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9767 VEC_UNSIGNEDO): Add definitions.
9768 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9769 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9770 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9771 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9772 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9773 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9774 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9775 vec_unsignede, vec_unsignedo): Add builtin defines.
9776 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9778 * doc/extend.texi: Update the built-in documentation file for the
9779 new built-in functions.
9781 2017-06-29 Richard Biener <rguenther@suse.de>
9783 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9784 reduction chains to LOOP_VINFO_REDUCTIONS.
9785 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9786 SLP reductions after processing reduction chains.
9788 2017-06-29 Nathan Sidwell <nathan@acm.org>
9790 * builtins.c (fold_builtin_FUNCTION): Use
9791 lang_hooks.decl_printable_name.
9793 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
9796 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
9797 with only one label.
9798 * stmt.c (expand_case): Assert NCASES is greater than one.
9800 2017-06-29 Richard Biener <rguenther@suse.de>
9802 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
9804 (group_case_labels): Likewise.
9805 (find_taken_edge): Push sanity checking on val to workers...
9806 (find_taken_edge_cond_expr): ... here
9807 (find_taken_edge_switch_expr): ... and here, handle cases
9808 with just a default label.
9809 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
9810 (group_case_labels): Likewise.
9811 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
9812 group_case_labels does anything cleanup the CFG again.
9814 2017-06-29 Bin Cheng <bin.cheng@arm.com>
9816 PR tree-optimization/81196
9817 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
9818 exit condition comparing two IVs.
9820 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
9822 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
9823 profile to the dummy entry at the end of the list of architectures.
9824 * config/arm/arm-cpu-cdata.h: Regenerated.
9826 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9827 Michael Collison <michael.collison@arm.com>
9830 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9832 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9833 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9834 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9835 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9836 of shift when the shift amount is masked with constant equal to
9837 the size of the mode.
9838 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9841 2017-06-29 Martin Liska <mliska@suse.cz>
9843 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9845 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9848 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9851 (edge_growth_cache_entry::edge_growth_cache_entry): New
9853 (reset_edge_growth_cache): Update to use constructor.
9855 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9857 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9858 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9859 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9861 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9863 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9864 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9866 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9868 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9869 (*-linux-uclibc*): Add t-uclibc tmake_file.
9870 * config/t-musl: New.
9871 * config/t-uclibc: New.
9873 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9875 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9877 (gen_comm_data): Emit architectural setting of arch_prof.
9878 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9880 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9881 (armv8-m.base, armv8-m.main): Likewise.
9882 * arm-protos.h (arm_build_target): Add profile field.
9883 (arch_option): Likewise.
9884 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9886 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9887 arm_active_target.profile.
9889 2017-06-28 Richard Biener <rguenther@suse.de>
9892 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9893 TYPE_OVERFLOW_WRAPS.
9894 * match.pd (negate_expr_p): Likewise.
9895 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9896 fold_build2, not fold_binary.
9898 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9900 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9901 Convert memory address to Pmode.
9902 (aarch64_print_operand): Assert MEM operands are always Pmode.
9904 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9907 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9908 Remove redundant if.
9909 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9910 * config/arm/aarch-common-protos.h
9911 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9912 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9914 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9917 * multiple_target.c (create_dispatcher_calls): Set the default
9918 clone to be static, not public.
9920 2017-06-28 Richard Biener <rguenther@suse.de>
9922 * tree-vect-loop.c (vectorizable_reduction): Move special
9923 cond reduction IV var creation ...
9924 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9925 parameter. Use STMT_VINFO_VECTYPE.
9926 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9929 2017-06-28 Martin Liska <mliska@suse.cz>
9932 * ipa-visibility.c (non_local_p): Handle visibility.
9934 2017-06-28 Martin Liska <mliska@suse.cz>
9937 * common.opt: Add IntegerRange to various options.
9938 * opt-functions.awk (integer_range_info): New function.
9939 * optc-gen.awk: Add integer_range_info to cl_options struct.
9940 * opts-common.c (decode_cmdline_option): Handle
9941 CL_ERR_INT_RANGE_ARG.
9942 (cmdline_handle_error): Likewise.
9943 * opts.c (print_filtered_help): Show valid interval in
9944 when --help is provided.
9945 * opts.h (struct cl_option): Add range_min and range_max fields.
9946 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9948 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9950 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9951 (x * C EQ/NE y * C): New transformation.
9953 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9955 * genmultilib (combination_space): Accept '+' in option names.
9957 2017-06-28 Martin Liska <mliska@suse.cz>
9960 * asan.c (instrument_derefs): Bail out inner references
9961 that are hard register variables.
9963 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9966 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9967 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9969 2017-06-28 Richard Biener <rguenther@suse.de>
9971 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9972 (vect_get_slp_defs): Adjust.
9973 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9975 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9977 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9978 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9979 (vectorizable_reduction): Adjust.
9980 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9982 (vect_get_slp_defs): Likewise.
9983 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9984 (vectorizable_bswap): Adjust.
9985 (vectorizable_call): Likewise.
9986 (vectorizable_conversion): Likewise.
9987 (vectorizable_assignment): Likewise.
9988 (vectorizable_shift): Likewise.
9989 (vectorizable_operation): Likewise.
9990 (vectorizable_store): Likewise.
9991 (vectorizable_condition): Likewise.
9992 (vectorizable_comparison): Likewise.
9994 2017-06-28 Michael Collison <michael.collison@arm.com>
9997 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9999 (arm_gen_movmemqi): Removed unused variable 'i'.
10000 Convert 'for' loop into 'while' loop.
10001 (arm_expand_prologue): Remove last unnecessary set of insn.
10002 (thumb_pop): Remove unused variable 'pushed_words'.
10003 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10005 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10007 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10008 * config/s390/s390-protos.h: Add prototype of
10009 s390_rel_address_ok_p.
10010 * config/s390/s390.c (s390_got_symbol): New function.
10011 (s390_rel_address_ok_p): New function.
10012 (legitimize_pic_address): Use s390_rel_address_ok_p.
10013 (s390_load_got): Use s390_got_symbol.
10014 (s390_option_override): Issue error if
10015 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10016 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10018 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10020 2017-06-27 Andrew Pinski <apinski@cavium.com>
10022 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10023 (X * copysign (1.0, X)): New pattern.
10024 (X * copysign (1.0, -X)): New pattern.
10025 (copysign (-1.0, CST)): New pattern.
10027 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10029 * genmultilib (combination_space): Remove variable.
10030 Validate reuse rules against regular expression for any sequence
10031 of multilib options in any order.
10033 2017-06-27 Michael Collison <michael.collison@arm.com>
10035 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10036 call aarch64_split_simd_combine.
10037 * (aarch64_combine_internal<mode>): Delete pattern.
10038 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10039 Allow register and subreg operands.
10041 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10043 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10044 specific need, just fallback on defaults.
10045 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10047 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10048 Olivier Hainque <hainque@adacore.com>
10050 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10052 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10053 targets. Pick a default if no particular attempt applied.
10054 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10057 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10059 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10060 (x86_64-wrs-vxworks7): Likewise.
10062 2017-06-27 Marek Polacek <polacek@redhat.com>
10065 * ubsan.c (instrument_null): Check get_base_address's result for null.
10067 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10069 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10071 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10073 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10074 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10075 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10076 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10077 New function types.
10078 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10079 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10080 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10081 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10082 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10083 BUILT_IN_FEUPDATEENV): New builtins.
10084 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10085 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10086 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10087 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10089 (builtin_structptr_types): Adjust size.
10090 * tree.c (builtin_structptr_types): Add four entries.
10092 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10093 Olivier Hainque <hainque@adacore.com>
10095 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10096 (TLS_SYM): New local macro, forcing reference to __tls__ on
10097 link command lines for VxWorks 7 RTPs, triggering initialization
10099 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10100 OS features TLS support, true for RTPs on VxWorks 7.
10101 * config/vxworks.c (vxworks_override_options): Setup emutls
10104 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10106 * predict.c (test_prediction_value_range): Use -1U instead of -1
10107 to avoid narrowing conversion warning.
10108 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10109 to avoid narrowing conversion warning.
10110 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10112 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10114 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10116 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10117 64bit configurations.
10118 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10119 (SIZE_TYPE): Likewise.
10120 * config/vxworks.c (vxworks_emutls_var_fields): Use
10121 long_unsigned_type_node instead of unsigned_type_node as the offset
10122 field type, which is "pointer" mode in emutls.c.
10124 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10127 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10129 PR middle-end/81207
10130 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10131 gimple_vuse copying separately from gimple_vdef copying.
10133 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10135 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10138 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10139 Olivier Hainque <hainque@adacore.com>
10141 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10142 for all vxworks7 targets.
10143 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10144 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10145 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10146 variations for VX6/VX7 and 32/64bits later on in ...
10147 (VXWORKS_LIB_SPEC): Leverage new macros.
10148 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10149 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10151 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10153 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10154 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10156 2017-06-26 Carl Love <cel@us.ibm.com>
10158 * config/rs6000/rs6000-c.c: Add support for built-in functions
10159 vector bool char vec_reve (vector bool char);
10160 vector signed char vec_reve (vector signed char);
10161 vector unsigned char vec_reve (vector unsigned char);
10162 vector bool int vec_reve (vector bool int);
10163 vector signed int vec_reve (vector signed int);
10164 vector unsigned int vec_reve (vector unsigned int);
10165 vector bool long long vec_reve (vector bool long long);
10166 vector signed long long vec_reve (vector signed long long);
10167 vector unsigned long long vec_reve (vector unsigned long long);
10168 vector bool short vec_reve (vector bool short);
10169 vector signed short vec_reve (vector signed short);
10170 vector double vec_reve (vector double);
10171 vector float vec_reve (vector float);
10172 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10173 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10174 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10175 (altivec_vreve): New pattern.
10176 * config/rs6000/altivec.h (vec_reve): New define.
10177 * doc/extend.texi (vec_rev): Update the built-in documentation file
10178 for the new built-in functions.
10180 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10182 PR tree-optimization/71815
10183 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10185 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10187 (slsr_process_phi): Likewise.
10188 (replace_uncond_cands_and_profitable_phis): Don't replace a
10189 multiply candidate with a stride of 1 (copy or cast).
10190 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10192 (lowest_cost_path): Likewise.
10193 (total_savings): Likewise.
10195 2017-06-26 Richard Biener <rguenther@suse.de>
10198 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10199 Use def_builtin_pure for all gather builtins.
10201 2017-06-26 Richard Biener <rguenther@suse.de>
10203 PR tree-optimization/81203
10204 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10205 non-dominating BBs.
10207 2017-06-26 Marek Polacek <polacek@redhat.com>
10210 * doc/invoke.texi: Document -Wmultistatement-macros.
10212 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10214 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10215 arm_neon_ok_no_float_abi effective target.
10217 2017-06-26 Richard Biener <rguenther@suse.de>
10219 PR tree-optimization/80928
10220 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10221 (copy_bbs): Set BB_DUPLICATED flag early.
10222 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10224 (execute_on_shrinking_pred): Likewise.
10225 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10226 BB_DUPLICATED blocks.
10227 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10228 iterate over all PHIs considering removal of *gsi.
10230 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10232 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10235 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10237 * config/rs6000/rs6000-string.c: (expand_block_clear,
10238 do_load_for_compare, select_block_compare_mode,
10239 compute_current_alignment, expand_block_compare,
10240 expand_strncmp_align_check, expand_strn_compare,
10241 expand_block_move, rs6000_output_load_multiple)
10242 Move functions related to string/block move/compare
10243 to a separate file.
10244 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10245 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10246 for this function which is now used in two files.
10247 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10248 * config.gcc: Add rs6000-string.o to extra_objs for
10249 targets powerpc*-*-* and rs6000*-*-*.
10251 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10254 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10255 32-bit, since indexed is not valid for DImode.
10256 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10257 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10258 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10259 (define_peephole2 for Altivec d-form store): Likewise.
10262 * multiple_target.c (create_dispatcher_calls): Only create the
10263 dispatcher call if the function is the default clone of a
10264 versioned function.
10266 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10268 PR middle-end/80902
10269 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10270 a call, force the call to not be a tail call.
10272 2017-06-23 Jeff Law <law@redhat.com>
10274 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10277 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10279 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10280 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10281 handling for early expansion of vector shifts (sl,sr,sra,rl).
10282 (builtin_function_type): Add vector shift right instructions
10283 to the unsigned argument list.
10285 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10287 rtl-optimizatoin/79286
10288 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10289 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10290 trap. PIC register plus a const unspec without offset can never trap.
10292 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10294 * tree.h (builtin_structptr_type): New type.
10295 (builtin_structptr_types): Declare new array.
10296 * tree.c (builtin_structptr_types): New array.
10297 (free_lang_data, build_common_tree_nodes): Use it.
10299 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10302 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10305 2017-06-22 Matt Turner <mattst88@gmail.com>
10307 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10308 Lake models to skylake case. Assume skylake for unknown
10309 models with clflushopt.
10311 2017-06-22 Jeff Law <law@redhat.com>
10313 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10314 frame sizes that do not satisfy aarch64_uimm12_shift.
10316 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10318 * profile-count.h (apply_probability,
10319 apply_scale, probability_in): Fix checks for zero.
10321 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10323 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10324 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10326 2016-06-22 Richard Biener <rguenther@suse.de>
10328 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10329 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10330 REDUC_MAX_EXPR support.
10331 (vectorizable_reduction): Likewise.
10332 (vect_create_epilog_for_reduction): Likewise.
10334 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10336 * match.pd (A / (1 << B) -> A >> B): New.
10337 * generic-match-head.c: Include optabs-tree.h.
10338 * gimple-match-head.c: Likewise.
10339 * optabs-tree.h (target_supports_op_p): New.
10340 * optabs-tree.c (target_supports_op_p): New.
10342 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10344 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10345 $gcc_cv_ld --help output.
10346 (gcc_cv_ld_demangle): Likewise.
10347 (gcc_cv_ld_eh_frame_hdr): Likewise.
10348 (gcc_cv_ld_pie): Likewise.
10349 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10350 (gcc_cv_ld_buildid): Likewise.
10351 (gcc_cv_ld_sysroot): Likewise.
10352 (ld_bndplt_support): Likewise.
10353 (ld_pushpopstate_support): Likewise.
10354 * configure: Regenerate.
10355 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10357 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10360 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10361 operands indexes to avoid gap between operands and match_dups.
10363 2017-06-21 Andrew Pinski <apinski@cavium.com>
10365 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10366 Increment Arith_shift and Arith_shift_reg by 1.
10367 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10369 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10370 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10371 (aarch64_strip_extend): Add new argument and test for it.
10372 (aarch64_cheap_mult_shift_p): New function.
10373 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10374 add a cost if it is true.
10375 Update calls to aarch64_strip_extend.
10376 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10378 2017-06-21 Andrew Pinski <apinski@cavium.com>
10380 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10382 (thunderxt88): Likewise.
10383 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10384 (thunderx_prefetch_tune): New variable.
10385 (thunderx2t99_prefetch_tune): Update for the correct values.
10386 (thunderxt88_tunings): New variable.
10387 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10388 generic_prefetch_tune.
10389 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10391 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10393 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10394 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10395 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10396 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10397 (aarch64_atomic_cas<mode>, GPI): Likewise.
10399 2017-06-21 Martin Liska <mliska@suse.cz>
10401 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10402 statements on cold and hot labels.
10403 * predict.c (tree_estimate_probability_bb): Remove the
10404 prediction from this place.
10406 2017-06-21 Martin Liska <mliska@suse.cz>
10408 PR tree-optimization/79489
10409 * gimplify.c (maybe_add_early_return_predict_stmt): New
10411 (gimplify_return_expr): Call the function.
10412 * predict.c (tree_estimate_probability_bb): Remove handling
10414 * predict.def: Update comment about early return predictor.
10415 * gimple-predict.h (is_gimple_predict): New function.
10416 * predict.def: Change default value of early return to 66.
10417 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10419 * passes.def: Put pass_strip_predict_hints to the beginning of
10422 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10424 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10425 FUNCTION_DECL declarations.
10426 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10428 (dwaf2out_decl): Likewise.
10429 * godump.c (go_early_global_decl): Skip call to the real debug hook
10430 for FUNCTION_DECL declarations.
10431 * passes.c (rest_of_decl_compilation): Skip call to the
10432 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10433 -fdump-go-spec is passed.
10435 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10437 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10438 Reorder for compactness.
10439 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10440 (def_builtin_pure, def_builtin_pure2): New functions.
10441 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10443 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10445 * match.pd (nop_convert): New predicate.
10446 ((A +- CST1) +- CST2): Allow some NOP conversions.
10448 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10451 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10452 with ctors/dtors if GOVD_SHARED is set.
10454 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10456 * config/aarch64/aarch64.md (movti_aarch64):
10457 Emit mov rather than orr.
10458 (movtf_aarch64): Likewise.
10459 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10460 Emit mov rather than orr.
10462 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10464 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10465 Swap alternatives, make integer dup more expensive.
10467 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10469 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10470 Return true for non-tls symbols.
10472 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10474 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10475 (cortex-a75): Likewise.
10476 (cortex-a75.cortex-a55): Likewise.
10477 * config/aarch64/aarch64-tune.md: Regenerate.
10478 * doc/invoke.texi (-mtune): Document new values for -mtune.
10480 2017-06-21 Tom de Vries <tom@codesourcery.com>
10482 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10483 stack_size feature.
10484 (Effective-Target Keywords, Other attributes): Suggest using
10485 dg-add-options stack_size feature to get stack limit in stack_size
10486 effective target documentation.
10488 2017-06-21 Julian Brown <julian@codesourcery.com>
10489 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10491 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10492 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10493 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10495 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10496 attribute type list for neon_multiply.
10497 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10498 attribute type list for neon_multiply.
10499 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10500 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10501 attribute type list for neon_multiply.
10502 * config/arm/types.md (crypto_pmull): Add.
10503 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10504 attribute type list.
10506 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10508 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10511 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10513 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10514 to make sure not to dereference a NULL cost_classes_ptr pointer.
10516 2017-06-20 Carl Love <cel@us.ibm.com>
10518 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10519 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10520 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10521 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10522 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10523 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10524 VMULOSW): New enum "unspec" values.
10525 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10526 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10527 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10528 altivec_vmulosw): New patterns.
10529 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10530 VMULOSW): Add definitions.
10532 2017-06-20 Julia Koval <julia.koval@intel.com>
10534 * config/i386/i386.c: Fix rounding expand for new pattern.
10535 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10537 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10539 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10540 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10542 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10544 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10547 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10549 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10550 architecture, then by implementer ID.
10551 * config/aarch64/aarch64-tune.md: Regenerate.
10553 2017-06-20 Richard Biener <rguenther@suse.de>
10555 PR middle-end/81097
10556 * fold-const.c (split_tree): Fold to type before negating.
10558 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10560 * diagnostic-show-locus.c
10561 (selftest::test_fixit_deletion_affecting_newline): New function.
10562 (selftest::diagnostic_show_locus_c_tests): Call it.
10564 2017-06-20 Andreas Schwab <schwab@suse.de>
10567 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10570 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10572 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10573 __ARM_FEATURE_COPROC according to support.
10575 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10577 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10578 Rewritten to avoid overflow for > 32-bit pointers.
10581 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10582 by removing enum keyword.
10583 (ubsan_type_descriptor): Likewise. Formatting fix.
10586 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10587 splitter): Require TARGET_SSE2 in the condition.
10589 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10592 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10593 for doing vector set of SFmode on ISA 3.0.
10594 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10595 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10597 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10598 SFmode value into a V4SF variable that was extracted from another
10599 V4SF variable without converting the element to double precision
10600 and back to single precision vector format.
10601 (vsx_insert_extract_v4sf_p9_2): Likewise.
10603 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10605 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10606 in UWHI to avoid undefined overflow.
10609 * ubsan.h (enum ubsan_encode_value_phase): New.
10610 (ubsan_encode_value): Change second argument to
10611 enum ubsan_encode_value_phase with default value of
10612 UBSAN_ENCODE_VALUE_GENERIC.
10613 * ubsan.c (ubsan_encode_value): Change second argument to
10614 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10615 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10616 create_tmp_var_raw instead of create_tmp_var and use a
10618 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10619 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10620 ubsan_encode_value callers.
10623 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10624 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10625 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10627 2017-06-19 Richard Biener <rguenther@suse.de>
10629 PR middle-end/81118
10630 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10631 estimates if we changed anything.
10633 2017-06-19 Richard Biener <rguenther@suse.de>
10635 PR tree-optimization/80887
10636 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10637 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10638 simplified lookups, then reset mprts_hook.
10639 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10641 (try_to_simplify): Likewise.
10643 2017-06-19 Martin Liska <mliska@suse.cz>
10646 * gimplify.c (gimplify_switch_expr):
10647 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10649 2017-06-19 Martin Liska <mliska@suse.cz>
10651 * doc/install.texi: Document that PGO runs in 4 stages.
10653 2017-06-19 Martin Liska <mliska@suse.cz>
10656 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10657 to dispatcher function name.
10658 * multiple_target.c (replace_function_decl): New function.
10659 (create_dispatcher_calls): Redirect both edges and references.
10661 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10663 * profile-count.c (profile_count::dump): Dump quality.
10664 (profile_count::differs_from_p): Update for unsigned val.
10665 * profile-count.h (profile_count_quality): New enum.
10666 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10668 2017-06-19 Richard Biener <rguenther@suse.de>
10670 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10671 struct function as arg.
10672 (estimate_numbers_of_iterations): Export overload with loop arg.
10673 (free_numbers_of_iterations_estimates_loop): Use an overload of
10674 free_numbers_of_iterations_estimates instead.
10675 * tree-cfg.c (remove_bb): Adjust.
10676 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10677 * tree-parloops.c (gen_parallel_loop): Likewise.
10678 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10680 (tree_unroll_loops_completely): Likewise.
10681 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10682 Use an overload instead and export.
10683 (estimated_loop_iterations): Adjust.
10684 (max_loop_iterations): Likewise.
10685 (likely_max_loop_iterations): Likewise.
10686 (estimate_numbers_of_iterations): Take struct function as arg
10688 (loop_exits_before_overflow): Adjust.
10689 (free_numbers_of_iterations_estimates_loop): Use an overload.
10690 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10691 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10693 2017-06-19 Richard Biener <rguenther@suse.de>
10696 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10697 RANGE_EXPR conservatively.
10699 2017-06-16 Carl Love <cel@us.ibm.com>
10701 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10702 definitions for vec_float, vec_float2, vec_floato,
10703 vec_floate built-ins.
10704 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10705 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10707 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10708 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10709 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10710 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10711 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10712 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10713 vec_floato): Add builtin defines.
10714 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10715 Update the built-in documentation file for the new built-in
10718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10720 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10721 (mthumb): Mark as the negative of -marm.
10723 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10725 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10727 (ARM Options, -mtune): Document that this accepts the same
10728 extension options as -mcpu.
10729 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10733 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10734 permitted extensions.
10736 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10738 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10739 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10740 (armv8-m.main): Add option +nodsp.
10741 * config/arm/arm-cpu-cdata.h: Regenerated.
10743 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10745 * config/arm/t-fuchsia: New file.
10746 * config.gcc (arm*-*-fuchsia*): Use it.
10748 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10750 * config/arm/t-symbian: Rewrite for new option infrastructure.
10752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10754 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10755 (MULTILIB_REQUIRED): Likewise.
10757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10759 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10760 (MULTILIB_RESUE): Likewise.
10761 (MULTILIB_MATCHES): Likewise.
10762 (MULTLIB_REQUIRED): Likewise.
10764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10766 * config/arm/t-rtems: Rewrite for new option framework.
10768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10770 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10771 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10772 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10773 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10774 * config/arm/t-multilib: ... here.
10775 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10776 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10777 armv7-a and armv8*-a when A-profile libraries have not been built.
10778 * config/arm/t-rmprofile: Rewrite.
10780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10782 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10783 with a backslash. Remove the backslash after substituting unescaped
10785 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10789 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10790 the driver to pass through the default setting of -mfloat-abi.
10791 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
10793 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
10794 (all_feat_combs): New rule.
10795 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
10797 * config/arm/t-aprofile: Rewrite.
10799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10801 * config/arm/arm.h (FPUTYPE_AUTO): Define.
10802 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
10803 fpu is not specified by the user/command-line.
10804 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
10805 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
10806 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
10807 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
10808 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
10809 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
10811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10813 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
10814 * config/arm/t-arm-elf: Rewritten.
10816 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10818 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
10819 have some floating-point instructions.
10820 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
10821 (TARGET_MAYBE_HARD_FLOAT): New macro.
10822 * config/arm/arm-builtins.c (arm_init_builtins): Use
10823 TARGET_MAYBE_HARD_FLOAT.
10824 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10828 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10829 (configargs.h): Include it.
10830 (arm_print_hint_for_fpu_option): New function.
10831 (arm_parse_fpu_option): New function.
10832 (candidate_extension): New class.
10833 (arm_canon_for_multilib): New function.
10834 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10835 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10836 (ARCH_CANONICAL_SPECS): New macro.
10837 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10841 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10842 are set after handling multilib fragments. Set target_cpu_default2
10845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10847 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10849 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10850 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10852 (check_arch): Likewise.
10853 * config/arm/arm.c (arm_configure_build_target): Handle
10854 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10855 options in the default.
10857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10859 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10860 when an option is an alias of another.
10861 * config/arm/parsecpu.awk (optalias): New parser token.
10862 (gen_comm_data): Mark non-alias options as such. Emit entries
10863 for extension aliases.
10864 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10865 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10866 (armv6kz, armv6zk, armv6t2): Likewise.
10867 (armv7): Make vfpv3-d16 an alias.
10868 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10870 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10871 Sort in canonical order.
10872 (armv8-a): Sort in canonical order.
10873 (armv8.1-a, armv8.2-a): Likewise.
10874 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10876 (cortex-a9): Sort in canonical order.
10877 * config/arm/arm.c (selftests.h): Include it.
10878 (arm_test_cpu_arch_data): New function.
10879 (arm_run_self_tests): New function.
10880 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10881 (targetm): Move declaration to the end of the file.
10882 * arm-cpu-cdata.h: Regenerated.
10884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10886 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10887 call to target_mode_check describing the type of option passed.
10888 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10889 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10890 arm_parse_cpu_option_name to match parameters against list of
10892 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10893 arm_arch_core_flags data structure.
10894 * config/arm/arm-cpu_cdata.h: Regenerated.
10896 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10898 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10900 (arm_print_hint_for_cpu_option): Likewise.
10901 (arm_print_hint_for_arch_option): Likewise.
10902 (arm_parse_cpu_option_name): Likewise.
10903 (arm_parse_arch_option_name): Likewise.
10904 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10905 of entries in the all_fpus list.
10906 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10907 (arm_parse_cpu_option_name): Declare.
10908 (arm_parse_arch_option_name): Declare.
10909 (arm_parse_option_features): Declare.
10910 (arm_intialize_isa): Declare.
10911 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10913 (gen_comm_data): ... here. Make definitions non-static.
10914 * config/arm/arm-cpu-data.h: Regenerated.
10915 * config/arm/arm-cpu-cdata.h: Regenerated.
10917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10919 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10920 (cpu_arch_extension): New structure.
10921 (cpu_arch_option, arch_option, cpu_option): New structures.
10922 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10923 architecture types.
10924 (gen_data): Generate new format data tables.
10925 * config/arm/arm.c (cpu_tune): New structure.
10926 (cpu_option, processors): Delete.
10927 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10928 (arm_print_hint_for_cpu_option): ... this and ...
10929 (arm_print_hint_for_arch_option): ... this.
10930 (arm_parse_arch_cpu_name): Delete. Replace with ...
10931 (arm_parse_cpu_option_name): ... this and ...
10932 (arm_parse_arch_option_name): ... this.
10933 (arm_unrecognized_feature): Change type of target parameter to
10935 (arm_parse_arch_cpu_features): Delete. Replace with ...
10936 (arm_parse_option_features): ... this.
10937 (arm_configure_build_target): Rework to use new configuration data
10939 (arm_print_tune_info): Rework for new configuration data tables.
10940 * config/arm/arm-cpu-data.h: Regenerated.
10941 * config/arm/arm-cpu.h: Regenerated.
10943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10945 * Makefile.in (OBJS): Move sbitmap.o from here ...
10946 (OBJS-libcommon): ... to here.
10948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10950 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10951 (ISA_ALL_CRYPTO): New macro.
10952 (ISA_ALL_SIMD): New macro
10953 (ISA_ALL_FP): New macro.
10954 * config/arm/arm.c (fpu_bitlist): Update initializer.
10955 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10957 (arm9e): Add fpu. Add option for nofp
10958 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10959 (arm926ej-s, arm1026ej-s): Likewise.
10960 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10961 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10962 neon-fp16, neon-vfpv4, nofp and nosimd.
10963 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10964 (cortex-a8): Add fpu. Add option for nofp.
10965 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10966 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10967 (cortex-r4f): Add fpu.
10968 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10969 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10971 (cortex-r8): Likewise.
10972 (cortex-m4): Add fpu. Add option for nofp.
10973 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10974 (cortex-a17.cortex-a7): Likewise.
10975 (cortex-a32): Add fpu. Add options for crypto and nofp.
10976 (cortex-a35, cortex-a53): Likewise.
10977 (cortex-a57): Add fpu. Add option for crypto.
10978 (cortex-a72, cortex-a73): Likewise.
10979 (exynos-m1): Likewise.
10980 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10981 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10982 (cortex-m33): Add fpu. Add option for nofp.
10983 * config/arm/arm-cpu-cdata.h: Regenerated
10984 * config/arm/arm-cpu-data.h: Regenerated.
10986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10988 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10989 (armv5te, armv5tej): Likewise.
10990 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10991 (armv7): Add options fp and vfpv3-d16.
10992 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10993 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10995 (armv7ve): Likewise.
10996 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10997 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10998 (armv8-a): Add nocrypto option.
10999 (armv8.1-a, armv8.2-a): Likewise.
11000 (armv8-m.main): add options fp, fp.dp and nofp.
11002 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11004 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11006 (armv8-a+crc): Delete.
11007 (armv8.1-a): Add options simd, crypto and nofp.
11008 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11009 (armv8.2-a+fp16): Delete.
11010 (armv8-m.main): Add option dsp.
11011 (armv8-m.main+dsp): Delete.
11012 (cortex-a8): Add fpu. Add nofp option.
11013 (cortex-a9): Add fpu. Add nofp and nosimd options.
11014 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11015 link to main cpu and architecture data structures.
11016 (gen_comm_data): Only put isa attributes from the main architecture
11018 (option): New statement for architecture and CPU entries.
11019 * arm.c (struct cpu_option): New structure.
11020 (struct processors): Add entry for options.
11021 (arm_unrecognized_feature): New function.
11022 (arm_parse_arch_cpu_name): Ignore any characters after the first
11024 (arm_parse_arch_cpu_feature): New function.
11025 (arm_configure_build_target): Separate out any CPU and architecture
11026 features and parse separately. Don't error out if -mfpu=auto is
11027 used with only an architecture string.
11028 (arm_print_asm_arch_directives): New function.
11029 (arm_file_start): Call it.
11030 * config/arm/arm-cpu-cdata.h: Regenerated.
11031 * config/arm/arm-cpu-data.h: Likewise.
11032 * config/arm/arm-tables.opt: Likewise.
11034 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11036 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11037 assembler when it is not -mfpu=auto.
11039 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11041 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11042 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11043 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11044 (ASM_CPU_SPEC): Rewrite.
11045 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11046 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11047 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11048 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11049 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11050 copied string is NUL-terminated. Also strip any characters prefixed
11052 (arm_rewrite_selected_arch): New function.
11053 (arm_rewrite_march): New function.
11055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11057 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11058 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11059 (march, mcpu, mtune): Convert to string-based options.
11060 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11061 (arm_parse_arch_cpu_name): New function.
11062 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11063 identify selected architecture or CPU.
11064 (arm_option_save): New function.
11065 (TARGET_OPTION_SAVE): Redefine.
11066 (arm_option_restore): Restore string options.
11067 (arm_option_print): Print string options.
11069 2017-06-16 Martin Sebor <msebor@redhat.com>
11071 PR tree-optimization/80933
11072 PR tree-optimization/80934
11073 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11074 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11075 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11076 (gimple_fold_builtin): Call them.
11078 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11080 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11081 as unlikely; update profile.
11083 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11085 * predict.c (force_edge_cold): Handle declaring edges impossible
11088 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11090 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11092 (try_unroll_loop_completely): Fix reporting.
11094 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11096 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11098 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11101 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11102 if given a non-constant argument for an intrinsic which requires a
11105 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11107 * profile.c (compare_freqs): New function.
11108 (branch_prob): Sort edge list.
11109 (find_spanning_tree): Assume that the list is priority sorted.
11111 2017-06-16 Richard Biener <rguenther@suse.de>
11113 PR tree-optimization/81090
11114 * passes.def (pass_record_bounds): Remove.
11115 * tree-pass.h (make_pass_record_bounds): Likewise.
11116 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11117 make_pass_record_bounds): Likewise.
11118 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11119 not free niter estimates at the beginning but at the end.
11120 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11122 2017-06-16 Richard Biener <rguenther@suse.de>
11124 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11125 initializer to workaround ICE in host GCC 4.8.
11127 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11129 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11131 (clone_inlined_nodes): Update.
11133 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11135 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11136 prefetch settings, and enable prefetching by default at -O3.
11138 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11140 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11141 Set flag_prefetch_loop_arrays according to tuning data.
11143 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11145 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11146 New tune structure.
11147 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11148 [Unrelated to main purpose of the patch] Place the pointer field last
11149 to enable type checking errors when tune structure are wrongly merged.
11150 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11151 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11152 (thunderx2t99_prefetch_tune): New tune constants.
11153 (tune_params *_tunings): Update all tunings (no functional change).
11154 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11155 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11156 from tunings structures.
11158 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11161 * ubsan.c (instrument_null): Add T argument, use it instead
11162 of computing it based on IS_LHS.
11163 (instrument_object_size): Likewise.
11164 (pass_ubsan::execute): Adjust instrument_null and
11165 instrument_object_size callers to pass gimple_get_lhs or
11166 gimple_assign_rhs1 result to it. Use instrument_null instead of
11167 calling get_base_address and instrument_mem_ref. Handle
11168 aggregate call arguments for object-size sanitization.
11170 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11172 PR tree-optimization/81089
11173 * tree-vrp.c (is_masked_range_test): Validate operands of
11176 2017-06-15 Martin Sebor <msebor@redhat.com>
11179 * dumpfile.c (dump_register): Avoid calling memset to initialize
11180 a class with a default ctor.
11181 * gcc.c (struct compiler): Remove const qualification.
11182 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11183 * hash-table.h: Ditto.
11184 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11186 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11187 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11189 * params.h (struct param_info): Make struct members non-const.
11190 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11191 with default initialization.
11192 * vec.h (vec_copy_construct, vec_default_construct): New helper
11194 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11195 with vec_copy_construct.
11196 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11197 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11198 * doc/invoke.texi (-Wclass-memaccess): Document.
11200 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11202 * emit-rtl.h (is_leaf): Update comment about local
11203 register allocator.
11205 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11208 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11209 for a variable to have a section before checking if the section has a
11211 Set section to.persistent if persistent attribute is set.
11212 Warn if .persistent attribute is used on an automatic variable.
11214 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11216 PR rtl-optimization/80474
11217 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11219 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11221 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11224 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11226 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11227 (casesi): Emit bounds checking as RTL.
11228 (casesi_internal_mips16_<mode>): Remove bounds checking.
11230 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11232 * config/xtensa/xtensa.c (xtensa_option_override): Append
11233 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11234 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11235 xtensa_doloop_hooks): Define unconditionally.
11236 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11238 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11239 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11240 for it in xtensa_option_override.
11241 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11242 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11244 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11246 * doc/cppopts.texi: Document '-' special value to -MF.
11248 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11250 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11251 (cortex_a53_fconst): Likewise.
11252 (cortex_a53_fpmul): Likewise.
11253 (cortex_a53_f_load_64): Likewise.
11254 (cortex_a53_f_load_many): Likewise.
11255 (cortex_a53_advsimd_alu): Likewise.
11256 (cortex_a53_advsimd_alu_q): Likewise.
11257 (cortex_a53_advsimd_mul): Likewise.
11258 (cortex_a53_advsimd_mul_q): Likewise.
11259 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11260 Add missing fmul, r2f_cvt and fconst cases.
11262 2017-06-14 Richard Biener <rguenther@suse.de>
11264 PR middle-end/81088
11265 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11267 (fold_binary_loc): When associating do not treat pre-existing
11268 TREE_OVERFLOW on literal constants as a reason to allow
11269 TREE_OVERFLOW on associated literal constants.
11271 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11273 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11274 (MASK_FEATURES): New macro.
11275 * config/sparc/sparc.c (sparc_option_override): Remove the special
11276 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11278 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11280 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11281 a division of 0 if non-call exceptions are enabled.
11283 2017-06-14 Andrew Pinski <apinski@cavium.com>
11284 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11287 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11288 Improve vector initialization code gen for only variable case.
11290 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11292 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11294 2017-06-14 Richard Biener <rguenther@suse.de>
11296 PR tree-optimization/81083
11297 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11300 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11302 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11303 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11304 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11305 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11306 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11307 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11309 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11311 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11312 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11314 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11316 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11318 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11320 * config/rs6000/t-rtems: Don't handle SPE.
11322 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11324 * config/rs6000/t-linux: Don't handle SPE.
11326 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11328 * config/rs6000/eabispe.h: Delete file.
11330 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11332 * config/rs6000/t-spe: Delete file.
11334 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11336 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11337 (rs6000_legitimate_offset_address_p): Return false for anything in
11338 V2SImode or V2SFmode.
11340 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11342 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11343 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11344 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11346 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11347 (rs6000_legitimate_offset_address_p): Ditto.
11348 (rs6000_emit_move): Ditto.
11349 (rs6000_init_builtins): Remove V4HI_type_node.
11351 2017-06-13 Martin Liska <mliska@suse.cz>
11354 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11355 (gate_asan): Likewise.
11356 * asan.h (asan_no_sanitize_address_p): Remove the function.
11357 (sanitize_flags_p): New function.
11358 * builtins.def: Fix coding style.
11359 * common.opt: Use renamed enum value.
11360 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11361 * doc/extend.texi: Document no_sanitize attribute.
11362 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11363 to SANITIZE_UNDEFINED_NONDEFAULT.
11364 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11365 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11366 Use sanitize_flags_p.
11367 * gimplify.c (gimplify_function_tree): Likewise.
11368 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11369 * opts.c (parse_no_sanitize_attribute): New function.
11370 (common_handle_option): Use renamed enum value.
11371 * opts.h (parse_no_sanitize_attribute): Declare.
11372 * tree.c (sanitize_flags_p): New function.
11373 * tree.h: Declared here.
11374 * tsan.c: Use sanitize_flags_p.
11375 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11376 (instrument_mem_ref): Likewise.
11377 (instrument_bool_enum_load): Likewise.
11378 (do_ubsan_in_current_function): Remove the function.
11379 (pass_ubsan::execute): Use sanitize_flags_p.
11380 * ubsan.h: Remove do_ubsan_in_current_function
11381 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11382 (dump_function_to_file): Use it here.
11384 2017-06-13 Martin Jambor <mjambor@suse.cz>
11386 PR tree-optimization/80803
11387 PR tree-optimization/81063
11388 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11389 (propagate_subaccesses_across_link): Enqueue subtree whenever
11390 necessary instead of relying on the caller.
11392 2017-06-13 Martin Jambor <mjambor@suse.cz>
11394 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11395 that have a first_link.
11396 (sort_and_splice_var_accesses): Do not check first_link before
11398 (subtree_mark_written_and_enqueue): Likewise.
11399 (propagate_all_subaccesses): Likewise and do not stop at first
11400 parent with a first_link.
11402 2017-06-13 Martin Jambor <mjambor@suse.cz>
11404 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11407 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11409 * match.pd: New pattern.
11411 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11413 * tree-vrp.c (is_masked_range_test): New function.
11414 (register_edge_assert_for): Determine ranges for
11417 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11419 PR tree-optimization/67328
11420 * fold-const.c (maskable_range_p): New function.
11421 (build_range_check): Generate bittests if possible.
11423 2017-06-13 Martin Liska <mliska@suse.cz>
11425 * gimple-pretty-print.c (dump_probability): Add new argument.
11426 (dump_edge_probability): Dump both probability and count.
11427 (dump_gimple_label): Likewise.
11428 (dump_gimple_bb_header): Likewise.
11430 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11433 * config/avr/avr-devices.c: Fix indentation.
11434 * config/avr/gen-avr-mmcu-specs.c: Dito.
11436 2017-06-13 Richard Biener <rguenther@suse.de>
11438 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11439 instead get vector type from stmt_info.
11440 (vectorizable_reduction): Adjust. Remove dead code.
11442 2017-06-13 Richard Biener <rguenther@suse.de>
11444 PR middle-end/81065
11445 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11446 case of C * (x * C2 + C3).
11447 (fold_addr_of_array_ref_difference): Properly fold index difference.
11449 2017-06-12 David S. Miller <davem@davemloft.net>
11452 * config/sparc/sparc.md (return expander): Emit frame blockage if
11453 function uses alloca.
11455 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11457 * combine.c (make_field_assignment): Check len rather than the mode
11458 precision when calling force_to_mode.
11460 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11462 Support multilibs and devices that see flash in RAM address range.
11465 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11466 (avr_mcu_t) <flash_pm_offset>: New field.
11467 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11468 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11469 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11470 (AVR_TINY_PM_OFFSET): Remove macro.
11471 * config/avr/avr.opt (-mshort-calls): New option.
11472 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11473 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11474 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11475 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11476 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11477 instead of avr_arch->have_jmp_call.
11478 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11479 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11480 avr_arch->flash_pm_offset to define.
11481 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11482 new field flash_pm_offset. Add entry for avrxmega3.
11483 (avr_texinfo): Add entry for avrxmega3.
11484 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11485 attiny212, attiny214,
11486 attiny412, attiny414, attiny416, attiny417,
11487 attiny814, attiny816, attiny817,
11488 attiny1614, attiny1616, attiny1617,
11489 attiny3214, attiny3216, attiny3217.
11490 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11491 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11492 (avr_print_operand_address) [AVR_TINY]: Same.
11493 (avr_asm_init_sections) <readonly_data_section>: Only patch
11494 callback if avr_arch->flash_pm_offset = 0.
11495 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11496 for rodata if avr_arch->flash_pm_offset != 0.
11497 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11498 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11499 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11500 (m_options): Append opt_rcall.
11501 (m_dirnames): Append dir_rcall.
11502 * config/avr/t-multilib: Regenerate.
11504 * configure.ac [target=avr]: Check whether avrxmega3 default
11505 linker description file works as needed.
11506 * configure: Regenerate.
11507 * doc/avr-mmcu.texi: Regenerate.
11508 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11509 <__AVR_ARCH__>: Document avrxmega3 and 103.
11510 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11511 <__AVR_SHORT_CALLS__>: Document it.
11512 <__AVR_PM_BASE_ADDRESS__>: Document it.
11513 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11514 (AVR Variable Attributes) <progmem>: Document this is
11515 not needed for avrxmega3.
11516 (AVR Named Address Spaces) <__flash>: Dito.
11518 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11520 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11522 2017-06-12 Doug Rupp <rupp@adacore.com>
11524 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11525 Append vxworks-stdint.h to the tm_file list.
11526 * config/vxworks-stdint.h: New file.
11528 2017-06-12 Martin Liska <mliska@suse.cz>
11530 PR tree-optimization/81041
11531 * tree-profile.c (gimple_gen_ic_func_profiler):
11532 Create an extra BB in profile-generate
11533 (gimple_gen_time_profiler): Likewise.
11535 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11537 PR tree-optimization/81003
11538 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11539 (update_range_test): Use it instead of force_gimple_operand_gsi.
11541 2017-06-12 Richard Biener <rguenther@suse.de>
11543 PR tree-optimization/81053
11544 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11545 with backedge value not defined in loop. Simplify def stmt
11548 2017-06-11 Tom de Vries <tom@codesourcery.com>
11551 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11553 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11554 nvptx_cannot_force_const_mem.
11556 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11558 * opts.c (finish_options): Move test for flag_split_stack after
11559 it has been initialized.
11561 2017-06-11 Jason Merrill <jason@redhat.com>
11563 * tree.h (id_equal): New.
11564 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11565 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11566 instead of strcmp of IDENTIFIER_POINTER.
11568 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11570 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11571 (mark_all_inlined_calls_cdtor): Fix formating.
11572 (inline_transform): Rescale profile before inlining.
11574 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11576 * cgraph.h (cgraph_edge::clone): Update prototype.
11577 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11578 (cgraph_node::create_clone): Update.
11579 (cgraph_node::create_version_clone): Update.
11580 * tree-inline.c (copy_bb): Update.
11581 (expand_call_inline): Update.
11583 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11585 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11586 factored out from ...
11587 (rs6000_emit_prologue): ... here.
11589 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11591 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11592 factored out from ...
11593 (rs6000_emit_prologue): ... here.
11595 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11597 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11599 (handle_missing_profiles): Fix computation of tp_first_run.
11600 (counts_to_freqs): Do not touch freqs when count is 0.
11602 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11604 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11607 2017-06-10 Tom de Vries <tom@codesourcery.com>
11609 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11610 attributes): Document signal effective target.
11612 2017-06-10 Tom de Vries <tom@codesourcery.com>
11614 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11615 Document effective target stack_size.
11617 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11619 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11620 to the edit_context if they can be auto-applied.
11622 2017-06-9 Ian Lance Taylor <iant@golang.org>
11624 * opts.c (finish_options): If -fsplit-stack, disable implicit
11625 -forder-blocks-and-partition.
11626 * doc/invoke.texi (Optimize Options): Document that when using
11627 -fsplit-stack -forder-blocks-and-partition is not implicitly
11630 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11632 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11633 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11634 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11635 * builtins.def (abort, trap, unreachable): Declare cold.
11636 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11637 * tree-core.h (ECF_COLD): New.
11638 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11639 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11641 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11643 * predict.c (unlikely_executed_stmt_p): Cleanup.
11645 2017-06-09 Richard Biener <rguenther@suse.de>
11647 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11648 model if the ref is always written to.
11650 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11652 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11654 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11656 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11659 2017-06-09 Tom de Vries <tom@codesourcery.com>
11662 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11663 "target cannot support label values" when encountering LABEL_REF.
11665 2017-06-09 Martin Liska <mliska@suse.cz>
11667 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11668 (gimple_gen_ic_func_profiler): Emit direct comparison
11669 of __gcov_indirect_call_callee with NULL.
11670 (gimple_gen_time_profiler): Change probability from
11671 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11673 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11675 * profile.c (edge_gcov_counts): Turn to pointer.
11676 (compute_branch_probabilities, compute_branch_probabilities): Update.
11677 (branch_prob): Do not clear edge_gcov_count.
11678 * profile.h (edge_gcov_counts): Turn to pointer.
11679 (edge_gcov_count): Update.
11681 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11683 * gimple.h (gimple_check_failed): Mark cold.
11685 2017-06-09 Richard Biener <rguenther@suse.de>
11687 PR tree-optimization/66623
11688 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11689 refactor check_reduction into two parts, properly computing
11690 whether we have to check reduction validity for outer loop
11693 2017-06-09 Richard Biener <rguenther@suse.de>
11695 PR tree-optimization/79483
11696 * graphite-scop-detection.c (order): New global.
11697 (get_order): Compute bb to order mapping that satisfies code
11698 generation constraints.
11699 (cmp_pbbs): New helper.
11700 (build_scops): Start domwalk at entry block, sort generated
11703 2017-06-09 Richard Biener <rguenther@suse.de>
11705 PR middle-end/81007
11706 * ipa-polymorphic-call.c
11707 (ipa_polymorphic_call_context::restrict_to_inner_class):
11708 Skip FIELD_DECLs with error_mark_node type.
11709 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11712 2017-06-09 Martin Liska <mliska@suse.cz>
11714 * predict.c (struct branch_predictor): New struct.
11715 (test_prediction_value_range): New test.
11716 (predict_c_tests): New function.
11717 * selftest-run-tests.c (selftest::run_tests): Run the function.
11718 * selftest.h: Declare new tests.
11720 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11723 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11724 gen_add3_insn did not fail.
11725 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11726 r0, construct that number in a temporary reg and add that reg to r0.
11727 If asked to put the result in r0 as well, fail.
11729 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11731 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11732 for early expansion of vec_eqv.
11734 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11736 PR middle-end/81005
11737 * ubsan.c (instrument_null): Avoid pointless code temporary.
11738 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11740 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11744 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11747 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11750 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11752 * predict.c (unlikely_executed_edge_p): Move ahead.
11753 (probably_never_executed_edge_p): Use it.
11755 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11757 PR middle-end/79988
11758 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11759 gimple_call_builtin_p call.
11761 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11763 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11764 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11765 rtl_check_failed_type2, rtl_check_failed_code1,
11766 rtl_check_failed_code2, rtl_check_failed_code_mode,
11767 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11768 rtvec_check_failed_bounds, rtl_check_failed_flag,
11769 _fatal_insn_not_found, _fatal_insn): Likewise.
11770 * tree.h (tree_contains_struct_check_failed,
11771 tree_check_failed, tree_not_check_failed,
11772 tree_class_check_failed, tree_range_check_failed,
11773 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11774 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11775 tree_operand_check_failed, omp_clause_check_failed,
11776 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11779 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11781 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11782 flag_branch_probabilities.
11783 * ipa-inline.c (edge_badness): Likewise.
11784 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11785 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11786 * predict.c (maybe_hot_frequency_p): Likewise.
11787 (probably_never_executed): Likewise.
11788 * sched-ebb.c (schedule_ebbs): Likewise.
11789 * sched-rgn.c (find_single_block_region): Likewise.
11790 * tracer.c (tail_duplicate): Likewise.
11792 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11794 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
11795 longer requires x_flag_profile_use.
11797 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11799 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
11800 instead of flag_reorder_blocks_and_partition.
11801 * dbxout.c (dbxout_function_end): Likewise.
11802 * dwarf2out.c (gen_subprogram_die): Likewise.
11803 * haifa-sched.c (sched_create_recovery_edges): Likewise.
11804 * hw-doloop.c (reorg_loops): Likewise.
11805 * varasm.c (assemble_start_function,
11806 assemble_end_function): Likewise.
11807 (decide_function_section): Do not check for
11808 flag_reorder_blocks_and_partition.
11810 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11812 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11814 (chkp_get_hard_register_fake_addr_expr): Ditto.
11815 (chkp_build_addr_expr): Add check for hard reg case.
11816 (chkp_parse_array_and_component_ref): Ditto.
11817 (chkp_find_bounds_1): Ditto.
11818 (chkp_process_stmt): Don't generate bounds store for
11821 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11823 * predict.c (maybe_hot_bb_p): Do not check profile status.
11824 (maybe_hot_edge_p): Likewise.
11825 (probably_never_executed): Check for zero counts even if profile
11827 (unlikely_executed_edge_p): New function.
11828 (unlikely_executed_stmt_p): New function.
11829 (unlikely_executed_bb_p): New function.
11830 (set_even_probabilities): Use unlikely predicates.
11831 (combine_predictions_for_bb): Likewise.
11832 (predict_paths_for_bb): Likewise.
11833 (predict_paths_leading_to_edge): Likewise.
11834 (determine_unlikely_bbs): New function.
11835 (estimate_bb_frequencies): Use it.
11836 (compute_function_frequency): Use zero counts even if profile is
11838 * profile-count.h: Fix typo.
11840 2017-08-08 Julia Koval <julia.koval@intel.com>
11842 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11843 _mm512_mask_cvtsepi16_storeu_epi8,
11844 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11845 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11846 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11847 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11848 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11849 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11850 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11851 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11852 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11853 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11854 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11855 __builtin_ia32_pmovuswb256mem_mask,
11856 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11857 __builtin_ia32_pmovwb512mem_mask): New builtins.
11859 2017-08-08 Julia Koval <julia.koval@intel.com>
11861 PR target/73350,80862
11862 * config/i386/subst.md (round): Fix round pattern.
11863 * config/i386/i386.c (ix86_erase_embedded_rounding):
11864 Fix erasing rounding for the fixed pattern.
11866 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11868 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11870 2017-06-08 Martin Liska <mliska@suse.cz>
11872 PR gcov-profile/80911
11873 * gcov.c (block_info::block_info): New constructor.
11875 2017-06-07 Carl Love <cel@us.ibm.com>
11877 * config/rs6000/rs6000-c: The return type of the following
11878 built-in functions was implemented as int not long long. Fix sign
11879 of return value for the unsigned version of vec_mulo and vec_mule.
11880 vector unsigned long long vec_bperm (vector unsigned long long,
11881 vector unsigned char)
11882 vector signed long long vec_mule (vector signed int,
11884 vector unsigned long long vec_mule (vector unsigned int,
11885 vector unsigned int)
11886 vector signed long long vec_mulo (vector signed int,
11888 vector unsigned long long vec_mulo (vector unsigned int,
11889 vector unsigned int)
11890 * doc/extend.texi: Fix the documentation for the built-in
11893 2017-06-07 Carl Love <cel@us.ibm.com>
11896 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11899 2017-06-07 Carl Love <cel@us.ibm.com>
11901 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11902 support, Generate doublehv for signed int/float for BE case only.
11904 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11906 * doc/invoke.texi (mcx16): Rewrite.
11908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11910 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11911 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11912 *mov<mode>_softfloat, and an anonymous splitter): Use
11913 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11915 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11917 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11919 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11920 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11921 (rs6000_debug_reg_global): Adjust.
11922 (rs6000_init_hard_regno_mode_ok): Adjust.
11923 (rs6000_dbx_register_number): Adjust.
11924 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11925 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11926 Remove SPE_ACC and SPEFSCR.
11927 (REG_ALLOC_ORDER): Ditto.
11928 (FRAME_POINTER_REGNUM): Change to 111.
11929 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11930 (REG_CLASS_NAMES): Ditto.
11931 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11932 (REGISTER_NAMES): Ditto.
11933 (ADDITIONAL_REG_NAMES): Ditto.
11934 (rs6000_reg_names): Ditto.
11935 * config/rs6000/rs6000.md: Renumber some register number
11938 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11940 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11942 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11943 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11945 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11946 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11947 Delete the SPE high registers.
11948 (REG_ALLOC_ORDER): Ditto.
11949 (enum reg_class): Remove SPE_HIGH_REGS.
11950 (REG_CLASS_NAMES): Ditto.
11951 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11952 (REGISTER_NAMES): Ditto.
11953 (rs6000_reg_names): Ditto.
11954 * doc/tm.texi.in: Remove SPE as example.
11955 * doc/tm.texi: Regenerate.
11957 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11959 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11960 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11961 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11962 * config/rs6000/rs6000.md (type): Remove "brinc".
11964 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11966 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11967 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11968 * config/rs6000/linuxspe.h: Delete file.
11969 * config/rs6000/rs6000.md: Don't include spe.md.
11970 * config/rs6000/spe.h: Delete file.
11971 * config/rs6000/spe.md: Delete file.
11972 * config/rs6000/t-rs6000: Remove spe.md.
11974 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11976 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11977 (reg_or_none500mem_operand): Delete.
11978 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11979 instead of reg_or_none500mem_operand.
11981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11983 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11984 handling of SPE flags.
11985 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11987 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11989 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11991 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11992 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11993 paired_divv2sf3): Similar.
11994 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11995 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11996 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11998 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11999 Rename the paired_* instruction patterns.
12000 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12002 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12003 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12004 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12005 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12006 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12007 PAIRED_VECTOR_MODE.
12008 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12009 (spe_func_has_64bit_regs_p): Delete.
12010 (spe_expand_predicate_builtin): Delete.
12011 (spe_expand_evsel_builtin): Delete.
12012 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12013 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12014 (invalid_e500_subreg): Delete.
12015 (rs6000_legitimize_address): Always force_reg op2 as well, for
12016 paired single memory accesses.
12017 (rs6000_member_type_forces_blk): Delete.
12018 (rs6000_spe_function_arg): Delete.
12019 (rs6000_expand_unop_builtin): Delete SPE handling.
12020 (rs6000_expand_binop_builtin): Ditto.
12021 (spe_expand_stv_builtin): Delete.
12022 (bdesc_2arg_spe): Delete.
12023 (spe_expand_builtin): Delete.
12024 (spe_expand_predicate_builtin): Delete.
12025 (spe_expand_evsel_builtin): Delete.
12026 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12027 (spe_init_builtins): Delete.
12028 (spe_func_has_64bit_regs_p): Delete.
12029 (savres_routine_name): Delete "info" parameter. Adjust callers.
12030 (rs6000_emit_stack_reset): Ditto.
12031 (rs6000_dwarf_register_span): Delete.
12032 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12033 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12034 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12036 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12038 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12039 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12040 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12041 mulv2sf3, divv2sf3): Delete expanders.
12043 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12045 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12047 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12049 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12050 * config/rs6000/rs6000.c: Ditto.
12052 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12054 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12055 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12056 comparison_operator.
12058 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12060 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12061 * config/rs6000/rs6000.opt: Ditto.
12062 * config/rs6000/t-rtems: Ditto.
12064 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12066 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12067 TARGET_E500_SINGLE by 0, simplify.
12068 * config/rs6000/rs6000.c: Ditto.
12069 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12070 (spe_build_register_parallel): Delete.
12071 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12072 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12073 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12074 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12075 (E500_CONVERT): Delete.
12076 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12080 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12081 * config/rs6000/dfp.md: Ditto.
12082 (negdd2, *negdd2_fpr): Merge.
12083 (absdd2, *absdd2_fpr): Merge.
12084 (negtd2, *negtd2_fpr): Merge.
12085 (abstd2, *abstd2_fpr): Merge.
12086 * config/rs6000/e500.h: Delete file.
12087 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12088 TARGET_FPRS by 1 and simplify.
12089 * config/rs6000/rs6000-c.c: Ditto.
12090 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12091 TARGET_DF_SPE by 0.
12092 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12094 * config/rs6000/rs6000.md: Ditto.
12095 (floatdidf2, *floatdidf2_fpr): Merge.
12096 (move_from_CR_gt_bit): Delete.
12097 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12098 (E500_CR_IOR_COMPARE): Delete.
12099 (All patterns that require !TARGET_FPRS): Delete.
12100 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12102 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12104 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12106 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12108 * graphds.c (add_edge): Intitialize edge's attached data.
12109 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12110 pointer parameter. Call pointed function on each edge during
12111 graph traversing. Skip traversing the edge when the function
12113 (graphds_dfs, graphds_scc): Ditto.
12114 (for_each_edge): New parameter. Pass the new parameter to callback
12116 * graphds.h (skip_edge_callback): New function pointer type.
12117 (graphds_dfs, graphds_scc): New function pointer parameter.
12118 (graphds_edge_callback, for_each_edge): New parameter.
12120 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12122 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12123 out code checking if runtime alias check is possible to below ...
12124 Call the new function.
12125 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12126 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12128 2017-06-07 Marek Polacek <polacek@redhat.com>
12131 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12132 TYPE_OVERFLOW_WRAPS check.
12134 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12136 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12137 if versioning is required.
12138 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12139 peeling with the check for versioning.
12141 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12143 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12144 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12145 Set true to new parameter if new ssa variable is defined.
12146 (vect_gen_vector_loop_niters): Refactor. Set range information
12147 for the new vector loop bound variable.
12148 (vect_do_peeling): Ditto.
12150 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12152 * tree-affine.c (ssa.h): Include header file.
12153 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12154 has wrapping overflow behavior.
12156 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12158 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12160 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12162 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12163 (tree_to_aff_combination): ... here.
12165 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12167 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12168 reg_pressure model function.
12169 (ivopts_global_cost_for_size): Delete.
12170 (determine_set_costs, iv_ca_recount_cost): Call new model function
12171 ivopts_estimate_reg_pressure.
12173 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12175 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12176 expensive than udiv. Remove floating point cases from mod.
12178 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12180 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12181 Increase idiv cost.
12183 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12185 * config/aarch64/aarch64.md
12186 (copysignsf3): Fix mask generation.
12188 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12190 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12192 (class dump_manager): Add register_dumps method.
12193 * dumpfile.c: Include langhooks.h.
12194 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12195 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12196 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12197 (dump_manager::dump_register): Start with 512 entries instead of 32.
12198 (dump_manager::register_dumps): New method.
12199 * toplev.c (general_init): Instead of invoking register_dumps
12200 langhook, invoke register_dumps method on the dump manager.
12201 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12204 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12206 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12207 Say that # requires an associated define_split to exist, and that
12208 the define_split must be suitable for use after register allocation.
12210 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12212 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12213 (compute_outgoing_frequencies): Also initialize zero counts.
12214 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12215 around loops; preserve more of profile when nothing changes.
12217 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12219 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12221 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12222 * config/arm/arm-cpu-cdata.h: Regenerate.
12223 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12224 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12225 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12226 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12227 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12229 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12231 * config/arm/t-rmprofile: Likewise.
12232 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12234 2017-06-06 David S. Miller <davem@davemloft.net>
12237 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12238 blockage if function uses alloca.
12240 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12242 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12243 New "uid" fields to hold pretty-print IDs of group and ref.
12244 Memory references are now identified as <group_id>:<ref_id>
12245 instead of using [random] addresses.
12246 (dump_mem_details): Simplify, no functional change.
12247 (dump_mem_ref): Simplify and make output more concise.
12248 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12249 (find_or_create_group): Initialize group uid.
12250 (record_ref): Initialize ref uid. Improve debug output.
12251 (prune_group_by_reuse, should_issue_prefetch_p,)
12252 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12253 (mark_nontemporal_store, determine_loop_nest_reuse):
12254 Improve debug output.
12256 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12258 * dbgcnt.def (prefetch): New debug counter.
12259 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12260 (schedule_prefetches): Stop issueing prefetches if debug counter
12263 2017-06-06 Tom de Vries <tom@codesourcery.com>
12265 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12266 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12268 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12270 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12271 Use aarch64_reg_or_zero predicate for operand 4.
12272 (aarch64_compare_and_swap<mode> define_insn_and_split):
12273 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12274 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12276 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12278 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12279 (arm_compute_save_core_reg_mask): This.
12280 (thumb1_compute_save_reg_mask): Rename into ...
12281 (thumb1_compute_save_core_reg_mask): This.
12282 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12283 (arm_compute_frame_layout): Likewise.
12285 2017-06-06 Richard Biener <rguenther@suse.de>
12287 PR tree-optimization/80974
12288 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12289 keep or clear leaders SSA info.
12291 2017-06-06 Tom de Vries <tom@codesourcery.com>
12293 * config/nvptx/nvptx.c (split_mode_p): New function.
12294 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12296 2017-06-06 Tom de Vries <tom@codesourcery.com>
12298 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12300 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12303 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12306 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12308 * shrink-wrap.c (handle_simple_exit): Update profile.
12309 (try_shrink_wrapping): Upate profile.
12311 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12313 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12314 (tree_guess_outgoing_edge_probabilities): New.
12315 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12316 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12318 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12320 * ipa-split.c (split_function): Initialize return bb profile.
12322 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12324 * profile.c (compute_branch_probabilities): Also initialize
12325 EXIT_BLOCK profile.
12327 2017-06-06 Richard Biener <rguenther@suse.de>
12329 PR tree-optimization/80928
12330 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12331 (vect_analyze_loop_operations): Properly guard analysis for
12333 (vect_transform_loop): Likewise.
12334 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12335 (vect_model_induction_cost): Do not cost for pure SLP.
12336 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12337 of induction in inner loop vectorization.
12338 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12339 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12340 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12342 (vect_analyze_slp_cost_1): Cost induction.
12343 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12344 (vect_get_slp_vect_defs): Likewise.
12345 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12346 (vect_transform_stmt): Handle SLP reductions.
12347 * tree-vectorizer.h (vectorizable_induction): Adjust.
12349 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12351 * config/rs6000/rs6000.c (make_resolver_func): Update
12352 init_lowered_empty_function call.
12354 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12356 * doc/invoke.texi: Document the -fprofile-abs-path option.
12357 * common.opt (fprofile-abs-path): New option.
12358 * gcov-io.h (gcov_write_filename): Declare.
12359 * gcov-io.c (gcov_write_filename): New function.
12360 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12361 * profile.c (output_location): Likewise.
12363 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12365 * shring-wrap.c: Revert accidental commit.
12367 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12369 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12371 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12373 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12375 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12376 profile in callgraph edge.
12377 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12378 (apply_scale): Likewise.
12379 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12380 Also scale profile when inlining function with zero profile.
12381 (initialize_cfun): Update exit block profile even when it is zero.
12382 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12383 when profile is read.
12385 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12387 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12388 (CLONE_*): New constants to define the processors we can generate
12389 code for with the target_clone attribute.
12390 (rs6000_clone_map): New array to identify which clone processors
12391 the current program is running on.
12392 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12393 target_clone attribute.
12394 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12395 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12396 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12397 (cpu_expand_builtin): Add support for target_clone attribute.
12398 (rs6000_valid_attribute_p): Allow "default" attribute.
12399 (get_decl_name): New debug function to simplify printing the
12400 current function name in debugging statements.
12401 (rs6000_clone_priority): New functions to support the target_clone
12402 attribute, and be able to generate code to switch between ISA 2.05
12403 through ISA 3.0 (power6 through power9).
12404 (rs6000_compare_version_priority): Likewise.
12405 (rs6000_get_function_versions_dispatcher): Likewise.
12406 (make_resolver_func): Likewise.
12407 (add_condition_to_bb): Likewise.
12408 (dispatch_function_versions): Likewise.
12409 (rs6000_generate_version_dispatcher_body): Likewise.
12410 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12411 (fusion_gpr_load_p): Fix a spacing issue.
12412 * doc/extend.texi (Common Function Attributes): Document that the
12413 PowerPC supports the target_clone attribute.
12415 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12417 * config/arm/arm.h: explain F symbol found in description of ARM
12418 register allocation in its legend.
12420 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12422 * config/mips/frame-header-opt.c: Include profile-count.h.
12423 * config/riscv/riscv.c: Include profile-count.h
12425 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12427 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12429 (sm_set_flag_if_changed): Add bbs field.
12430 (execute_sm_if_changed_flag_set): Pass BBS.
12431 (execute_sm): Update.
12433 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12435 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12438 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12440 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12441 (peephole2): New peephole2 to emit the above.
12442 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12444 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12446 * config/aarch64/aarch64.c (define_peephole2 above
12447 *sub_<shift>_<mode>): New peephole.
12449 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12451 * config/i386/i386.c (make_resolver_func): Update.
12452 * Makefile.in: Add profile-count.h and profile-count.o
12453 * auto-profile.c (afdo_indirect_call): Update to new API.
12454 (afdo_set_bb_count): Update.
12455 (afdo_propagate_edge): Update.
12456 (afdo_propagate_circuit): Update.
12457 (afdo_calculate_branch_prob): Update.
12458 (afdo_annotate_cfg): Update.
12459 * basic-block.h: Include profile-count.h
12460 (struct edge_def): Turn count to profile_count.
12461 (struct basic_block_def): Likewie.
12462 (REG_BR_PROB_BASE): Move to profile-count.h
12463 (RDIV): Move to profile-count.h
12464 * bb-reorder.c (max_entry_count): Turn to profile_count.
12465 (find_traces): Update.
12466 (rotate_loop):Update.
12467 (connect_traces):Update.
12468 (sanitize_hot_paths):Update.
12469 * bt-load.c (migrate_btr_defs): Update.
12470 * cfg.c (RDIV): Remove.
12471 (init_flow): Use alloc_block.
12472 (alloc_block): Uninitialize count.
12473 (unchecked_make_edge): Uninitialize count.
12474 (check_bb_profile): Update.
12475 (dump_edge_info): Update.
12476 (dump_bb_info): Update.
12477 (update_bb_profile_for_threading): Update.
12478 (scale_bbs_frequencies_int): Update.
12479 (scale_bbs_frequencies_gcov_type): Update.
12480 (scale_bbs_frequencies_profile_count): New.
12481 * cfg.h (update_bb_profile_for_threading): Update.
12482 (scale_bbs_frequencies_profile_count): Declare.
12483 * cfgbuild.c (compute_outgoing_frequencies): Update.
12484 (find_many_sub_basic_blocks): Update.
12485 * cfgcleanup.c (try_forward_edges): Update.
12486 (try_crossjump_to_edge): Update.
12487 * cfgexpand.c (expand_gimple_tailcall): Update.
12488 (construct_exit_block): Update.
12489 * cfghooks.c (verify_flow_info): Update.
12490 (dump_bb_for_graph): Update.
12491 (split_edge): Update.
12492 (make_forwarder_block): Update.
12493 (duplicate_block): Update.
12494 (account_profile_record): Update.
12495 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12496 (get_estimated_loop_iterations): Update.
12497 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12498 (single_likely_exit): Update.
12499 * cfgloopmanip.c (scale_loop_profile): Update.
12501 (set_zero_probability): Update.
12502 (lv_adjust_loop_entry_edge): Update.
12503 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12504 (purge_dead_edges): Update.
12505 (rtl_account_profile_record): Update.
12506 * cgraph.c (cgraph_node::create): Uninitialize count.
12507 (symbol_table::create_edge): Uninitialize count.
12508 (cgraph_update_edges_for_call_stmt_node): Update.
12509 (cgraph_edge::dump_edge_flags): Update.
12510 (cgraph_node::dump): Update.
12511 (cgraph_edge::maybe_hot_p): Update.
12512 * cgraph.h: Include profile-count.h
12513 (create_clone), create_edge, create_indirect_edge): Update.
12514 (cgraph_node): Turn count to profile_count.
12515 (cgraph_edge0: Likewise.
12516 (make_speculative, clone): Update.
12517 (create_edge): Update.
12518 (init_lowered_empty_function): Update.
12519 * cgraphclones.c (cgraph_edge::clone): Update.
12520 (duplicate_thunk_for_node): Update.
12521 (cgraph_node::create_clone): Update.
12522 * cgraphunit.c (cgraph_node::analyze): Update.
12523 (cgraph_node::expand_thunk): Update.
12524 * final.c (dump_basic_block_info): Update.
12525 * gimple-streamer-in.c (input_bb): Update.
12526 * gimple-streamer-out.c (output_bb): Update.
12527 * graphite.c (print_global_statistics): Update.
12528 (print_graphite_scop_statistics): Update.
12529 * hsa-brig.c: Include basic-block.h.
12530 * hsa-dump.c: Include basic-block.h.
12531 * hsa-gen.c (T sum_slice): Update.
12532 (convert_switch_statements):Update.
12533 * hsa-regalloc.c: Include basic-block.h.
12534 * ipa-chkp.c (chkp_produce_thunks): Update.
12535 * ipa-cp.c (struct caller_statistics): Update.
12536 (init_caller_stats): Update.
12537 (gather_caller_stats): Update.
12538 (ipcp_cloning_candidate_p): Update.
12539 (good_cloning_opportunity_p): Update.
12540 (get_info_about_necessary_edges): Update.
12541 (dump_profile_updates): Update.
12542 (update_profiling_info): Update.
12543 (update_specialized_profile): Update.
12544 (perhaps_add_new_callers): Update.
12545 (decide_about_value): Update.
12546 (ipa_cp_c_finalize): Update.
12547 * ipa-devirt.c (struct odr_type_warn_count): Update.
12548 (struct decl_warn_count): Update.
12549 (struct final_warning_record): Update.
12550 (possible_polymorphic_call_targets): Update.
12551 (ipa_devirt): Update.
12552 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12553 * ipa-icf.c (sem_function::merge): Update.
12554 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12555 * ipa-inline.c (compute_uninlined_call_time): Update.
12556 (compute_inlined_call_time): Update.
12557 (want_inline_small_function_p): Update.
12558 (want_inline_self_recursive_call_p): Update.
12559 (edge_badness): Update.
12560 (lookup_recursive_calls): Update.
12561 (recursive_inlining): Update.
12562 (inline_small_functions): Update.
12563 (dump_overall_stats): Update.
12564 (dump_inline_stats): Update.
12565 * ipa-profile.c (ipa_profile_generate_summary): Update.
12566 (ipa_propagate_frequency): Update.
12567 (ipa_profile): Update.
12568 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12569 * ipa-utils.c (ipa_merge_profiles): Update.
12570 * loop-doloop.c (doloop_modify): Update.
12571 * loop-unroll.c (report_unroll): Update.
12572 (unroll_loop_runtime_iterations): Update.
12573 * lto-cgraph.c (lto_output_edge): Update.
12574 (lto_output_node): Update.
12575 (input_node): Update.
12576 (input_edge): Update.
12577 (merge_profile_summaries): Update.
12578 * lto-streamer-in.c (input_cfg): Update.
12579 * lto-streamer-out.c (output_cfg): Update.
12580 * mcf.c (create_fixup_graph): Update.
12581 (adjust_cfg_counts): Update.
12582 (sum_edge_counts): Update.
12583 * modulo-sched.c (sms_schedule): Update.
12584 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12585 * predict.c (maybe_hot_count_p): Update.
12586 (probably_never_executed): Update.
12587 (dump_prediction): Update.
12588 (combine_predictions_for_bb): Update.
12589 (propagate_freq): Update.
12590 (handle_missing_profiles): Update.
12591 (counts_to_freqs): Update.
12592 (rebuild_frequencies): Update.
12593 (force_edge_cold): Update.
12594 * predict.h: Include profile-count.h
12595 (maybe_hot_count_p, counts_to_freqs): UPdate.
12596 * print-rtl-function.c: Do not include cfg.h
12597 * print-rtl.c: Include basic-block.h
12598 * profile-count.c: New file.
12599 * profile-count.h: New file.
12600 * profile.c (is_edge_inconsistent): Update.
12601 (correct_negative_edge_counts): Update.
12602 (is_inconsistent): Update.
12603 (set_bb_counts): Update.
12604 (read_profile_edge_counts): Update.
12605 (compute_frequency_overlap): Update.
12606 (compute_branch_probabilities): Update; Initialize and deinitialize
12608 (branch_prob): Update.
12609 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12610 (edge_gcov_count): New.
12611 (bb_gcov_count): New.
12612 * shrink-wrap.c (try_shrink_wrapping): Update.
12613 * tracer.c (better_p): Update.
12614 * trans-mem.c (expand_transaction): Update.
12615 (ipa_tm_insert_irr_call): Update.
12616 (ipa_tm_insert_gettmclone_call): Update.
12617 * tree-call-cdce.c: Update.
12618 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12619 (gimple_duplicate_sese_tail): Update.
12620 (gimple_account_profile_record): Update.
12621 (execute_fixup_cfg): Update.
12622 * tree-inline.c (copy_bb): Update.
12623 (copy_edges_for_bb): Update.
12624 (initialize_cfun): Update.
12625 (freqs_to_counts): Update.
12626 (copy_cfg_body): Update.
12627 (expand_call_inline): Update.
12628 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12629 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12630 (try_unroll_loop_completely): Update.
12631 (try_peel_loop): Update.
12632 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12633 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12634 * tree-ssa-loop-split.c (connect_loops): Update.
12635 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12636 * tree-ssa-reassoc.c (branch_fixup): Update.
12637 * tree-ssa-tail-merge.c (replace_block_by): Update.
12638 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12639 (compute_path_counts): Update.
12640 (update_profile): Update.
12641 (recompute_probabilities): Update.
12642 (update_joiner_offpath_counts): Update.
12643 (estimated_freqs_path): Update.
12644 (freqs_to_counts_path): Update.
12645 (clear_counts_path): Update.
12646 (ssa_fix_duplicate_block_edges): Update.
12647 (duplicate_thread_path): Update.
12648 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12649 (struct switch_conv_info): Update.
12650 * tree-tailcall.c (decrease_profile): Update.
12651 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12652 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12653 * value-prof.c (check_counter): Update.
12654 (gimple_divmod_fixed_value): Update.
12655 (gimple_mod_pow2): Update.
12656 (gimple_mod_subtract): Update.
12657 (gimple_ic_transform): Update.
12658 (gimple_stringop_fixed_value): Update.
12659 * value-prof.h (gimple_ic): Update.
12661 2017-06-02 Carl Love <cel@us.ibm.com>
12663 * config/rs6000/rs6000-c: Add support for built-in functions
12664 vector double vec_doublee (vector signed int);
12665 vector double vec_doublee (vector unsigned int);
12666 vector double vec_doublee (vector float);
12667 vector double vec_doubleh (vector signed int);
12668 vector double vec_doubleh (vector unsigned int);
12669 vector double vec_doubleh (vector float);
12670 vector double vec_doublel (vector signed int);
12671 vector double vec_doublel (vector unsigned int);
12672 vector double vec_doublel (vector float);
12673 vector double vec_doubleo (vector signed int);
12674 vector double vec_doubleo (vector unsigned int);
12675 vector double vec_doubleo (vector float);.
12676 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12677 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12679 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12680 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12681 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12683 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12684 vec_doublel, vec_doubleh.
12685 * doc/extend.texi: Update the built-in documentation file for the
12686 new built-in functions.
12688 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12691 * ipa-inline-analysis.c (free_growth_caches): Set
12692 edge_removal_hook_holder to NULL after removing it.
12694 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12696 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12697 comparision with zero.
12699 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12700 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12701 for early expansion of vec_min and vec_max builtins.
12702 (builtin_function_type): Add min/max unsigned variants to those
12703 identified as having unsigned arguments.
12705 2017-06-02 Olivier Hainque <hainque@adacore.com>
12707 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12709 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12711 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12712 Use VALL_F16 iterator rather than VALL.
12714 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12716 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12717 Emit CBNZ inside loop when doing a strong exchange and comparing
12718 against zero. Generate the CC flags after the loop.
12720 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12722 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12723 (dl_section_ref): New.
12724 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12725 On AIX, append an expression to subtract the size of the
12726 section length to dl_section_ref.
12728 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12731 for early expansion of vector absolute builtins.
12733 2017-06-02 Richard Biener <rguenther@suse.de>
12735 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12736 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12738 2017-06-02 Richard Biener <rguenther@suse.de>
12740 PR tree-optimization/80948
12741 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12742 stmt order as well.
12744 2017-06-02 Richard Biener <rguenther@suse.de>
12746 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12748 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12749 for inductions as relevant.
12751 2017-06-02 Richard Biener <rguenther@suse.de>
12753 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12754 (vectorizable_induction): ... this. Remove dead code.
12756 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12758 * builtins. (expand_builtin_alloca): Remove second parameter and
12759 infer its value from the first parameter instead.
12760 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12762 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12764 PR rtl-optimization/80903
12765 * loop-doloop.c (add_test): Unshare sequence.
12767 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12769 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12771 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12773 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12775 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12776 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12777 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12778 (xlogue_layout::get_stub_rtx): Make static.
12779 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12780 (xlogue_layout::compute_stub_managed_regs): Rename to...
12781 (xlogue_layout::count_stub_managed_regs): ...this.
12782 (xlogue_layout::is_stub_managed_reg): New function.
12783 (xlogue_layout::m_stub_names): Rename to...
12784 (xlogue_layout::s_stub_names): ...this, make static.
12785 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12786 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12787 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12788 xlogue_layout::s_stub_names): Instantiate statics.
12789 (stub_managed_regs): Remove.
12790 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12791 (disable_call_ms2sysv_xlogues): Rename to...
12792 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
12793 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
12795 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
12796 change after reload_completed.
12797 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
12799 (ix86_expand_prologue): Likewise.
12800 (ix86_expand_epilogue): Likewise.
12801 (ix86_expand_split_stack_prologue): Likewise.
12802 (ix86_compute_frame_layout): Remove frame parameter ...
12803 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
12804 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
12806 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
12807 (ix86_frame): Move from here ...
12808 * config/i386/i386.h (ix86_frame): ... to here.
12809 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
12810 complete ix86_frame data structure instead. Remove some_ld_name.
12812 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12814 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
12815 symbols that hold a DECL_VALUE_EXPR.
12817 2017-06-01 Martin Jambor <mjambor@suse.cz>
12819 PR tree-optimization/80898
12820 * tree-sra.c (process_subtree_disqualification): Removed.
12821 (disqualify_candidate): Do not acll
12822 process_subtree_disqualification.
12823 (subtree_mark_written_and_enqueue): New function.
12824 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12825 RHS has been disqualified and re-queue LHS if necessary. Apart
12826 from that, ignore disqualified RHS.
12828 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12830 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12831 address fetch for z10 or later.
12833 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12835 * config/arc/arc.md (tst_movb): Add guard when splitting.
12837 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12839 * config/arc/arc.c (arc_can_eliminate): Test against
12840 arc_frame_pointer_needed.
12842 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12844 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12845 to prevent store reordering.
12846 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12847 (type): Add block type.
12848 (stack_tie): Define special instruction to be used in
12851 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12853 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12854 constraint. It is not valid for the pattern.
12855 (noncommutative_binary_comparison): Likewise.
12857 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12859 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12862 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12864 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12865 be used by the reg-alloc.
12867 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12869 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12870 reg-alloc when having mul64 or mul32x16 instructions.
12871 (mulsidi3): Likewise.
12872 (umulsidi3): Likewise.
12873 (mulsi32x16): New pattern.
12874 (mulsi64): Likewise.
12875 (mulsidi64): Likewise.
12876 (umulsidi64): Likewise.
12877 (MUL32x16_REG): Define.
12878 (mul64_600): Use MUL32x16_REG.
12879 (mac64_600): Likewise.
12880 (umul64_600): Likewise.
12881 (umac64_600): Likewise.
12883 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12885 * config/arc/arc.md (mulsi3_700): Make it commutative.
12887 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12889 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12891 (*sign_extendsidi2_insn): Likewise for movstosw.
12893 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12895 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12896 the type of the input discriminant value. Convert the
12897 discriminant value of signedness vary.
12899 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12901 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12902 Add to -Wall section.
12904 2017-06-01 Richard Biener <rguenther@suse.de>
12906 PR middle-end/66313
12907 * fold-const.c (fold_plusminus_mult_expr): If the factored
12908 factor may be zero use a wrapping type for the inner operation.
12909 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12910 and handle moved defs.
12911 (process_assignment): Properly guard the unary op case. Return a
12912 tri-state indicating that moving the stmt before the call may allow
12913 to continue. Pass through to_move.
12914 (find_tail_calls): Handle moving unrelated defs before
12917 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12920 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12921 splitter result in the canonical way.
12923 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12925 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12926 also for 32bit target. Update insn attributes.
12927 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12929 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12931 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12932 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12933 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12934 (_mm_maskz_min_ss): New intrinsics.
12936 2017-05-31 Martin Liska <mliska@suse.cz>
12938 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12939 Change comment style to one we normally use.
12940 (vectorizable_reduction): Likewise.
12941 (vectorizable_induction): Likewise.
12942 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12943 (vectorizable_call): Likewise.
12944 (vectorizable_simd_clone_call): Likewise.
12945 (vectorizable_conversion): Likewise.
12946 (vectorizable_assignment): Likewise.
12947 (vectorizable_shift): Likewise.
12948 (vectorizable_operation): Likewise.
12949 (vectorizable_store): Likewise.
12950 (vectorizable_load): Likewise.
12951 * tree-vectorizer.h: Likewise.
12953 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12955 * passes.c (emergency_dump_function): New.
12956 * tree-pass.h (emergency_dump_function): Declare.
12957 * plugin.c (plugins_internal_error_function): Remove.
12958 * plugin.h (plugins_internal_error_function): Remove declaration.
12959 * toplev.c (internal_error_function): New static function. Use it...
12960 (general_init): ...here.
12962 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12964 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12965 (arc_rtx_costs): Add costs for new patterns.
12966 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12967 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12969 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12971 * tree-ssa-strlen.c (get_next_strinfo): New function.
12972 (get_stridx_plus_constant): Use it.
12973 (zero_length_string): Likewise.
12974 (adjust_related_strinfos): Likewise.
12975 (adjust_last_stmt): Likewise.
12977 2017-05-31 Richard Biener <rguenther@suse.de>
12980 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12981 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12983 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12985 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12986 loop_vinfo argument and use of dependence distance vectors.
12987 Check instead whether the two references differ only in their
12988 initial value and assume that they have the same alignment if the
12989 difference is a multiple of the vector alignment.
12990 (vect_analyze_data_refs_alignment): Update call accordingly.
12992 2017-05-31 Martin Liska <mliska@suse.cz>
12995 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12997 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12999 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13000 (create_intersect_range_checks): Move from ...
13001 * tree-data-ref.c (create_intersect_range_checks_index)
13002 (create_intersect_range_checks): ... to here.
13003 (create_runtime_alias_checks): New function factored from ...
13004 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13005 here. Call above function.
13006 * tree-data-ref.h (create_runtime_alias_checks): New function.
13008 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13010 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13011 segment length for dr_b and compute it in wide_int.
13013 2017-05-31 Richard Biener <rguenther@suse.de>
13015 PR tree-optimization/80906
13016 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13017 and pass through iv_map.
13018 (copy_bb_and_scalar_dependences): Adjust.
13019 (translate_pending_phi_nodes): Likewise.
13020 (copy_loop_close_phi_args): Handle code-generating IVs instead
13023 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13025 * diagnostic-color.c (color_dict): Add "type-diff".
13026 (parse_gcc_colors): Update comment.
13027 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13028 -fdiagnostics-show-template-tree and -fno-elide-type.
13029 (GCC_COLORS): Add type-diff to example.
13031 (-fdiagnostics-show-template-tree): New.
13032 (-fno-elide-type): New.
13033 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13034 the pp_format_decoder callback. Call any m_format_postprocessor's
13036 (pretty_printer::pretty_printer): Initialize
13037 m_format_postprocessor.
13038 (pretty_printer::~pretty_printer): Delete any
13039 m_format_postprocessor.
13040 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13041 (class format_postprocessor): New class.
13042 (struct pretty_printer::format_decoder): Document the new parameters.
13043 (struct pretty_printer::m_format_postprocessor): New field.
13044 * tree-diagnostic.c (default_tree_printer): Update for new
13045 bool and const char ** params.
13046 * tree-diagnostic.h (default_tree_printer): Likewise.
13048 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13050 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13051 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13052 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13053 rs6000_gen_cell_microcode code.
13054 (rs6000_final_prescan_insn): Delete.
13055 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13056 "warn-cell-microcode" entries.
13057 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13058 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13059 throughout. Change cc_reg_not_micro_cr0_operand to
13060 cc_reg_not_cr0_operand throughout.
13061 (*extendhi<mode>2_noload): Delete.
13062 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13063 (mwarn-cell-microcode): Delete.
13064 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13065 -mgen-cell-microcode and -mwarn-cell-microcode.
13067 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13070 * config/i386/constraints.md (Yd): New constraint.
13072 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13073 and (?Yd, r) alternatives. Update insn attributes.
13074 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13075 and (?*Yd, r) alternatives. Update insn attributes.
13076 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13078 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13080 * gimplify.c (gimplify_modify_expr): Don't create a
13081 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13084 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13086 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13088 2017-05-30 Richard Biener <rguenther@suse.de>
13090 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13091 and reduc_def fields.
13092 (STMT_VINFO_REDUC_TYPE): New define.
13093 (STMT_VINFO_REDUC_DEF): Likewise.
13094 (vect_force_simple_reduction): Adjust prototype.
13095 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13096 (vect_is_simple_reduction): Remove check_reduction argument.
13097 (vect_force_simple_reduction): Adjust and set
13098 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13099 (vectorizable_reduction): Do not re-do reduction analysis
13100 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13101 * tree-parloops.c (gather_scalar_reductions): Adjust.
13103 2017-05-30 Richard Biener <rguenther@suse.de>
13105 PR middle-end/80901
13106 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13109 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13111 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13112 Introduce unknown_misalignment parameter and remove vf.
13113 (vect_peeling_hash_get_lowest_cost):
13114 Pass unknown_misalignment parameter.
13115 (vect_enhance_data_refs_alignment):
13116 Fix unsupportable data ref treatment.
13118 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13120 * tree-vect-data-refs.c (vect_get_data_access_cost):
13121 Workaround for SLP handling.
13122 (vect_enhance_data_refs_alignment):
13123 Compute costs for doing no peeling at all, compare to the best
13124 peeling costs so far and avoid peeling if cheaper.
13126 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13128 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13129 Return peeling info and set costs to zero for unlimited cost
13131 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13132 with unknown misalignment. Compute and costs for unknown
13133 misalignment, compare them to the costs for known misalignment
13134 and choose the cheapest for peeling.
13136 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13138 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13139 (vect_get_peeling_costs_all_drs): Create function.
13140 (vect_peeling_hash_get_lowest_cost):
13141 Use vect_get_peeling_costs_all_drs.
13142 (vect_peeling_supportable): Create function.
13143 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13145 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13147 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13148 DR_HAS_NEGATIVE_STEP.
13149 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13150 (vect_enhance_data_refs_alignment): Use.
13151 (vect_duplicate_ssa_name_ptr_info): Use.
13152 * tree-vectorizer.h (dr_misalignment): Use.
13153 (known_alignment_for_access_p): Use.
13155 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13158 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13160 (has_section_name): New function.
13162 2017-05-30 Martin Liska <mliska@suse.cz>
13165 * auto-profile.c (get_function_decl_from_block): Fix
13168 2017-05-30 Richard Biener <rguenther@suse.de>
13170 PR middle-end/80876
13171 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13173 2017-05-30 Martin Liska <mliska@suse.cz>
13175 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13176 * dumpfile.h (struct dump_file_info): Remove ctors.
13178 2017-05-30 Martin Liska <mliska@suse.cz>
13180 * predict.def: Fix GNU coding style.
13182 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13184 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13185 Mark 'to' argument with ATTRIBUTE_UNUSED.
13187 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13189 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13190 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13191 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13194 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13196 * doc/install.texi (Options specification): Restore entry of
13197 --enable-sjlj-exceptions.
13199 2017-05-27 Michael Eager <eager@eagercon.com>
13202 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13204 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13206 * config/microblaze/microblaze.h
13207 (FIXED_REGISTERS): Update in macro.
13208 (CALL_USED_REGISTERS): Update in macro.
13210 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13212 * doc/install.texi: Add links to macOS binary distributions.
13214 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13218 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13220 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13222 2017-05-26 Martin Liska <mliska@suse.cz>
13224 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13226 2017-05-26 Martin Liska <mliska@suse.cz>
13228 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13229 always leading ';; '.
13230 (dump_bb_info): Likewise.
13231 (brief_dump_cfg): Likewise.
13232 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13233 * dumpfile.c: Remove usage of TDF_VERBOSE.
13234 * dumpfile.h (enum dump_kind): Likewise.
13235 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13236 * print-tree.c (debug_verbose): Remove.
13237 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13238 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13239 * tree-diagnostic.c (default_tree_printer): Replace
13240 TDF_DIAGNOSTIC with TDF_SLIM.
13242 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13244 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13245 in parameter loop, rather than loop_vinfo.
13246 (create_intersect_range_checks): Ditto.
13247 (vect_create_cond_for_alias_checks): Update call to above functions.
13249 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13251 PR tree-optimization/80815
13252 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13253 for merging runtime alias checks. Handle negative DR_STEPs.
13255 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13257 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13259 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13260 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13261 out code pruning runtime alias checks.
13262 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13263 factored out from above.
13264 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13266 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13268 (prune_runtime_alias_test_list): New decalaration.
13270 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13272 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13273 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13274 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13275 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13276 (operator==, comp_dr_with_seg_len_pair): Ditto.
13277 (vect_prune_runtime_alias_test_list): Ditto.
13279 2017-05-26 Martin Liska <mliska@suse.cz>
13282 * params.def: Bound partial-inlining-entry-probability param.
13284 2017-05-26 Marek Polacek <polacek@redhat.com>
13287 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13290 2017-05-26 Richard Biener <rguenther@suse.de>
13292 PR tree-optimization/80842
13293 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13296 2017-05-26 Richard Biener <rguenther@suse.de>
13298 PR tree-optimization/80844
13299 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13301 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13303 * doc/md.texi (Machine Constraints): Update x86 family
13304 machine constraints section to match 'config/i386/constraints.md'.
13306 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13308 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13310 2017-05-25 Nathan Sidwell <nathan@acm.org>
13312 * doc/invoke.texi (--enable-languages): Update documentation.
13314 2017-05-25 Martin Liska <mliska@suse.cz>
13316 * dumpfile.c: Add TDF_FOLDING.
13317 * dumpfile.h (enum dump_kind): Likewise.
13318 * genmatch.c (dt_simplify::gen_1): Use it.
13320 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13322 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13324 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13326 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13327 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13329 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13331 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13332 * match.pd (X == C): Rewrite it here.
13333 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13334 with_certain_nonzero_bits2): New predicates.
13335 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13337 2017-05-24 Nathan Sidwell <nathan@acm.org>
13339 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13342 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13345 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13347 * config/powerpcspe: New port. Files are copied from the rs6000
13348 port, with "rs6000" in filenames replaced by "powerpcspe".
13350 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13352 PR rtl-optimization/80754
13353 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13355 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13357 * config/sparc/sparc.md (length): Return the correct value for -mflat
13358 sibcalls to match output_sibcall.
13360 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13364 * config/rs6000/rs6000.c (struct machine_function): Add new field
13366 (rs6000_get_separate_components): Init that field, use it.
13367 (rs6000_components_for_bb): Use the field.
13369 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13371 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13373 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13375 PR middle-end/80823
13376 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13378 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13381 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13382 address against address_operand predicate.
13383 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13385 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13387 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13388 parameters passed indirectly.
13390 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13392 * config/i386/i386.md (*movdi_internal): Remove SSE4
13393 alternative 18 (?r, *v). Update insn attributes.
13394 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13395 Update insn attributes.
13396 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13397 Update insn attributes.
13398 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13399 alternative 1 (r, v). Remove isa attribute.
13400 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13401 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13402 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13404 2017-05-23 Tom de Vries <tom@codesourcery.com>
13406 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13409 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13411 * cgraphunit.c (symbol_table::process_new_functions): Update.
13412 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13413 (inline_generate_summary): Rename to ...
13414 (ipa_fn_summary_generate): ... this one.
13415 (inline_read_summary): Rename to ...
13416 (ipa_fn_summary_read): ... this one.
13417 (inline_write_summary): Rename to ...
13418 (ipa_fn_summary_write): ... this one.
13419 (inline_free_summary): Rename to ...
13420 (ipa_free_fn_summary): ... this one.
13421 (pass_data_local_fn_summary, pass_local_fn_summary,
13422 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13423 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13424 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13425 make_pass_ipa_fn_summary): New.
13426 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13427 inline_write_summary, inline_free_summary): Remove.
13428 (ipa_free_fn_summary) : New.
13429 * ipa-inline.c (ipa_inline): Update.
13430 (pass_ipa_inline): Do not generate summaries.
13431 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13433 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13434 and add pass_ipa_fn_summary.
13435 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13437 (make_pass_inline_parameters): Remove.
13439 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13441 * omp-low.c (struct omp_context): Remove "default_kind" member.
13444 * omp-offload.c (execute_oacc_device_lower): Remove the
13445 parallelism dimensions function attributes for unparallelized
13446 OpenACC kernels constructs.
13448 2017-05-23 Martin Liska <mliska@suse.cz>
13450 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13452 (cgraph_edge::make_speculative): Likewise.
13453 (cgraph_edge::resolve_speculation): Likewise.
13454 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13455 (cgraph_node::dump): Likewise.
13456 * cgraph.h: Likewise.
13457 * cgraphunit.c (analyze_functions): Likewise.
13458 (symbol_table::compile): Likewise.
13459 * ipa-cp.c (print_all_lattices): Likewise.
13460 (determine_versionability): Likewise.
13461 (initialize_node_lattices): Likewise.
13462 (ipcp_verify_propagated_values): Likewise.
13463 (estimate_local_effects): Likewise.
13464 (update_profiling_info): Likewise.
13465 (create_specialized_node): Likewise.
13466 (perhaps_add_new_callers): Likewise.
13467 (decide_about_value): Likewise.
13468 (decide_whether_version_node): Likewise.
13469 (identify_dead_nodes): Likewise.
13470 (ipcp_store_bits_results): Likewise.
13471 * ipa-devirt.c (dump_targets): Likewise.
13472 (ipa_devirt): Likewise.
13473 * ipa-icf.c (sem_item::dump): Likewise.
13474 (sem_function::equals): Likewise.
13475 (sem_variable::equals): Likewise.
13476 (sem_item_optimizer::read_section): Likewise.
13477 (sem_item_optimizer::execute): Likewise.
13478 (congruence_class::dump): Likewise.
13479 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13480 (dump_inline_summary): Likewise.
13481 (estimate_node_size_and_time): Likewise.
13482 (inline_analyze_function): Likewise.
13483 * ipa-inline-transform.c (inline_call): Likewise.
13484 * ipa-inline.c (report_inline_failed_reason): Likewise.
13485 (want_early_inline_function_p): Likewise.
13486 (edge_badness): Likewise.
13487 (update_edge_key): Likewise.
13488 (inline_small_functions): Likewise.
13489 * ipa-profile.c (ipa_profile): Likewise.
13490 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13491 (ipa_make_edge_direct_to_target): Likewise.
13492 (remove_described_reference): Likewise.
13493 (ipa_impossible_devirt_target): Likewise.
13494 (propagate_controlled_uses): Likewise.
13495 (ipa_print_node_params): Likewise.
13496 (ipcp_transform_function): Likewise.
13497 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13498 (propagate_pure_const): Likewise.
13499 * ipa-reference.c (generate_summary): Likewise.
13500 (read_write_all_from_decl): Likewise.
13501 (propagate): Likewise.
13502 (ipa_reference_read_optimization_summary): Likewise.
13503 * ipa-utils.c (ipa_merge_profiles): Likewise.
13504 * ipa.c (walk_polymorphic_call_targets): Likewise.
13505 (symbol_table::remove_unreachable_nodes): Likewise.
13506 (ipa_single_use): Likewise.
13507 * passes.c (execute_todo): Likewise.
13508 * predict.c (drop_profile): Likewise.
13509 * symtab.c (symtab_node::get_dump_name): New function.
13510 (symtab_node::dump_name): Likewise.
13511 (symtab_node::dump_asm_name): Likewise.
13512 (symtab_node::dump_references): Likewise.
13513 (symtab_node::dump_referring): Likewise.
13514 (symtab_node::dump_base): Likewise.
13515 (symtab_node::debug_symtab): Likewise.
13516 * tree-sra.c (convert_callers_for_node): Likewise.
13517 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13518 * value-prof.c (init_node_map): Likewise.
13520 2017-05-23 Martin Liska <mliska@suse.cz>
13522 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13523 and symtab_node::debug_symtab to symbol_table::debug.
13524 * cgraphunit.c (analyze_functions): Use the renamed function.
13525 (symbol_table::compile): Likewise.
13526 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13527 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13528 * passes.c (execute_todo): Likewise.
13529 * symtab.c (symbol_table::dump): New function.
13530 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13532 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13534 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13535 that nonconst implies exec.
13537 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13539 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13540 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13541 (inline_edge_summary_vec): Turn into ...
13542 (ipa_call_summaries): ... this one.
13543 (redirect_to_unreachable, edge_set_predicate,
13544 evaluate_properties_for_edge, inline_summary_alloc,
13545 reset_ipa_call_summary, reset_inline_summary,
13546 inline_summary_t::duplicate): Update.
13547 (inline_edge_duplication_hook): Turn to ...
13548 (ipa_call_summary_t::duplicate): ... this one.
13549 (inline_edge_removal_hook): Turn to ...
13550 (ipa_call_summary_t::remove): ... this one.
13551 (dump_inline_edge_summary): Turn to ...
13552 (dump_ipa_call_summary): ... this one.
13553 (estimate_function_body_sizes): Update.
13554 (inline_update_callee_summaries): Update.
13555 (remap_edge_change_prob): Update.
13556 (remap_edge_summaries): Update.
13557 (inline_merge_summary): Update.
13558 (do_estimate_edge_time): Update.
13559 (inline_generate_summary): Update.
13560 (inline_read_section): Update.
13561 (inline_read_summary): Update.
13562 (inline_free_summary): Update.
13563 * ipa-inline.c (can_inline_edge_p): Update.
13564 (compute_inlined_call_time): Update.
13565 (want_inline_small_function_p): Update.
13566 (edge_badness): Update.
13567 (early_inliner): Update.
13568 * ipa-inline.h (inline_edge_summary): Turn to ...
13569 (ipa_call_summary): ... this one.
13570 (ipa_call_summary_t): New class.
13571 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13572 (ipa_call_summaries): New.
13573 (inline_edge_summary): Remove.
13574 (estimate_edge_growth): Update.
13575 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13576 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13577 * ipa-split.c (execute_split_functions): Update.
13578 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13580 2017-05-23 Tom de Vries <tom@codesourcery.com>
13582 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13583 attributes): Document rdrand effective target.
13585 2017-05-23 Tom de Vries <tom@codesourcery.com>
13587 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13588 attributes): Sort alphabetically.
13590 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13592 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13594 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13597 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13598 V2DF/V2DI splat into two separate patterns, one that handles
13599 registers, and the other that only handles memory. Drop support
13600 for splatting from a GPR on ISA 2.07 and then splitting the
13601 splat into direct move and splat.
13602 (vsx_splat_<mode>_reg): Likewise.
13603 (vsx_splat_<mode>_mem): Likewise.
13605 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13607 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13609 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13611 PR middle-end/80809
13612 * omp-low.c (finish_taskreg_remap): New function.
13613 (finish_taskreg_scan): If unit size of ctx->record_type
13614 is non-constant, unshare the size expression and replace
13615 decls in it with possible outer var refs.
13617 PR middle-end/80809
13618 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13619 GOVD_SHARED rather than GOVD_PRIVATE with it.
13620 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13621 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13623 PR middle-end/80853
13624 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13625 as last argument to build_outer_var_ref for pointer bases of array
13626 section reductions.
13628 2017-05-19 Martin Sebor <msebor@redhat.com>
13630 * print-tree.c (print_node): Print DECL_READ_P flag.
13632 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13634 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13635 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13636 * cgraph.c: Likewise.
13637 * cgraphunit.c: Likewise.
13638 * gengtype.c: Likewise.
13639 * ipa-cp.c: Likewise.
13640 * ipa-devirt.c: Likewise.
13641 * ipa-icf.c: Likewise.
13642 * ipa-predicate.c: Likewise.
13643 * ipa-profile.c: Likewise.
13644 * ipa-prop.c: Likewise.
13645 * ipa-split.c: Likewise.
13647 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13648 edge_predicate_pool, dump_inline_hints,
13649 inline_summary::account_size_time, redirect_to_unreachable,
13650 edge_set_predicate, set_hint_predicate,
13651 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13652 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13653 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13654 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13655 ipa_call_summary_t::remove, initialize_growth_caches,
13656 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13657 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13658 mark_modified, unmodified_parm_1, unmodified_parm,
13659 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13660 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13661 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13662 will_be_nonconstant_predicate, record_modified_bb_info,
13663 get_minimal_bb, record_modified, param_change_prob,
13664 phi_result_unknown_predicate, predicate_for_phi_result,
13665 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13666 estimate_function_body_sizes, compute_inline_parameters,
13667 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13668 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13669 inline_update_callee_summaries, remap_edge_change_prob,
13670 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13671 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13672 inline_analyze_function, inline_summary_t::insert,
13673 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13674 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13675 inline_free_summary): Move to ipa-fnsummary.h
13676 (predicate_t): Remove.
13677 * ipa-fnsummary.c: New file.
13678 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13679 (enum inline_hints_vals, inline_hints, agg_position_info,
13680 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13681 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13682 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13683 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13684 inline_read_summary, inline_write_summary, inline_free_summary,
13685 inline_analyze_function, initialize_inline_failed,
13686 inline_merge_summary, inline_update_overall_summary,
13687 compute_inline_parameters): Move to ipa-fnsummary.h
13688 * ipa-fnsummary.h: New file.
13689 * ipa-inline-transform.h: Include ipa-inline.h.
13690 * ipa-inline.c: LIkewise.
13692 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13694 * ipa-inline.c (edge_badness): Use inlined_time instead of
13695 inline_summaries->get.
13697 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13699 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13701 2017-05-22 Nathan Sidwell <nathan@acm.org>
13703 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13704 (fdump-lang): Document 'raw' option.
13705 * dumpfile.h (TDI_tu): Delete.
13706 * dumpfile.c (dump_files): Remove translation-unit.
13707 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13709 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13711 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13712 command option from $(AWK) call.
13713 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13714 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13715 [FORMAT]: Remove handling of variable.
13716 * config/avr/t-multilib: Regenerate.
13718 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13720 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13722 (dump_inline_summary): Do not print self_time.
13723 (estimate_function_body_sizes): Do not set self_time.
13724 (compute_inline_parameters): Likewise.
13725 (inline_read_section, inline_write_summary): Do not stream self_time.
13726 * ipa-inline.h (inline_summary): Drop self_time.
13728 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13730 * ipa-inline-analysis.c (account_size_time): Rename to ...
13731 (inline_summary::account_size_time): ... this one.
13732 (reset_ipa_call_summary): Turn to ...
13733 (ipa_call_summary::reset): ... this one.
13734 (reset_inline_summary): Turn to ...
13735 (inline_summary::reset): ... this one.
13736 (inline_summary_t::remove): Update.
13737 (inline_summary_t::duplicate): Update.
13738 (ipa_call_summary_t::remove): Update.
13739 (dump_inline_summary): Update.
13740 (estimate_function_body_sizes): Update.
13741 (compute_inline_parameters): Update.
13742 (estimate_node_size_and_time): Update.
13743 (inline_merge_summary): Update.
13744 (inline_update_overall_summary): Update.
13745 (inline_read_section): Update.
13746 (inline_write_summary): Update.
13747 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13748 add account_size_time and reset member functions.
13749 (ipa_call_summary): Add reset function.
13750 * ipa-predicate.h (predicate::operator &): Constify.
13752 2017-05-22 Richard Biener <rguenther@suse.de>
13754 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13756 2017-05-19 Jason Merrill <jason@redhat.com>
13758 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13760 2017-05-19 Marek Polacek <polacek@redhat.com>
13763 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13764 TYPE_OVERFLOW_WRAPS checks.
13766 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13768 * tree-core.h (enum omp_clause_default_kind): Add
13769 "OMP_CLAUSE_DEFAULT_PRESENT".
13770 * tree-pretty-print.c (dump_omp_clause): Handle it.
13771 * gimplify.c (enum gimplify_omp_var_data): Add
13772 "GOVD_MAP_FORCE_PRESENT".
13773 (gimplify_adjust_omp_clauses_1): Map it to
13774 "GOMP_MAP_FORCE_PRESENT".
13775 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13777 * gimplify.c (oacc_default_clause): Clarify.
13779 2017-05-19 Nathan Sidwell <nathan@acm.org>
13781 LANG_HOOK_REGISTER_DUMPS
13782 * toplev.c (general_init): Call register dump lang hook.
13783 * doc/invoke.texi: Document -fdump-lang option family.
13784 * dumpfile.c (dump_files): Remove class dump here.
13785 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13786 * dumpfile.h (tree_dump_index): Remove TDI_class.
13787 * langhooks-def.h (lhd_register_dumps): Declare.
13788 (LANG_HOOKS_REGISTER_DUMPS): Define.
13789 (LANG_HOOKS_INITIALIZER): Add it.
13790 * langhooks.c (lhd_register_dumps): Define.
13791 * langhooks.h (struct lang_hooks): Add register_dumps.
13793 2017-05-19 Nathan Sidwell <nathan@acm.org>
13795 * context.h (context::set_passes): New.
13796 * context.c (context::context): Do not create pass manager.
13797 * toplev.c (general_init): Create pass manager here.
13799 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
13801 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
13802 use this splitter if two add or or instructions would also work for
13803 the constant we want to generate.
13805 2017-05-19 Richard Biener <rguenther@suse.de>
13808 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
13809 predicate evaluation.
13811 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13813 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
13815 * ipa-inline.c (want_inline_small_function_p): Do not cast to
13818 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13820 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13821 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13822 (inline_edge_summary_vec): Turn into ...
13823 (ipa_call_summaries): ... this one.
13824 (redirect_to_unreachable, edge_set_predicate,
13825 evaluate_properties_for_edge, inline_summary_alloc,
13826 reset_ipa_call_summary, reset_inline_summary,
13827 inline_summary_t::duplicate): Update.
13828 (inline_edge_duplication_hook): Turn to ...
13829 (ipa_call_summary_t::duplicate): ... this one.
13830 (inline_edge_removal_hook): Turn to ...
13831 (ipa_call_summary_t::remove): ... this one.
13832 (dump_inline_edge_summary): Turn to ...
13833 (dump_ipa_call_summary): ... this one.
13834 (estimate_function_body_sizes): Update.
13835 (inline_update_callee_summaries): Update.
13836 (remap_edge_change_prob): Update.
13837 (remap_edge_summaries): Update.
13838 (inline_merge_summary): Update.
13839 (do_estimate_edge_time): Update.
13840 (inline_generate_summary): Update.
13841 (inline_read_section): Update.
13842 (inline_read_summary): Update.
13843 (inline_free_summary): Update.
13844 * ipa-inline.c (can_inline_edge_p): Update.
13845 (compute_inlined_call_time): Update.
13846 (want_inline_small_function_p): Update.
13847 (edge_badness): Update.
13848 (early_inliner): Update.
13849 * ipa-inline.h (inline_edge_summary): Turn to ...
13850 (ipa_call_summary): ... this one.
13851 (ipa_call_summary_t): New class.
13852 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13853 (ipa_call_summaries): New.
13854 (inline_edge_summary): Remove.
13855 (estimate_edge_growth): Update.
13856 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13857 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13858 * ipa-split.c (execute_split_functions): Update.
13859 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13861 2017-05-19 Richard Biener <rguenther@suse.de>
13863 PR middle-end/80764
13864 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13866 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13868 * config/rs6000/rs6000.c (struct machine_function): Add field
13869 fpr_is_wrapped_separately.
13870 (rs6000_get_separate_components): Use 64 components. Handle the
13871 new FPR components.
13872 (rs6000_components_for_bb): Handle the FPR components.
13873 (rs6000_emit_prologue_components): Handle the FPR components.
13874 (rs6000_emit_epilogue_components): Handle the FPR components.
13875 (rs6000_set_handled_components): Handle the FPR components.
13876 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13877 that are already separately shrink-wrapped.
13878 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13879 that are already separately shrink-wrapped.
13881 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13884 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13887 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13888 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13889 cases where the register allocator uses a move and an offsettable
13890 memory operation to/from a FPR register on ISA 2.06/2.07.
13891 (define_peephole2 for Altivec d-form store): Likewise.
13893 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13896 * config/i386/mmx.md (*mov<mode>_internal): Enable
13897 alternatives 11, 12, 13 and 14 also for 32bit targets.
13898 Remove alternatives 15, 16, 17 and 18.
13899 * config/i386/sse.md (vec_concatv2di): Change
13900 alternative (!x, *y) to (x, ?!*Yn).
13902 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13904 * dumpfile.h (enum dump_kind): Remove stray comma.
13906 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13908 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13909 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13910 predicate::num_conditions
13911 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13912 (CHANGED): turn into predicate::changed.
13913 (agg_position_info): Move to ipa-predicate.h
13914 (add_condition, predicate::add_clause, predicate::operator &=,
13915 predicate::or_with, predicate::evaluate, predicate::probability,
13916 dump_condition, dump_clause, predicate::dump,
13917 predicate::remap_after_duplication, predicate::remap_after_inlining,
13918 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13919 (evaluate_conditions_for_known_args): Update.
13920 (set_cond_stmt_execution_predicate): Update.
13921 * ipa-inline.h: Include ipa-predicate.h
13922 (condition, inline_param_summary, conditions, agg_position_info,
13923 predicate): Move to ipa-predicate.h
13924 * ipa-predicate.c: New file.
13925 * ipa-predicate.h: New file.
13927 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13929 * final.c (leaf_function_p): Check we are not in a sequence.
13931 2017-05-18 Martin Liska <mliska@suse.cz>
13933 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13934 * dumpfile.c (dump_register): Use new enum dump_kind.
13935 (get_dump_file_name): Likewise.
13936 (dump_enable_all): Likewise.
13937 (dump_switch_p_1): Likewise.
13938 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13939 * dumpfile.h (enum dump_kind): New enum type.
13940 (struct dump_file_info): Create constructor and
13941 format fields and comments.
13942 * passes.c (pass_manager::register_one_dump_file):
13944 * statistics.c (statistics_early_init): Likewise.
13945 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13946 TDF_TREE with TDF_SLIM.
13947 (gather_memory_references_ref): Likewise.
13949 2017-05-18 Martin Liska <mliska@suse.cz>
13951 * vec.h (struct vnull): Use it.
13953 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13955 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13956 (true_predicate, false_predicate, true_predicate_p,
13957 false_predicate_p): Remove.
13958 (single_cond_predicate, not_inlined_predicate): Turn to member function
13960 (add_condition): Update.
13961 (add_clause): Turn to...
13962 (predicate::add_clause): ... this one; update; allow passing NULL
13964 (and_predicates): Turn to ...
13965 (predicate::operator &=): ... this one.
13966 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13967 (or_predicates): Turn to ...
13968 (predicate::or_with): ... this one.
13969 (evaluate_predicate): Turn to ...
13970 (predicate::evaluate): ... this one.
13971 (predicate_probability): Turn to ...
13972 (predicate::probability): ... this one.
13973 (dump_condition): Update.
13974 (dump_predicate): Turn to ...
13975 (predicate::dump): ... this one.
13976 (account_size_time): Update.
13977 (edge_set_predicate): Update.
13978 (set_hint_predicate): UPdate.
13979 (evaluate_conditions_for_known_args): Update.
13980 (evaluate_properties_for_edge): Update.
13981 (remap_predicate_after_duplication): Turn to...
13982 (predicate::remap_after_duplication): ... this one.
13983 (remap_hint_predicate_after_duplication): Update.
13984 (inline_summary_t::duplicate): UPdate.
13985 (dump_inline_edge_summary): Update.
13986 (dump_inline_summary): Update.
13987 (set_cond_stmt_execution_predicate): Update.
13988 (set_switch_stmt_execution_predicate): Update.
13989 (compute_bb_predicates): Update.
13990 (will_be_nonconstant_expr_predicate): Update.
13991 (will_be_nonconstant_predicate): Update.
13992 (phi_result_unknown_predicate): Update.
13993 (predicate_for_phi_result): Update.
13994 (array_index_predicate): Update.
13995 (estimate_function_body_sizes): Update.
13996 (estimate_node_size_and_time): Update.
13997 (estimate_ipcp_clone_size_and_time): Update.
13998 (remap_predicate): Rename to ...
13999 (predicate::remap_after_inlining): ... this one.
14000 (remap_hint_predicate): Update.
14001 (inline_merge_summary): Update.
14002 (inline_update_overall_summary): Update.
14003 (estimate_size_after_inlining): Update.
14004 (read_predicate): Rename to ...
14005 (predicate::stream_in): ... this one.
14006 (read_inline_edge_summary): Update.
14007 (write_predicate): Rename to ...
14008 (predicate::stream_out): ... this one.
14009 (write_inline_edge_summary): Update.
14010 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14011 (clause_t): Turn to uint32_t
14012 (predicate): Turn to class; implement constructor and operators
14014 (size_time_entry): Update.
14015 (inline_summary): Update.
14016 (inline_edge_summary): Update.
14018 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14020 * fold-const.c (fold_binary_loc): Move transformation...
14021 * match.pd (C - X CMP X): ... here.
14023 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14025 * config/sparc/sparc.c (sparc_option_override): Set function
14026 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14027 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14029 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14031 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14033 2017-05-18 Marek Polacek <polacek@redhat.com>
14036 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14037 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14038 gimple_assign_load_p.
14040 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14042 PR middle-end/80692
14043 * real.c (do_compare): Give decimal_do_compare preference over
14044 comparing just the signs.
14046 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14048 * doc/md.texi (Canonicalization of Instructions): Describe the
14049 canonical form of instructions that inherently set a condition
14052 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14054 PR middle-end/80775
14055 * tree-cfg.c: Move deletion of unreachable case statements to after
14056 the merging of consecutive case labels.
14058 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14060 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14061 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14062 restoring of callee-saved registers.
14064 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14066 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14067 * config/visium/visium.c (single_set_and_flags): Likewise.
14068 * config/visium/visium.md (Substitutions): Likewise.
14070 2017-05-17 Martin Liska <mliska@suse.cz>
14072 * cfg.c: Introduce dump_flags_t type and
14073 use it instead of int type.
14075 * cfghooks.c: Likewise.
14076 * cfghooks.h (struct cfg_hooks): Likewise.
14077 * cfgrtl.c: Likewise.
14078 * cfgrtl.h: Likewise.
14079 * cgraph.c (cgraph_node::get_body): Likewise.
14080 * coretypes.h: Likewise.
14081 * domwalk.c: Likewise.
14082 * domwalk.h: Likewise.
14083 * dumpfile.c (struct dump_option_value_info): Likewise.
14084 (dump_enable_all): Likewise.
14085 (dump_switch_p_1): Likewise.
14086 (opt_info_switch_p): Likewise.
14087 * dumpfile.h (enum tree_dump_index): Likewise.
14088 (struct dump_file_info): Likewise.
14089 * genemit.c: Likewise.
14090 * generic-match-head.c: Likewise.
14091 * gengtype.c (open_base_files): Likewise.
14092 * gimple-pretty-print.c: Likewise.
14093 * gimple-pretty-print.h: Likewise.
14094 * graph.c (print_graph_cfg): Likewise.
14095 * graphite-scop-detection.c (dot_all_sese): Likewise.
14096 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14097 * loop-unroll.c (report_unroll): Likewise.
14098 * passes.c (pass_manager::register_one_dump_file): Likewise.
14099 * print-tree.c: Likewise.
14100 * statistics.c: Likewise.
14101 * tree-cfg.c: Likewise.
14102 * tree-cfg.h: Likewise.
14103 * tree-dfa.c: Likewise.
14104 * tree-dfa.h: Likewise.
14105 * tree-dump.c (dump_function): Likewise.
14106 * tree-dump.h (struct dump_info): Likewise.
14107 * tree-pretty-print.c: Likewise.
14108 * tree-pretty-print.h: Likewise.
14109 * tree-ssa-live.c: Likewise.
14110 * tree-ssa-live.h: Likewise.
14111 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14112 * tree-vect-loop.c: Likewise.
14113 * tree-vect-slp.c: Likewise.
14115 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14116 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14118 PR tree-optimization/80457
14119 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14120 of all arguments to a statement as scalar_to_vec operations.
14121 (vectorizable_call): Adjust call to vect_model_simple_cost for
14123 (vectorizable_conversion): Likewise.
14124 (vectorizable_assignment): Likewise.
14125 (vectorizable_shift): Likewise.
14126 (vectorizable_operation): Likewise.
14127 (vectorizable_comparison): Likewise.
14128 (vect_is_simple_cond): Record the def types for operands.
14129 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14130 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14131 for statement argument count.
14133 2017-05-16 Carl Love <cel@us.ibm.com>
14135 * config/rs6000/rs6000-c: Add support for built-in functions
14136 vector unsigned long long vec_bperm (vector unsigned long long,
14137 vector unsigned char)
14138 vector signed long long vec_mule (vector signed int,
14140 vector unsigned long long vec_mule (vector unsigned int,
14141 vector unsigned int)
14142 vector signed long long vec_mulo (vector signed int,
14144 vector unsigned long long vec_mulo (vector unsigned int,
14145 vector unsigned int)
14146 vector signed char vec_sldw (vector signed char,
14147 vector signed char,
14149 vector unsigned char vec_sldw (vector unsigned char,
14150 vector unsigned char,
14152 vector signed short vec_sldw (vector signed short,
14153 vector signed short,
14155 vector unsigned short vec_sldw (vector unsigned short,
14156 vector unsigned short,
14158 vector signed int vec_sldw (vector signed int,
14161 vector unsigned int vec_sldw (vector unsigned int,
14162 vector unsigned int,
14164 vector signed long long vec_sldw (vector signed long long,
14165 vector signed long long,
14167 vector unsigned long long vec_sldw (vector unsigned long long,
14168 vector unsigned long long,
14170 * config/rs6000/rs6000-c: Add support for built-in functions
14171 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14172 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14173 * doc/extend.texi: Update the built-in documentation for the
14174 new built-in functions.
14176 2017-05-16 Marek Polacek <polacek@redhat.com>
14180 * tree.c (save_expr): Don't fold the expression.
14182 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14184 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14185 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14186 and (?*y,m). Update insn attributes.
14188 2017-05-16 Martin Liska <mliska@suse.cz>
14190 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14191 flags argument of print_gimple_stmt, print_gimple_expr,
14192 print_generic_stmt and print_generic_expr.
14193 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14194 * coretypes.h: Likewise.
14195 * except.c (dump_eh_tree): Likewise.
14196 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14197 * gimple-pretty-print.h: Likewise.
14198 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14199 (backprop::push_to_worklist): Likewise.
14200 (backprop::pop_from_worklist): Likewise.
14201 (backprop::process_use): Likewise.
14202 (backprop::intersect_uses): Likewise.
14203 (note_replacement): Likewise.
14204 * gimple-ssa-store-merging.c
14205 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14206 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14207 (pass_store_merging::execute): Likewise.
14208 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14209 (ssa_base_cand_dump_callback): Likewise.
14210 (dump_incr_vec): Likewise.
14211 (replace_refs): Likewise.
14212 (replace_mult_candidate): Likewise.
14213 (create_add_on_incoming_edge): Likewise.
14214 (create_phi_basis): Likewise.
14215 (insert_initializers): Likewise.
14216 (all_phi_incrs_profitable): Likewise.
14217 (introduce_cast_before_cand): Likewise.
14218 (replace_one_candidate): Likewise.
14219 * gimplify.c (gimplify_expr): Likewise.
14220 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14221 (set_rename): Likewise.
14222 (rename_uses): Likewise.
14223 (copy_loop_phi_nodes): Likewise.
14224 (add_close_phis_to_merge_points): Likewise.
14225 (copy_loop_close_phi_args): Likewise.
14226 (copy_cond_phi_args): Likewise.
14227 (graphite_copy_stmts_from_block): Likewise.
14228 (translate_pending_phi_nodes): Likewise.
14229 * graphite-poly.c (print_pdr): Likewise.
14230 (dump_gbb_cases): Likewise.
14231 (dump_gbb_conditions): Likewise.
14232 (print_scop_params): Likewise.
14233 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14234 (build_cross_bb_scalars_use): Likewise.
14235 (gather_bbs::before_dom_children): Likewise.
14236 * hsa-dump.c (dump_hsa_immed): Likewise.
14237 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14238 (get_replacement_map): Likewise.
14239 * ipa-inline-analysis.c (dump_condition): Likewise.
14240 (estimate_function_body_sizes): Likewise.
14241 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14242 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14243 * ipa-prop.c (ipa_dump_param): Likewise.
14244 (ipa_print_node_jump_functions_for_edge): Likewise.
14245 (ipa_modify_call_arguments): Likewise.
14246 (ipa_modify_expr): Likewise.
14247 (ipa_dump_param_adjustments): Likewise.
14248 (ipa_dump_agg_replacement_values): Likewise.
14249 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14250 * ipa-pure-const.c (check_stmt): Likewise.
14251 (pass_nothrow::execute): Likewise.
14252 * ipa-split.c (execute_split_functions): Likewise.
14253 * omp-offload.c (dump_oacc_loop_part): Likewise.
14254 (dump_oacc_loop): Likewise.
14255 * trans-mem.c (tm_log_emit): Likewise.
14256 (tm_memopt_accumulate_memops): Likewise.
14257 (dump_tm_memopt_set): Likewise.
14258 (dump_tm_memopt_transform): Likewise.
14259 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14260 (print_loop): Likewise.
14261 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14262 (chkp_gather_checks_info): Likewise.
14263 (chkp_get_check_result): Likewise.
14264 (chkp_remove_check_if_pass): Likewise.
14265 (chkp_use_outer_bounds_if_possible): Likewise.
14266 (chkp_reduce_bounds_lifetime): Likewise.
14267 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14268 (chkp_mark_completed_bounds): Likewise.
14269 (chkp_register_incomplete_bounds): Likewise.
14270 (chkp_mark_invalid_bounds): Likewise.
14271 (chkp_maybe_copy_and_register_bounds): Likewise.
14272 (chkp_build_returned_bound): Likewise.
14273 (chkp_get_bound_for_parm): Likewise.
14274 (chkp_build_bndldx): Likewise.
14275 (chkp_get_bounds_by_definition): Likewise.
14276 (chkp_generate_extern_var_bounds): Likewise.
14277 (chkp_get_bounds_for_decl_addr): Likewise.
14278 * tree-chrec.c (chrec_apply): Likewise.
14279 * tree-data-ref.c (dump_data_reference): Likewise.
14280 (dump_subscript): Likewise.
14281 (dump_data_dependence_relation): Likewise.
14282 (analyze_overlapping_iterations): Likewise.
14283 * tree-inline.c (expand_call_inline): Likewise.
14284 (tree_function_versioning): Likewise.
14285 * tree-into-ssa.c (dump_defs_stack): Likewise.
14286 (dump_currdefs): Likewise.
14287 (dump_names_replaced_by): Likewise.
14288 (dump_update_ssa): Likewise.
14289 (update_ssa): Likewise.
14290 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14291 * tree-parloops.c (build_new_reduction): Likewise.
14292 (try_create_reduction_list): Likewise.
14293 (ref_conflicts_with_region): Likewise.
14294 (oacc_entry_exit_ok_1): Likewise.
14295 (oacc_entry_exit_single_gang): Likewise.
14296 * tree-pretty-print.h: Likewise.
14297 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14298 (get_scalar_evolution): Likewise.
14299 (add_to_evolution): Likewise.
14300 (get_loop_exit_condition): Likewise.
14301 (analyze_evolution_in_loop): Likewise.
14302 (analyze_initial_condition): Likewise.
14303 (analyze_scalar_evolution): Likewise.
14304 (instantiate_scev): Likewise.
14305 (number_of_latch_executions): Likewise.
14306 (gather_chrec_stats): Likewise.
14307 (final_value_replacement_loop): Likewise.
14308 (scev_const_prop): Likewise.
14309 * tree-sra.c (dump_access): Likewise.
14310 (disqualify_candidate): Likewise.
14311 (create_access): Likewise.
14312 (reject): Likewise.
14313 (maybe_add_sra_candidate): Likewise.
14314 (create_access_replacement): Likewise.
14315 (analyze_access_subtree): Likewise.
14316 (analyze_all_variable_accesses): Likewise.
14317 (sra_modify_assign): Likewise.
14318 (initialize_constant_pool_replacements): Likewise.
14319 (find_param_candidates): Likewise.
14320 (decide_one_param_reduction): Likewise.
14321 (replace_removed_params_ssa_names): Likewise.
14322 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14323 * tree-ssa-copy.c (dump_copy_of): Likewise.
14324 (copy_prop_visit_cond_stmt): Likewise.
14325 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14326 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14327 (record_equivalences_from_stmt): Likewise.
14328 * tree-ssa-dse.c (compute_trims): Likewise.
14329 (delete_dead_call): Likewise.
14330 (delete_dead_assignment): Likewise.
14331 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14332 (forward_propagate_into_cond): Likewise.
14333 (pass_forwprop::execute): Likewise.
14334 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14335 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14337 (move_computations_worker): Likewise.
14338 (execute_sm): Likewise.
14339 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14340 (remove_exits_and_undefined_stmts): Likewise.
14341 (remove_redundant_iv_tests): Likewise.
14342 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14343 (adjust_iv_update_pos): Likewise.
14344 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14345 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14346 (value_replacement): Likewise.
14347 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14348 * tree-ssa-pre.c (print_pre_expr): Likewise.
14349 (get_representative_for): Likewise.
14350 (create_expression_by_pieces): Likewise.
14351 (insert_into_preds_of_block): Likewise.
14352 (eliminate_insert): Likewise.
14353 (eliminate_dom_walker::before_dom_children): Likewise.
14354 (eliminate): Likewise.
14355 (remove_dead_inserted_code): Likewise.
14356 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14357 * tree-ssa-reassoc.c (get_rank): Likewise.
14358 (eliminate_duplicate_pair): Likewise.
14359 (eliminate_plus_minus_pair): Likewise.
14360 (eliminate_not_pairs): Likewise.
14361 (undistribute_ops_list): Likewise.
14362 (eliminate_redundant_comparison): Likewise.
14363 (update_range_test): Likewise.
14364 (optimize_range_tests_var_bound): Likewise.
14365 (optimize_vec_cond_expr): Likewise.
14366 (rewrite_expr_tree): Likewise.
14367 (rewrite_expr_tree_parallel): Likewise.
14368 (linearize_expr): Likewise.
14369 (break_up_subtract): Likewise.
14370 (linearize_expr_tree): Likewise.
14371 (attempt_builtin_powi): Likewise.
14372 (attempt_builtin_copysign): Likewise.
14373 (transform_stmt_to_copy): Likewise.
14374 (transform_stmt_to_multiply): Likewise.
14375 (dump_ops_vector): Likewise.
14376 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14377 (print_scc): Likewise.
14378 (set_ssa_val_to): Likewise.
14379 (visit_reference_op_store): Likewise.
14380 (visit_use): Likewise.
14381 (sccvn_dom_walker::before_dom_children): Likewise.
14382 (run_scc_vn): Likewise.
14383 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14385 (expr_hash_elt::print): Likewise.
14386 (const_and_copies::pop_to_marker): Likewise.
14387 (const_and_copies::record_const_or_copy_raw): Likewise.
14388 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14389 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14390 (dump_predicates): Likewise.
14391 (find_uninit_use): Likewise.
14392 (warn_uninitialized_phi): Likewise.
14393 (pass_late_warn_uninitialized::execute): Likewise.
14394 * tree-ssa.c (verify_vssa): Likewise.
14395 (verify_ssa): Likewise.
14396 (maybe_optimize_var): Likewise.
14397 * tree-vrp.c (dump_value_range): Likewise.
14398 (dump_all_value_ranges): Likewise.
14399 (dump_asserts_for): Likewise.
14400 (register_edge_assert_for_2): Likewise.
14401 (vrp_visit_cond_stmt): Likewise.
14402 (vrp_visit_switch_stmt): Likewise.
14403 (vrp_visit_stmt): Likewise.
14404 (vrp_visit_phi_node): Likewise.
14405 (simplify_cond_using_ranges_1): Likewise.
14406 (fold_predicate_in): Likewise.
14407 (evrp_dom_walker::before_dom_children): Likewise.
14408 (evrp_dom_walker::push_value_range): Likewise.
14409 (evrp_dom_walker::pop_value_range): Likewise.
14410 (execute_early_vrp): Likewise.
14412 2017-05-16 Richard Biener <rguenther@suse.de>
14414 * dwarf2out.c (loc_list_from_tree_1): Do not create
14415 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14417 2017-05-16 Richard Biener <rguenther@suse.de>
14419 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14421 (note_variable_value_in_expr): If we resolved the decl ref
14422 do not push to the stack.
14424 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14426 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14427 operations in fast-math mode.
14428 (vaddq_f16): Likewise.
14429 (vmul_f16): Likewise.
14430 (vmulq_f16): Likewise.
14431 (vsub_f16): Likewise.
14432 (vsubq_f16): Likewise.
14433 * config/arm/neon.md (add<mode>3): New.
14435 (fma:<VH:mode>3): New. Also remove outdated comment.
14438 2017-05-16 Martin Liska <mliska@suse.cz>
14442 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14443 (odr_types_equivalent_p): Likewise.
14445 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14447 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14449 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14452 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14453 non-interunit SSE move alternatives with '?'.
14454 (zero-extendsidi peephole2): New peephole to skip intermediate
14455 general register in SSE zero-extend sequence.
14457 2017-05-15 Jeff Law <law@redhat.com>
14459 * reorg.c (relax_delay_slots): Create a new variable to hold
14460 the temporary target rather than clobbering TARGET_LABEL.
14462 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14463 missing argument to extract_bit_field call.
14464 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14466 2017-05-15 Martin Liska <mliska@suse.cz>
14469 * gcc.c (process_command): Do not allow empty argument of -o option.
14471 2017-05-15 Renlin Li <renlin.li@arm.com>
14473 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14474 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14475 * config/aarch64/constraints.md (Usf): Add long call check.
14476 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14477 (call_value): Likewise.
14478 (sibcall): Likewise.
14479 (sibcall_value): Likewise.
14481 (call_value_insn): New.
14482 (sibcall_insn): Update rtx pattern.
14483 (sibcall_value_insn): Likewise.
14484 (call_internal): Remove.
14485 (call_value_internal): Likewise.
14486 (sibcall_internal): Likewise.
14487 (sibcall_value_internal): Likewise.
14488 (call_reg): Likewise.
14489 (call_symbol): Likewise.
14490 (call_value_reg): Likewise.
14491 (call_value_symbol): Likewise.
14493 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14496 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14498 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14500 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14501 compatible with CCGOCmode and with CCZmode.
14503 2017-05-14 Martin Sebor <msebor@redhat.com>
14505 PR middle-end/77671
14506 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14507 (gimple_fold_builtin_snprintf): Same.
14508 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14509 (gimple_fold_builtin_snprintf): Same.
14510 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14511 of character types.
14512 (is_call_safe): New function.
14513 (try_substitute_return_value): Call it.
14514 (try_simplify_call): New function.
14515 (pass_sprintf_length::handle_gimple_call): Call it.
14517 2017-05-14 Martin Sebor <msebor@redhat.com>
14519 PR middle-end/80669
14520 * builtins.c (expand_builtin_stpncpy): Simplify.
14522 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14524 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14525 * config/i386/i386.h
14526 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14527 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14528 (struct machine_function): Add new members call_ms2sysv,
14529 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14530 (struct machine_frame_state): New fields sp_realigned and
14531 sp_realigned_offset.
14532 * config/i386/i386.c
14533 (enum xlogue_stub): New enum.
14534 (enum xlogue_stub_sets): New enum.
14535 (class xlogue_layout): New class.
14536 (struct ix86_frame): New fields stack_realign_allocate_offset,
14537 stack_realign_offset and outlined_save_offset. Modify comments to
14538 detail stack layout when using out-of-line stubs.
14539 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14540 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14541 -mcall-ms2sysv-xlogues.
14542 (stub_managed_regs): New static variable.
14543 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14544 registers managed by out-of-line stub.
14545 (disable_call_ms2sysv_xlogues): New function.
14546 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14547 m->call_ms2sysv when appropriate and compute frame layout for
14549 (sp_valid_at, fp_valid_at): New inline functions.
14550 (choose_basereg): New function.
14551 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14553 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14554 Use align parameter of choose_baseaddr to generated aligned SSE movs
14556 (pro_epilogue_adjust_stack): Modify to track
14557 machine_frame_state::sp_realigned.
14558 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14559 (ix86_nsaved_sseregs): Likewise.
14560 (ix86_emit_save_regs): Likewise.
14561 (ix86_emit_save_regs_using_mov): Likewise.
14562 (ix86_emit_save_sse_regs_using_mov): Likewise.
14563 (get_scratch_register_on_entry): Likewise.
14564 (gen_frame_set): New function.
14565 (gen_frame_load): Likewise.
14566 (gen_frame_store): Likewise.
14567 (emit_outlined_ms2sysv_save): Likewise.
14568 (emit_outlined_ms2sysv_restore): Likewise.
14569 (ix86_expand_prologue): Modify stack re-alignment code and call
14570 emit_outlined_ms2sysv_save when appropriate.
14571 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14572 parameter rtx_insn *insn, which allows the function to be used to only
14573 generate the notes.
14574 (ix86_expand_epilogue): Modify validity checks of frame and stack
14575 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14576 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14577 * config/i386/predicates.md
14578 (save_multiple): New predicate.
14579 (restore_multiple): Likewise.
14580 * config/i386/sse.md
14581 (save_multiple<mode>): New pattern.
14582 (save_multiple_realign<mode>): Likewise.
14583 (restore_multiple<mode>): Likewise.
14584 (restore_multiple_and_return<mode>): Likewise.
14585 (restore_multiple_leave_return<mode>): Likewise.
14586 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14588 2017-05-14 Julia Koval <julia.koval@intel.com>
14590 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14591 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14592 (__builtin_ia32_xsetbv): New builtins.
14593 * config/i386/i386.c (ix86_expand_special_args_builtin):
14595 (ix86_expand_builtin): Special expand for new intrinsics.
14596 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14597 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14598 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14600 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14602 * cfganal.c (inverted_post_order_compute): Change argument type
14604 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14605 * df-core.c (rest_of_handle_df_initialize): Adjust.
14606 (rest_of_handle_df_finish): Likewise.
14607 (df_analyze_1): Likewise.
14608 (df_analyze): Likewise.
14609 (loop_inverted_post_order_compute): Change argument to be a vec *.
14610 (df_analyze_loop): Adjust.
14611 (df_get_n_blocks): Likewise.
14612 (df_get_postorder): Likewise.
14613 * df.h (struct df_d): Change field to be a vec.
14614 * lcm.c (compute_laterin): Adjust.
14615 (compute_available): Likewise.
14616 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14617 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14618 * tree-ssa-pre.c (compute_antic): Likewise.
14620 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14622 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14623 (depth_first_search::depth_first_search): Change structure init
14624 function to this constructor.
14625 (depth_first_search::add_bb): Rename function to this member.
14626 (depth_first_search::execute): Likewise.
14627 (flow_dfs_compute_reverse_finish): Adjust.
14629 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14631 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14632 (longest_simple_path): Likewise.
14633 * shrink-wrap.c (spread_components): Likewise.
14634 (disqualify_problematic_components): Likewise.
14635 (emit_common_heads_for_components): Likewise.
14636 (emit_common_tails_for_components): Likewise.
14637 (insert_prologue_epilogue_for_components): Likewise.
14639 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14641 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14644 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14646 * df-core.c (df_set_blocks): Start using auto_bitmap.
14647 (df_compact_blocks): Likewise.
14648 * df-problems.c (df_rd_confluence_n): Likewise.
14649 * df-scan.c (df_insn_rescan_all): Likewise.
14650 (df_process_deferred_rescans): Likewise.
14651 (df_update_entry_block_defs): Likewise.
14652 (df_update_exit_block_uses): Likewise.
14653 (df_entry_block_bitmap_verify): Likewise.
14654 (df_exit_block_bitmap_verify): Likewise.
14655 (df_scan_verify): Likewise.
14656 * lra-constraints.c (lra_constraints): Likewise.
14657 (undo_optional_reloads): Likewise.
14658 (lra_undo_inheritance): Likewise.
14659 * lra-remat.c (calculate_gen_cands): Likewise.
14660 (do_remat): Likewise.
14661 * lra-spills.c (assign_spill_hard_regs): Likewise.
14662 (spill_pseudos): Likewise.
14663 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14664 (bitmap_set_subtract_values): Likewise.
14666 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14668 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14669 management with auto_bitmap.
14670 (fix_inter_tick): Likewise.
14671 (fix_recovery_deps): Likewise.
14672 * ira.c (add_store_equivs): Likewise.
14673 (find_moveable_pseudos): Likewise.
14674 (split_live_ranges_for_shrink_wrap): Likewise.
14675 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14676 (rtx_reuse_manager::seen_def_p): Likewise.
14677 (rtx_reuse_manager::set_seen_def): Likewise.
14678 * print-rtl.h (class rtx_reuse_manager): Likewise.
14680 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14682 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14684 (migrate_btr_def): Likewise.
14685 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14686 * df-core.c (loop_post_order_compute): Likewise.
14687 (loop_inverted_post_order_compute): Likewise.
14688 * hsa-common.h: Likewise.
14689 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14690 * init-regs.c (initialize_uninitialized_regs): Likewise.
14691 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14692 (inline_small_functions): Likewise.
14693 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14694 * ira.c (combine_and_move_insns): Likewise.
14695 (build_insn_chain): Likewise.
14696 * loop-invariant.c (find_invariants): Likewise.
14697 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14698 * predict.c (tree_predict_by_opcode): Likewise.
14699 (predict_paths_leading_to): Likewise.
14700 (predict_paths_leading_to_edge): Likewise.
14701 (estimate_loops_at_level): Likewise.
14702 (estimate_loops): Likewise.
14703 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14704 (spread_components): Likewise.
14705 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14706 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14707 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14708 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14709 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14710 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14711 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14712 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14713 (mark_threaded_blocks): Likewise.
14714 (thread_through_all_blocks): Likewise.
14715 * tree-ssa.c (verify_ssa): Likewise.
14716 (execute_update_addresses_taken): Likewise.
14717 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14719 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14721 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14723 (post_order_compute): Likewise.
14724 (inverted_post_order_compute): Likewise.
14725 (pre_and_rev_post_order_compute_fn): Likewise.
14727 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14729 * genrecog.c (int_set::int_set): Explicitly construct our
14730 auto_vec base class.
14731 * vec.h (auto_vec::auto_vec): New constructor.
14733 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14735 * bitmap.h (class auto_bitmap): New constructor taking
14736 bitmap_obstack * argument.
14738 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14740 * bitmap.h (class auto_bitmap): Change type of m_bits to
14741 bitmap_head, and adjust ctor / dtor and member operators.
14743 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14745 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14746 when returned register mode doesn't match original mode.
14748 2017-05-12 Jeff Law <law@redhat.com>
14749 Jakub Jelinek <jakub@redhat.com>
14751 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14752 we look for cc setter after the compare-elim changes.
14753 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14754 within the vector to match what compare-elim now expects.
14755 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14756 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14758 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14759 after the compare-elim changes.
14760 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14761 the vector to match what compare-elim now expects.
14762 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14763 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14764 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14765 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14766 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14768 * config/visium/visium.c (single_set_and_flags): Fix where
14769 we look for cc setter after the compare-elim changes.
14770 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14771 with the vector to match what compare-elim now expects.
14772 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14773 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14774 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14775 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14776 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14777 (neg<mode>2_insn_set_overflow): Likewise.
14779 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14781 PR middle-end/79794
14782 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14783 maybe_expand_insn call, set ops[0].target. If still set after call,
14784 set alt_rtl. Add extra arg to recursive calls.
14785 (extract_bit_field): Add alt_rtl argument. Pass to
14787 * expmed.h (extract_bit_field): Fix prototype.
14788 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14789 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14790 to extract_bit_field_calls.
14791 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
14792 Pass alt_rtl to extract_bit_field calls.
14793 * calls.c (store_unaligned_arguments_into_psuedos)
14794 load_register_parameters): Pass extra NULL to extract_bit_field calls.
14795 * optabs.c (maybe_legitimize_operand): Clear op->target when call
14797 * optabs.h (struct expand_operand): Add target bitfield.
14799 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14801 * compare-elim.c (try_eliminate_compare): Canonicalize
14802 operation with embedded compare to
14803 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
14804 (set (reg) (operation)].
14806 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
14808 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14811 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
14812 cost of adding a carry flag for ADC instruction.
14813 [case MINUS]: Ignore the cost of subtracting a carry flag
14814 for SBB instruction.
14816 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
14818 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
14820 * config/rs6000/bmiintrin.h: New file.
14821 * config/rs6000/bmi2intrin.h: New file.
14822 * config/rs6000/x86intrin.h: New file.
14824 2017-05-12 Jeff Law <law@redhat.com>
14826 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14829 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14831 PR middle-end/80707
14832 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14834 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14836 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14837 early expansion of vector divide builtins.
14838 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14839 builtins identified as having unsigned arguments.
14841 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14843 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14844 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14845 expansion of vector logical operations (and, andc, or, xor,
14848 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14850 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14851 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14853 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14855 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14856 early GIMPLE expansion of vector multiplies.
14858 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14860 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14861 TARGET_HAVE_MOVT conditional.
14862 (movt splitter): Likewise.
14864 2017-05-12 Richard Biener <rguenther@suse.de>
14866 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14867 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14868 Fold all stmts not inplace.
14870 2017-05-12 Richard Biener <rguenther@suse.de>
14872 PR tree-optimization/80713
14873 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14874 inserted_exprs bit for not removed stmts.
14876 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14878 PR middle-end/69921
14879 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14880 parallelized" attribute for parallelized OpenACC kernels.
14881 * omp-offload.c (execute_oacc_device_lower): Use it.
14883 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14884 Set "oacc kernels" attribute.
14885 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14886 parameter. Adjust all users.
14887 (oacc_fn_attrib_kernels_p): Remove function.
14888 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14889 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14890 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14891 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14892 assert "oacc kernels" attribute is set.
14894 2017-05-11 Carl Love <cel@us.ibm.com>
14896 * config/rs6000/rs6000-c: Add support for built-in functions
14897 vector unsigned char vec_popcnt (vector signed char)
14898 vector unsigned char vec_popcnt (vector unsigned char)
14899 vector unsigned short vec_popcnt (vector signed short)
14900 vector unsigned short vec_popcnt (vector unsigned short)
14901 vector unsigned int vec_popcnt (vector signed int)
14902 vector unsigned int vec_popcnt (vector unsigned int)
14903 vector unsigned long long vec_popcnt (vector signed long long)
14904 vector unsigned long long vec_popcnt (vector unsigned long long)
14905 vector signed long long vec_slo (vector signed long long,
14906 vector signed char)
14907 vector signed long long vec_slo (vector signed long long,
14908 vector unsigned char)
14909 vector unsigned long long vec_slo (vector unsigned long long,
14910 vector signed char)
14911 vector unsigned long long vec_slo (vector unsigned long long,
14912 vector unsigned char)
14913 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14914 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14915 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14916 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14917 * doc/extend.texi: Update the built-in documentation file for the
14918 new built-in functions.
14920 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14922 * attribs.h (sorted_attr_string): Move machine independent
14923 functions for target clone support from the i386 port to common
14924 code. Rename ix86_function_versions to common_function_versions.
14925 Rename make_name to make_unique_name.
14926 (common_function_versions): Likewise.
14927 (make_unique_name): Likewise.
14928 (make_dispatcher_decl): Likewise.
14929 (is_function_default_version): Likewise.
14930 * attribs.c (attr_strcmp): Likewise.
14931 (sorted_attr_string): Likewise.
14932 (common_function_versions): Likewise.
14933 (make_unique_name): Likewise.
14934 (make_dispatcher_decl): Likewise.
14935 (is_function_default_version): Likewise.
14936 * config/i386/i386.c (attr_strcmp): Likewise.
14937 (sorted_attr_string): Likewise.
14938 (ix86_function_versions): Likewise.
14939 (make_name): Likewise.
14940 (make_dispatcher_decl): Likewise.
14941 (is_function_default_version): Likewise.
14942 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14944 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14947 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14948 Account for direct move costs for vec_construct of integer
14951 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14954 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14955 (UNSPEC_STX_ATOMIC): Ditto.
14956 (loaddi_via_sse): New insn.
14957 (storedi_via_sse): Ditto.
14958 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14959 Update corresponding peephole2 patterns.
14960 (atomic_storedi_fpu): Ditto.
14962 2017-05-11 Julia Koval <julia.koval@intel.com>
14964 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14965 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14967 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14968 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14969 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14971 2017-05-11 Nathan Sidwell <nathan@acm.org>
14973 * graphite-poly.c: Include dumpfile.h.
14975 * dumpfle.h (dump_function): Declare here ...
14976 * tree-dump.h (dump_function): ... not here.
14977 * dumpfile.c: #include tree-cfg.h.
14978 (dump_function): Move here from ...
14979 * tree-dump.c (dump_function): ... here.
14980 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14981 * graphite-poly.c: Don't include tree-dump.h.
14982 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14983 * print-tree.c: Likewise.
14984 * stor-layout.c: Likewise.
14985 * tree-nested.c: Likewise.
14987 * dumpfile.c (dump_start): Use TDF_FLAGS.
14988 (dump_enable_all): Fix TDF_KIND check thinko.
14990 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14992 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14993 array entries to represent two legal parameterizations of the
14994 overloaded __builtin_cmpb function, as represented by the
14995 P6_OV_BUILTIN_CMPB constant.
14996 (altivec_resolve_overloaded_builtin): Add special case handling
14997 for the __builtin_cmpb function, as represented by the
14998 P6_OV_BUILTIN_CMPB constant.
14999 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15000 (BU_P6_64BIT_2): New macro.
15001 (BU_P6_OVERLOAD_2): New macro
15002 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15003 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15004 (CMPB): Add overload support to represent both 32-bit and 64-bit
15005 compare-bytes function.
15006 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15007 support for TARGET_CMPB.
15008 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15009 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15010 documentation of the __builtin_cmpb overloaded built-in function.
15012 2017-05-11 Richard Biener <rguenther@suse.de>
15014 PR tree-optimization/80705
15015 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15016 bases are not vectorizable.
15018 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15020 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15021 when counting register pressure.
15023 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15025 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15026 (iv_ca_more_deps): Renamed to ...
15027 (iv_ca_compare_deps): ... this.
15028 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15030 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15032 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15034 (determine_group_iv_costs): ... here.
15035 (find_inv_vars_cb): Record inv var if it's not recorded before.
15037 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15039 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15040 (get_shiftadd_cost): Ditto.
15042 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15044 * tree-ssa-address.c: Include header file.
15045 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15047 (add_to_parts): Refactor.
15048 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15049 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15054 PR tree-optimization/53090
15055 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15057 (extract_cond_operands): Detect condition with IV on both sides
15058 and return COMP_IV_EXPR_2.
15059 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15060 (rewrite_use_compare): Simplify by removing call to function
15061 extract_cond_operands.
15063 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15065 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15066 (extract_cond_operands): Detect condition comparing against non-
15067 invariant bound and return appropriate enum value.
15068 (find_interesting_uses_cond): Update use of extract_cond_operands.
15069 Handle its return value accordingly.
15070 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15072 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15074 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15075 nonlinear iv_use computation in loop invariant sensitive way.
15077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15079 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15080 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15082 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15084 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15085 (dump_cand): Support iv_cand.inv_exprs.
15086 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15088 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15091 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15093 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15095 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15096 as local function. Include necessary header files.
15097 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15099 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15101 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15103 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15105 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15106 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15107 RSHIFT_EXPR and BIT_NOT_EXPR.
15109 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15111 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15112 (adjust_setup_cost): New parameter supporting round up adjustment.
15113 (struct address_cost_data): Delete.
15114 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15115 (split_address_cost, ptr_difference_cost): Delete.
15116 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15117 (struct ainc_cost_data): New struct.
15118 (get_address_cost_ainc): New function.
15119 (get_address_cost, get_computation_cost): Reimplement.
15120 (determine_group_iv_cost_address): Record inv_expr for all uses of
15122 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15123 (iv_ca_has_deps): Reimplemented to ...
15124 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15126 (iv_ca_extend): Call iv_ca_more_deps.
15128 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15130 * tree-ssa-address.c (struct mem_address): Move to header file.
15131 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15132 * tree-ssa-address.h (struct mem_address): Move from C file.
15133 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15137 * tree-affine.h (aff_combination_type): New interface.
15138 (aff_combination_zero_p): Remove static.
15139 (aff_combination_const_p): New interface.
15140 (aff_combination_singleton_var_p): New interfaces.
15142 2017-05-11 Richard Biener <rguenther@suse.de>
15144 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15145 Skip unreachable blocks and destinations.
15146 (eliminate): Move stmt removal and fixup ...
15147 (fini_eliminate): ... here. Skip inserted exprs.
15148 (pass_pre::execute): Move fini_pre after fini_eliminate.
15149 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15150 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15151 PRE to get rid of dead code that has invalid SSA form and
15152 split critical edges again.
15154 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15156 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15158 2017-05-11 Richard Biener <rguenther@suse.de>
15160 * passes.c (execute_function_todo): Verify loops if they are
15161 said to be up-to-date.
15162 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15163 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15165 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15168 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15169 handle calling assemble_external ourself.
15172 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15173 modes with zero size. Enhance comment.
15175 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15177 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15178 built-ins for vec_xl and vec_xst with short and char pointer
15181 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15183 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15184 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15185 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15186 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15187 (_mm_maskz_min_round_ss): New intrinsics.
15188 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15189 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15190 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15191 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15192 (__builtin_ia32_minss_mask_round): New builtins.
15193 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15194 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15195 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15197 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15198 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15200 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15203 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15205 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15206 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15207 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15208 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15209 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15210 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15211 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15212 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15213 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15214 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15215 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15216 (__builtin_ia32_mulss_mask_round): New builtins.
15217 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15218 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15219 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15221 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15222 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15224 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15227 2017-05-10 Julia Koval <julia.koval@intel.com>
15229 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15230 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15231 (_mm256_setr_m128i): New intrinsics.
15233 2017-05-10 Julia Koval <julia.koval@intel.com>
15235 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15236 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15237 (_mm_maskz_rcp14_ss): New intrinsics.
15238 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15239 (__builtin_ia32_rcp14ss_mask): New builtins.
15240 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15242 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15244 PR tree-optimization/51513
15245 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15246 (assert_unreachable_fallthru_edge_p): Use it.
15247 (group_case_labels_stmt): Likewise.
15248 * tree-cfg.h: Prototype it.
15249 * stmt.c: Include cfghooks.h and tree-cfg.h.
15250 (emit_case_dispatch_table) <gap_label>: New local variable.
15251 Use it to fill dispatch table gaps.
15252 Test for default_label before updating probabilities.
15253 (expand_case) <default_label>: Remove unneeded initialization.
15254 Test for unreachable default case statement and remove its edge.
15255 Set default_label accordingly.
15256 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15258 2017-05-10 Carl Love <cel@us.ibm.com>
15260 * config/rs6000/rs6000-c: Add support for built-in functions
15261 vector signed char vec_neg (vector signed char)
15262 vector signed short int vec_neg (vector short int)
15263 vector signed int vec_neg (vector signed int)
15264 vector signed long long vec_neg (vector signed long long)
15265 vector float vec_neg (vector float)
15266 vector double vec_neg (vector double)
15267 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15269 * config/rs6000/altivec.h: Add define for vec_neg
15270 * doc/extend.texi: Update the built-in documentation for the
15271 new built-in functions.
15273 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15275 PR tree-optimization/77644
15276 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15278 2017-05-10 Nathan Sidwell <nathan@acm.org>
15280 * dumpfile.h (TDI_lang_all): New.
15281 (TDF_KIND): New. Renumber others
15282 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15284 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15286 (get_dump_file_name): Adjust suffix generation.
15287 (dump_enable_all): Use TDF_KIND.
15288 * doc/invoke.texi (-fdump-lang-all): Document.
15290 * dumpfile.h: Tabify.
15292 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15295 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15296 Move member access before delete.
15298 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15300 * tree-inline.c (expand_call_inline): Split block at stmt
15303 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15306 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15307 are now unused after splitting mov{sf,sd}_hardfloat.
15308 (f32_lr2): Likewise.
15309 (f32_lm): Likewise.
15310 (f32_lm2): Likewise.
15311 (f32_li): Likewise.
15312 (f32_li2): Likewise.
15313 (f32_lv): Likewise.
15314 (f32_sr): Likewise.
15315 (f32_sr2): Likewise.
15316 (f32_sm): Likewise.
15317 (f32_sm2): Likewise.
15318 (f32_si): Likewise.
15319 (f32_si2): Likewise.
15320 (f32_sv): Likewise.
15321 (f32_dm): Likewise.
15322 (f32_vsx): Likewise.
15323 (f32_av): Likewise.
15324 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15325 For movsf, order stores so the VSX stores occur before the GPR
15326 store which encourages the register allocator to use a traditional
15327 FPR instead of a GPR. For movsd, order the stores so that the GPR
15328 store comes before the VSX stores to allow the power6 to work.
15329 This is due to the power6 not having a 32-bit integer store
15330 instruction from a FPR.
15331 (movsf_hardfloat): Likewise.
15332 (movsd_hardfloat): Likewise.
15334 2017-05-09 Martin Sebor <msebor@redhat.com>
15336 PR translation/80280
15337 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15340 PR translation/80280
15341 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15342 data member added in r247778.
15343 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15345 2017-05-09 Nathan Sidwell <nathan@acm.org>
15347 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15349 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15352 2017-05-09 Marek Polacek <polacek@redhat.com>
15354 * doc/invoke.texi: Fix typo.
15356 2017-05-09 Richard Biener <rguenther@suse.de>
15358 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15359 (vrp_val_is_min): Likewise.
15360 (set_value_range_to_value): Likewise.
15361 (set_value_range_to_nonnegative): Likewise.
15362 (gimple_assign_nonzero_p): Likewise.
15363 (gimple_stmt_nonzero_p): Likewise.
15364 (vrp_int_const_binop): Likewise. Remove unreachable case.
15365 (adjust_range_with_scev): Adjust comments.
15366 (compare_range_with_value): Likewise.
15367 (extract_range_from_phi_node): Likewise.
15368 (test_for_singularity): Likewise.
15370 2017-05-09 Richard Biener <rguenther@suse.de>
15372 * tree-vrp.c (get_single_symbol): Add assert that we don't
15373 get overflowed constants as invariant part.
15374 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15375 checks. Use wi::cmp instead of recursing for integer constants.
15376 (compare_values): Just ignore whether we assumed undefined
15377 overflow instead of failing the compare.
15378 (extract_range_for_var_from_comparison_expr): Add comment before the
15379 TREE_NO_WARNING sets.
15380 (test_for_singularity): Likewise.
15381 (extract_range_from_comparison): Do not disable optimization
15382 when we assumed undefined overflow.
15383 (extract_range_basic): Remove init of unused var.
15385 2017-05-09 Richard Biener <rguenther@suse.de>
15387 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15388 (extract_range_from_multiplicative_op_1): Adjust.
15389 (extract_range_from_binary_expr_1): Use int_const_binop.
15391 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15394 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15395 rs6000_store_data_bypass_p in seven define_bypass directives and
15396 in several comments.
15397 * config/rs6000/rs6000-protos.h: Add prototype for
15398 rs6000_store_data_bypass_p function.
15399 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15400 function implements slightly different (rs6000-specific) semantics
15401 than store_data_bypass_p, returning false rather than aborting
15402 with assertion error when arguments do not satisfy the
15403 requirements of store data bypass.
15404 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15405 rs6000_store_data_bypass_p.
15407 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15409 * config/xtensa/xtensa-protos.h
15410 (xtensa_initial_elimination_offset): New declaration.
15411 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15412 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15413 macro definition, add case for FRAME_POINTER_REGNUM when
15414 FRAME_GROWS_DOWNWARD.
15415 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15416 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15417 xtensa_initial_elimination_offset.
15419 2017-05-08 Nathan Sidwell <nathan@acm.org>
15421 * doc/invoke.texi: Alphabetize -fdump options.
15423 2017-05-08 Martin Sebor <msebor@redhat.com>
15425 PR translation/80280
15426 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15428 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15430 * target.def (compute_frame_layout): New optional target hook.
15431 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15432 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15433 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15435 * reload1.c (verify_initial_elim_offsets): Likewise.
15436 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15437 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15438 (arm_get_frame_offsets): Split up into this ...
15439 (arm_compute_frame_layout): ... and this function.
15441 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15443 * config/aarch64/constraints.md (Usa): New constraint.
15444 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15446 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15448 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15449 with_multilib_list after it has been checked.
15451 2017-05-08 Richard Biener <rguenther@suse.de>
15453 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15454 (bitmap_set_subtract_values): Likewise.
15456 2017-05-08 Richard Biener <rguenther@suse.de>
15458 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15459 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15461 (gimple_stmt_nonzero_warnv_p): Rename to ...
15462 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15464 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15465 (extract_range_basic): Adjust, do not disable propagation on
15466 strict overflow sensitive simplification.
15467 (vrp_visit_cond_stmt): Likewise.
15469 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15471 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15472 body size unconditionally.
15474 2017-05-07 Jeff Law <law@redhat.com>
15477 2017-05-06 Jeff Law <law@redhat.com>
15478 PR tree-optimization/78496
15479 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15482 PR tree-optimization/78496
15483 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15484 (simplify_stmt_using_ranges): Call it.
15485 (vrp_dom_walker::before_dom_children): Extract equivalences
15486 from an ASSERT_EXPR with an equality comparison against a
15489 2017-05-06 Jeff Law <law@redhat.com>
15491 PR tree-optimization/78496
15492 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15495 PR tree-optimization/78496
15496 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15497 (simplify_stmt_using_ranges): Call it.
15498 (vrp_dom_walker::before_dom_children): Extract equivalences
15499 from an ASSERT_EXPR with an equality comparison against a
15502 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15504 * lra-constraints.c (lra_copy_reg_equiv): New function.
15505 (split_reg): Use it to copy equivalence information from the
15506 original register to the spill register.
15508 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15510 PR rtl-optimization/75964
15511 * simplify-rtx.c (simplify_const_relational_operation): Remove
15512 invalid handling of comparisons of integer ABS.
15514 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15516 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15517 initialize to zero.
15518 (init_regs): Remove declaration.
15519 (function_arg_advance_32): Initialize error_p as boolean variable.
15521 2017-05-05 Nathan Sidwell <nathan@acm.org>
15523 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15524 lines. Use for (;;).
15526 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15528 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15529 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15530 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15531 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15532 VF=2 that require versioning.
15534 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15536 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15539 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15541 * diagnostic.h (diagnostic_override_option_index): Convert from
15542 macro to inline function.
15544 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15546 * diagnostic.c (last_module_changed_p): New function.
15547 (set_last_module): New function.
15548 (diagnostic_report_current_module): Convert macro usage to
15549 the above functions.
15550 * diagnostic.h (diagnostic_context::last_module): Strengthen
15551 from const line_map * to const line_map_ordinary *.
15552 (diagnostic_last_module_changed): Delete macro.
15553 (diagnostic_set_last_module): Delete macro.
15555 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15557 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15558 with diagnostic_report_diagnostic.
15559 (diagnostic_n_impl_richloc): Likewise.
15560 * diagnostic.h (report_diagnostic): Delete macro.
15561 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15562 with diagnostic_report_diagnostic.
15563 * substring-locations.c (format_warning_va): Likewise.
15565 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15567 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15568 save/restor of format_spec. Move option-printing code to...
15569 (print_option_information): ...this new function, and
15570 reimplement by simply printing to the pretty_printer,
15571 rather than appending to the format string.
15573 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15575 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15576 handling logic into...
15577 (update_effective_level_from_pragmas): ...this new function.
15579 2017-05-04 Andrew Waterman <andrew@sifive.com>
15581 * config/riscv/riscv.opt (mstrict-align): New option.
15582 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15583 (SLOW_UNALIGNED_ACCESS): Define.
15584 (riscv_slow_unaligned_access): Declare.
15585 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15587 (riscv_slow_unaligned_access): New variable.
15588 (rocket_tune_info): Set slow_unaligned_access to true.
15589 (optimize_size_tune_info): Set slow_unaligned_access to false.
15590 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15591 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15592 (riscv_option_override): Set riscv_slow_unaligned_access.
15593 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15595 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15597 * config/riscv/riscv.md: Unify indentation.
15599 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15604 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15606 (extendsi<mode>2): Add support for doing sign extension via
15607 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15608 don't have ISA 3.0 instructions.
15609 (extendsi<mode>2 splitter): Likewise.
15610 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15611 generate the normal insns since SImode can now go in vector
15612 registers. Disallow the special UNSPECs needed for previous
15613 machines to hide SImode being used. Add new insns
15614 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15615 (fix_trunc<mode>si2_stfiwx): Likewise.
15616 (fix_trunc<mode>si2_internal): Likewise.
15617 (fixuns_trunc<mode>si2): Likewise.
15618 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15619 (fctiw<u>z_<mode>_smallint): Likewise.
15620 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15621 of floating point to 32-bit integer from doing a direct move to
15622 the GPR registers to do a store.
15623 (fctiwz_<mode>): Break long line.
15625 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15627 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15628 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15629 (addr_list, addr_offset_valid_p): New.
15630 (split_address_groups): Check offset validity with above function.
15631 (gt-tree-ssa-loop-ivopts.h): Include header file.
15633 2017-05-05 Nathan Sidwell <nathan@acm.org>
15635 * config.gcc (arm*-*-*): Add missing 'fi'.
15637 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15639 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15640 included in -fopt-info does not matter.
15641 * doc/optinfo.texi (-fopt-info): Fix description of default
15642 behavour. Explicitly say order of options included in -fopt-info
15645 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15647 * config.gcc: Allow combinations of aprofile and rmprofile values for
15648 --with-multilib-list.
15649 * config/arm/t-multilib: New file.
15650 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15651 variables. Remove setting of ISA and floating-point ABI in
15652 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15653 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15654 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15655 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15657 * config/arm/t-rmprofile: Likewise except for the matches changes.
15658 * doc/install.texi (--with-multilib-list): Document the combination of
15659 aprofile and rmprofile values and warn about pitfalls in doing that.
15661 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15663 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15664 (movdi_aarch64): Likewise.
15666 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15668 PR tree-optimization/80632
15669 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15671 (build_arrays): Initialize it for virtual phis.
15672 (fix_phi_nodes): Use it for virtual phis.
15674 PR tree-optimization/80558
15675 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15676 [x, y] op z into [x op, y op z] for op & or | if conditions
15679 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15680 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15683 * config/arm/arm.md (use_literal_pool): Remove.
15684 (64-bit immediate split): No longer takes cost into consideration
15685 if arm_disable_literal_pool is enabled.
15686 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15687 used when arm_disable_literal_pool is enabled.
15688 (arm_max_const_double_inline_cost): Remove use of
15689 arm_disable_literal_pool.
15690 (push_minipool_fix): Add assert.
15691 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15692 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15693 (no_literal_pool_sf_immediate): New.
15695 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15697 PR tree-optimization/80613
15698 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15699 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15701 2017-05-05 Richard Biener <rguenther@suse.de>
15703 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15705 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15707 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15708 of this flag from insn conditions due to removal from r247495.
15710 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15712 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15714 (arm_early_store_addr_dep_ptr): Likewise.
15715 * config/arm/aarch-common-protos.h
15716 (arm_early_load_addr_dep_ptr): Add prototype.
15717 (arm_early_store_addr_dep_ptr): Likewise.
15718 * config/arm/cortex-a53.md: Add new bypasses.
15720 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15722 * tree.c (next_type_uid): Change type to unsigned.
15723 (type_hash_canon): Decrement back next_type_uid if
15724 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15725 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15728 2017-05-04 Martin Sebor <msebor@redhat.com>
15730 * builtins.c: Fix a trivial typo in a comment.
15732 PR middle-end/79234
15733 * builtins.c (check_sizes): Adjust to handle reading past the end.
15734 Avoid printing excessive upper bound of ranges. Use %E to print
15735 tree nodes instead of converting them to %wu.
15736 (expand_builtin_memchr): New function.
15737 (compute_dest_size): Rename...
15738 (compute_objsize): ...to this.
15739 (expand_builtin_memcpy): Adjust.
15740 (expand_builtin_mempcpy): Adjust.
15741 (expand_builtin_strcat): Adjust.
15742 (expand_builtin_strcpy): Adjust.
15743 (check_strncat_sizes): Adjust.
15744 (expand_builtin_strncat): Adjust.
15745 (expand_builtin_strncpy): Adjust and simplify.
15746 (expand_builtin_memset): Adjust.
15747 (expand_builtin_bzero): Adjust.
15748 (expand_builtin_memcmp): Adjust.
15749 (expand_builtin): Handle memcmp.
15750 (maybe_emit_chk_warning): Check strncat just once.
15752 2017-05-04 Martin Sebor <msebor@redhat.com>
15754 PR preprocessor/79214
15755 PR middle-end/79222
15756 PR middle-end/79223
15757 * builtins.c (check_sizes): Add inlining context and issue
15758 warnings even when -Wno-system-headers is set.
15759 (check_strncat_sizes): Same.
15760 (expand_builtin_strncat): Same.
15761 (expand_builtin_memmove): New function.
15762 (expand_builtin_stpncpy): Same.
15763 (expand_builtin): Handle memmove and stpncpy.
15765 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15767 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15768 which is not used any more.
15770 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15772 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15774 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15776 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15777 (cortexa53_tunings): Likewise.
15778 (cortexa57_tunings): Likewise.
15779 (cortexa72_tunings): Likewise.
15780 (cortexa73_tunings): Likewise.
15782 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15784 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15785 Set loop alignment to 8.
15787 2017-05-04 Martin Sebor <msebor@redhat.com>
15789 PR translation/80280
15790 * builtins.c (expand_builtin_object_size): Add missing quoting to
15791 %D and like directives.
15792 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
15793 (hsa_type_for_tree_type): Same.
15794 (verify_function_arguments): Same.
15795 * symtab.c (symbol_table::change_decl_assembler_name): Same.
15796 * varasm.c (get_section): Same.
15799 2017-05-04 Martin Sebor <msebor@redhat.com>
15801 PR translation/80280
15802 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
15804 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15806 * config/aarch64/aarch64.c (generic_addrcost_table):
15807 Change HI/TI mode setting.
15809 2017-05-04 Martin Jambor <mjambor@suse.cz>
15811 PR tree-optimization/80622
15812 * tree-sra.c (comes_initialized_p): New function.
15813 (build_accesses_from_assign): Only set write lazily when
15814 comes_initialized_p is false.
15815 (analyze_access_subtree): Use comes_initialized_p.
15816 (propagate_subaccesses_across_link): Assert !comes_initialized_p
15817 instead of testing for PARM_DECL.
15819 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15821 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
15822 constraint on operand 0 to allow more general addressing modes.
15823 Adjust output template.
15824 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15826 * config/aarch64/aarch64-protos.h
15827 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15828 * config/aarch64/constraints.md (Dp): New address constraint.
15829 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15832 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15834 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15835 update use of estimate_ipcp_clone_size_and_time.
15836 (estimate_local_effects): Update use of
15837 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15838 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15839 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15840 Return nonspecialized time.
15842 2017-05-04 Richard Biener <rguenther@suse.de>
15844 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15845 for the last VUSE which def dominates the PHI. Directly call
15847 (get_continuation_for_phi_1): Remove.
15849 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15851 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15852 to explain the use of truncating division. Cap the number of
15853 iterations to the maximum given by nb_iterations_upper_bound,
15856 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15858 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15859 * configure: Regenerate.
15860 * config.in: Regenerate.
15861 * config/i386/driver-mingw32.c: new file.
15862 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15863 * config.host: Link driver-mingw32.o on MinGW host.
15864 * doc/install.texi: Document new --enable-mingw-wildcard configure
15867 2017-05-04 Marek Polacek <polacek@redhat.com>
15869 PR tree-optimization/80612
15870 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15872 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15873 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15875 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15876 (movt splitter): Likewise.
15877 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15878 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15879 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15880 block for Thumb-1 with MOVT.
15881 (thumb2_legitimate_address_p): Move code block ...
15882 (can_avoid_literal_pool_for_label_p): ... into this new function.
15883 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15885 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15886 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15887 "M-profile targets with the MOVT instruction".
15889 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15891 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15892 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15893 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15895 2017-05-04 Martin Liska <mliska@suse.cz>
15897 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15898 variable cond_code.
15900 2017-05-04 Richard Biener <rguenther@suse.de>
15902 * tree.c (array_at_struct_end_p): Handle arrays at struct
15903 end with flexarrays more conservatively. Refactor and treat
15904 arrays of arrays or aggregates more strict. Fix
15905 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15906 * tree.c (array_at_struct_end_p): Adjust prototype.
15907 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15908 * gimple-fold.c (get_range_strlen): Likewise.
15909 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15911 2017-05-04 Richard Biener <rguenther@suse.de>
15913 PR tree-optimization/31130
15914 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15916 (supports_overflow_infinity): Likewise.
15917 (is_negative_overflow_infinity): Likewise.
15918 (is_positive_overflow_infinity): Likewise.
15919 (is_overflow_infinity): Likewise.
15920 (stmt_overflow_infinity): Likewise.
15921 (overflow_infinity_range_p): Likewise.
15922 (usable_range_p): Remove as always returning true.
15923 (make_overflow_infinity): Remove.
15924 (negative_overflow_infinity): Likewise.
15925 (positive_overflow_infinity): Likewise.
15926 (avoid_overflow_infinity): Likewise.
15927 (set_value_range): Adjust accordingly.
15928 (set_value_range_to_nonnegative): Likewise, remove now unused
15929 overflow_infinity arg.
15930 (vrp_operand_equal_p): Adjust.
15931 (update_value_range): Likewise.
15932 (range_int_cst_singleton_p): Likewise.
15933 (operand_less_p): Likewise.
15934 (compare_values_warnv): Likewise.
15935 (extract_range_for_var_from_comparison_expr): Likewise.
15936 (vrp_int_const_binop): Likewise.
15937 (zero_nonzero_bits_from_vr): Likewise.
15938 (extract_range_from_multiplicative_op_1): Likewise.
15939 (extract_range_from_binary_expr_1): Likewise.
15940 (extract_range_from_unary_expr): Likewise.
15941 (extract_range_from_comparison): Likewise.
15942 (extract_range_basic): Likewise.
15943 (adjust_range_with_scev): Likewise.
15944 (compare_ranges): Likewise.
15945 (compare_range_with_value): Likewise.
15946 (dump_value_range): Likewise.
15947 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15949 (simplify_cond_using_ranges): Adjust.
15951 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15953 * brig-builtins.def: Added a builtin for class_f64.
15954 * builtin-types.def: Added a builtin type needed by class_f64.
15956 2017-05-03 Jason Merrill <jason@redhat.com>
15958 * timevar.def: Add TV_CONSTEXPR.
15960 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15962 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15964 2017-05-03 Martin Jambor <mjambor@suse.cz>
15966 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15967 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15968 * ipa-cp.c (ipcp_propagate_stage): Do not call
15969 ipa_update_after_lto_read.
15970 * ipa-inline.c (ipa_inline): Likewise.
15972 2017-05-03 Martin Jambor <mjambor@suse.cz>
15974 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15975 tag. Added a default constructor and a destructor.
15976 (ipa_edge_args_sum_t): New class;
15977 (ipa_edge_args_sum): Declare.
15978 (ipa_edge_args_vector): Remove declaration.
15979 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15980 (ipa_free_edge_args_substructures): Remove declaration.
15981 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15982 (ipa_edge_args_info_available_for_edge_p): Likewise.
15983 * ipa-prop.c (ipa_edge_args_vector): Removed.
15984 (edge_removal_hook_holder): Likewise.
15985 (edge_duplication_hook_holder): Likewise.
15986 (ipa_edge_args_sum): New variable.
15987 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15988 ipa_edge_args_vector.
15989 (ipa_free_edge_args_substructures): Likewise.
15990 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15991 ipa_edge_args_vector.
15992 (ipa_edge_removal_hook): Turned into method
15993 ipa_edge_args_sum_t::remove.
15994 (ipa_edge_duplication_hook): Turned into method
15995 ipa_edge_args_sum_t::duplicate.
15996 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15997 registering edge hooks.
15998 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15999 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16000 ipa_edge_args_sum instead of ipa_edge_args_vector.
16001 * ipa-profile.c (ipa_profile): Likewise.
16003 2017-05-03 Martin Jambor <mjambor@suse.cz>
16005 * symbol-summary.h (function_summary): New method exists.
16006 (function_summary::symtab_removal): Deallocate through release.
16007 (call_summary): New class.
16008 (gt_ggc_mx): New overload.
16009 (gt_pch_nx): Likewise.
16010 (gt_pch_nx): Likewise.
16012 2017-05-03 Jeff Law <law@redhat.com>
16014 PR tree-optimization/78496
16015 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16016 from simplify_cond_using_ranges. Split off code to walk
16017 backwards through casts into ...
16018 (simplify_cond_using_ranges_2): New function.
16019 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16020 (execute_vrp): After identifying jump threads, call
16021 simplify_cond_using_ranges_2.
16023 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16026 * ipa-inline.h (inline_summary): Add ctor.
16027 (create_ggc): Do not use ggc_cleared_alloc.
16029 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16030 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16032 * gcc.c (handle_braces): Support escaping in switch matching
16034 * doc/invoke.texi (Spec Files): Document it.
16035 Remove superfluous @code markup in items.
16037 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16039 * diagnostic-show-locus.c (struct column_range): New struct.
16040 (get_affected_columns): New function.
16041 (get_printed_columns): New function.
16042 (struct correction): New struct.
16043 (correction::ensure_capacity): New function.
16044 (correction::ensure_terminated): New function.
16045 (struct line_corrections): New struct.
16046 (line_corrections::~line_corrections): New dtor.
16047 (line_corrections::add_hint): New function.
16048 (layout::print_trailing_fixits): Reimplement in terms of the new
16050 (selftest::test_overlapped_fixit_printing): New function.
16051 (selftest::diagnostic_show_locus_c_tests): Call it.
16053 2017-05-03 Nathan Sidwell <nathan@acm.org>
16055 Canonicalize canonical type hashing
16056 * tree.h (type_hash_canon_hash): Declare.
16057 * tree.c (type_hash_list, attribute_hash_list): Move into
16058 type_hash_canon_hash.
16059 (build_type_attribute_qual_variant): Break out hash code calc into
16060 type_hash_canon_hash.
16061 (type_hash_canon_hash): New. Generic type hash computation.
16062 (build_range_type_1, build_array_type_1, build_function_type,
16063 build_method_type_directly, build_offset_type, build_complex_type,
16064 make_vector_type): Call it.
16066 2017-05-03 Richard Biener <rguenther@suse.de>
16068 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16069 When all DRs have unknown misaligned do not always peel
16070 when there is a store but apply the same costing model as if
16071 there were only loads.
16073 2017-05-03 Richard Biener <rguenther@suse.de>
16076 PR tree-optimization/80492
16077 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16078 compare_base_decls returning dont-know properly.
16080 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16082 * config/arm/iterators.md (CCSI): New mode iterator.
16083 (arch): New mode attribute.
16084 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16085 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16086 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16087 code iterator for success result mode.
16088 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16089 the corresponding new insn generators.
16091 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16094 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16095 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16097 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16099 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16100 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16101 (DDR_A): Wrap DDR argument in brackets.
16102 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16103 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16104 (DDR_REVERSED_P): Likewise.
16106 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16108 PR tree-optimization/79472
16109 * tree-switch-conversion.c (struct switch_conv_info): Add
16110 contiguous_range and default_case_nonstandard fields.
16111 (collect_switch_conv_info): Compute contiguous_range and
16112 default_case_nonstandard fields, don't clear final_bb if
16113 contiguous_range and only the default case doesn't have the required
16115 (check_all_empty_except_final): Set default_case_nonstandard instead
16116 of failing if contiguous_range and the default case doesn't have empty
16118 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16119 and only the default case doesn't have the required constants. Skip
16121 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16122 if default_case_nonstandard.
16123 (build_constructors): Build constant 1 just once. Assert that default
16124 values aren't inserted in between cases if contiguous_range. Skip
16126 (build_arrays): Skip virtual phis.
16127 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16128 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16129 Handle virtual phis.
16130 (gen_inbound_check): Handle default_case_nonstandard case.
16131 (process_switch): Adjust check_final_bb caller. Call
16132 gather_default_values with the first non-default case instead of
16133 default case if default_case_nonstandard.
16135 2017-05-02 Nathan Sidwell <nathan@acm.org>
16137 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16138 check. Fix formatting.
16140 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16142 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16143 errors when comparing specialized and unspecialized times.
16145 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16147 * diagnostic-show-locus.c
16148 (layout::should_print_annotation_line_p): Make private.
16149 (layout::print_annotation_line): Make private.
16150 (layout::annotation_line_showed_range_p): Make private.
16151 (layout::show_ruler): Make private.
16152 (layout::print_source_line): Make private. Pass in line and
16153 line_width, rather than calling location_get_source_line. Drop
16155 (layout::print_leading_fixits): New method.
16156 (layout::print_any_fixits): Rename to...
16157 (layout::print_trailing_fixits): ...this, and make private.
16158 Don't print newline fixits.
16159 (diagnostic_show_locus): Move logic for printing one row into...
16160 (layout::print_line): ...this new function. Move the
16161 location_get_source_line call and error-handling from
16162 print_source_line to here. Call print_leading_fixits, and rename
16163 print_any_fixits to print_trailing_fixits.
16164 (selftest::test_fixit_insert_containing_newline): Update now that
16165 newlines are partially supported.
16166 (selftest::test_fixit_insert_containing_newline_2): New test.
16167 (selftest::test_fixit_replace_containing_newline): Update comments.
16168 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16169 * edit-context.c (class added_line): New class.
16170 (class edited_line): Describe newline handling in comment.
16171 (edited_line::actually_edited_p): New method.
16172 (edited_line::print_content): Delete redundant decl.
16173 (edited_line::m_predecessors): New field.
16174 (edited_file::print_content): Call edited_line::print_content.
16175 (edited_file::print_diff): Update to support newlines.
16176 (edited_file::print_diff_hunk): Likewise.
16177 (edited_file::print_run_of_changed_lines): New function.
16178 (edited_file::print_diff_line): Convert to...
16179 (print_diff_line): ...this.
16180 (edited_file::get_effective_line_count): New function.
16181 (edited_line::edited_line): Initialize new field m_predecessors.
16182 (edited_line::~edited_line): Clean up m_predecessors.
16183 (edited_line::apply_fixit): Handle newlines.
16184 (edited_line::get_effective_line_count): New function.
16185 (edited_line::print_content): New function.
16186 (edited_line::print_diff_lines): New function.
16187 (selftest::test_applying_fixits_insert_containing_newline): New
16189 (selftest::test_applying_fixits_replace_containing_newline): New
16191 (selftest::insert_line): New function.
16192 (selftest::test_applying_fixits_multiple_lines): Add example of
16194 (selftest::edit_context_c_tests): Call the new tests.
16196 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16198 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16199 parameter cand. Update dump information.
16200 (get_computation_cost): Update uses.
16202 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16204 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16205 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16206 (get_computation_at, rewrite_use_address): Update use of
16207 get_computation_aff.
16209 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16211 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16212 (get_computation): Delete.
16213 (get_computation_cost): Implement like get_computation_cost_at.
16214 Use get_computation_at.
16215 (get_computation_cost_at): Delete.
16216 (rewrite_use_nonlinear_expr): Use get_computation_at.
16217 (rewrite_use_compare, remove_unused_ivs): Ditto.
16219 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16221 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16223 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16225 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16226 (ivopts_global_cost_for_size): Rename parameter and update uses.
16227 (iv_ca_recount_cost): Update uses.
16228 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16229 candidates seperately in n_invs and n_cands.
16230 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16232 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16234 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16235 (find_inv_vars_cb): New.
16236 (find_depends): Renamed to ...
16237 (find_inv_vars): ... this.
16238 (add_candidate_1, force_var_cost): Call find_inv_vars.
16239 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16241 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16243 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16244 inv_vars. Add inv_exprs.
16245 (struct iv_cand): Rename depends_on to inv_vars.
16246 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16247 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16248 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16249 (dump_cand): Dump inv_vars.
16250 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16251 (record_invariant, find_depends, add_candidate_1): Ditto.
16252 (set_group_iv_cost, force_var_cost): Ditto.
16253 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16254 (get_computation_cost_at, get_computation_cost): Ditto.
16255 (determine_group_iv_cost_generic): Ditto.
16256 (determine_group_iv_cost_address): Ditto.
16257 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16258 (determine_group_iv_costs): Ditto.
16259 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16260 (iv_ca_set_remove_invariants): Renamed to ...
16261 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16262 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16263 (iv_ca_set_add_invariants): Renamed to ...
16264 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16265 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16266 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16267 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16268 (create_new_ivs): Remove useless dump.
16270 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16272 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16274 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16275 (iv_ca_set_no_cp, create_new_iv): Ditto.
16277 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16279 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16281 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16283 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16284 function tree_check2.
16286 2017-05-02 Martin Liska <mliska@suse.cz>
16288 * doc/gcov.texi: Add missing preposition.
16289 * gcov.c (function_info::function_info): Properly fill up
16290 all member variables.
16292 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16294 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16296 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16298 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16300 2017-05-02 Martin Liska <mliska@suse.cz>
16303 * lto-streamer-in.c (lto_read_tree_1): Remove
16304 LTO_STREAMER_DEBUG.
16305 * lto-streamer.c (struct tree_hash_entry): Likewise.
16306 (struct tree_entry_hasher): Likewise.
16307 (tree_entry_hasher::hash): Likewise.
16308 (tree_entry_hasher::equal): Likewise.
16309 (lto_streamer_init): Likewise.
16310 (lto_orig_address_map): Likewise.
16311 (lto_orig_address_get): Likewise.
16312 (lto_orig_address_remove): Likewise.
16313 * lto-streamer.h: Likewise.
16314 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16315 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16317 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16319 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16320 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16321 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16322 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16323 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16324 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16325 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16326 (mm_maskz_sub_ss): New intrinsics.
16327 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16328 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16329 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16330 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16331 (__builtin_ia32_subss_mask_round): New builtins.
16332 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16333 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16334 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16336 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16337 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16339 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16342 2017-05-02 Martin Jambor <mjambor@suse.cz>
16344 PR tree-optimization/78687
16345 * tree-sra.c (access): New field parent.
16346 (process_subtree_disqualification): New function.
16347 (disqualify_candidate): Call it.
16348 (build_accesses_from_assign): Reset write flag if creating an
16350 (build_access_subtree): Fill in parent field and also prpagate
16351 down grp_write flag.
16352 (create_artificial_child_access): New parameter set_grp_write, set
16353 grp_write to its value.
16354 (propagate_subaccesses_across_link): Also propagate grp_write flag
16356 (propagate_all_subaccesses): Push the closest parent back to work
16357 queue if add_access_to_work_queue returned true.
16359 2017-05-02 Richard Biener <rguenther@suse.de>
16361 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16362 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16363 -fstrict-overflow documentation.
16364 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16365 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16366 flag_strict_overflow.
16367 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16368 * lto-opts.c (lto_write_options): Do not stream it.
16369 * lto-wrapper.c (merge_and_complain): Do not handle it.
16370 * opts.c (default_options_table): Do not set -fstrict-overflow.
16371 (finish_options): Likewise do not clear it when sanitizing.
16372 * simplify-rtx.c (simplify_const_relational_operation): Do not
16373 test flag_strict_overflow.
16375 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16377 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16378 using enabled attribute.
16379 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16380 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16381 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16382 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16383 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16384 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16385 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16386 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16387 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16388 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16390 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16392 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16394 2017-05-02 Richard Biener <rguenther@suse.de>
16396 PR tree-optimization/80591
16398 2017-04-10 Richard Biener <rguenther@suse.de>
16400 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16403 2017-05-02 Richard Biener <rguenther@suse.de>
16405 PR tree-optimization/80549
16406 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16407 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16408 headers if they do not have a preheader.
16410 2017-05-02 Martin Liska <mliska@suse.cz>
16413 * common.opt: Fix typo.
16414 * doc/invoke.texi: Likewise.
16416 2017-05-01 Jan Beulich <jbeulich@suse.com>
16418 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16419 swapping, add (x,x,m,x,n) alternative.
16421 2017-05-01 Nathan Sidwell <nathan@acm.org>
16423 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16424 unnecessary unadjusted_alignment check.
16426 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16429 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16431 * gimplify.c (gimplify_cilk_detach): New function.
16432 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16433 * tree-core.h: Document EXPR_CILK_SPAWN.
16434 * tree.h (EXPR_CILK_SPAWN): Define.
16436 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16438 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16439 to use new fixit_hint representation, using the "replace" logic.
16440 (get_line_span_for_fixit_hint): Likewise.
16441 (layout::print_any_fixits): Likewise.
16442 (selftest::test_one_liner_many_fixits): Rename to...
16443 (selftest::test_one_liner_many_fixits_1): ...this, and update
16444 comment and expected output to reflect that the multiple fix-it
16445 hints are now consolidated into one insertion.
16446 (selftest::test_one_liner_many_fixits_2): New test.
16447 (selftest::test_diagnostic_show_locus_one_liner): Update for
16449 (selftest::test_fixit_consolidation): Update for fix-it API
16451 * diagnostic.c (print_parseable_fixits): Likewise.
16452 * edit-context.c (edited_line::m_line_events): Convert from
16453 auto_vec <line_event *> to auto_vec <line_event>.
16454 (class line_event): Convert from abstract base class to a concrete
16455 class, taking over the role of replace_event.
16456 (class insert_event): Delete.
16457 (class replace_event): Rename to class line_event. Convert to
16459 (edit_context::add_fixits): Reimplement.
16460 (edit_context::apply_insert): Delete.
16461 (edit_context::apply_replace): Rename to...
16462 (edit_context::apply_fixit): ...this. Convert to half-open range.
16463 (edited_file::apply_insert): Delete.
16464 (edited_file::apply_replace): Rename to...
16465 (edited_file::apply_fixit): ...this.
16466 (edited_line::~edited_line): Drop deletion of events.
16467 (edited_line::apply_insert): Delete.
16468 (edited_line::apply_replace): Rename to...
16469 (edited_line::apply_fixit): ...this. Convert to half-open range.
16470 Update for change to type of m_line_events.
16471 * edit-context.h (edit_context::apply_insert): Delete.
16472 (edit_context::apply_replace): Rename to...
16473 (edit_context::apply_fixit): ...this.
16475 2017-05-01 Martin Sebor <msebor@redhat.com>
16477 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16480 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16483 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16484 __get_cpuid_max returns 0.
16485 (__get_cpuid_count): Ditto.
16487 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16489 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16490 replacement expression is another instance of one of its arguments.
16492 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16495 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16496 check for stack push/pop autoinc.
16497 * config/i386/i386.c (ix86_agi_dependent): Return false
16498 if the only reason why modified_in_p returned true is that
16499 addr is SP based and set_insn is a push or pop.
16501 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16503 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16506 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16509 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16510 (account_size_time): Use two predicates - exec_pred and
16512 (evaluate_conditions_for_known_args): Compute both clause and
16514 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16515 (inline_summary_t::duplicate): Update.
16516 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16518 (compute_inline_parameters): Likewise.
16519 (estimate_edge_size_and_time): Update caluclation of time.
16520 (estimate_node_size_and_time): Compute both time and nonspecialized
16522 (estimate_ipcp_clone_size_and_time): Update.
16523 (inline_merge_summary): Update.
16524 (do_estimate_edge_time): Update.
16525 (do_estimate_edge_size): Update.
16526 (do_estimate_edge_hints): Update.
16527 (inline_read_section, inline_write_summary): Stream both new predicates.
16528 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16530 (compute_inlined_call_time): Cleanup.
16531 (big_speedup_p): Update.
16532 (edge_badness): Update.
16533 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16534 (size_time_entry): Replace predicate by exec_predicate and
16535 nonconst_predicate.
16536 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16537 (estimate_edge_time): Return also nonspec_time.
16538 (reset_edge_growth_cache): Update.
16540 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16542 PR rtl-optimization/80491
16543 * ifcvt.c (noce_process_if_block): When looking for x setter
16544 with missing else_bb, don't check only the insn right before
16545 cond_earliest, but look for the last insn that x is modified in
16546 within the same bb.
16548 PR rtl-optimization/80491
16549 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16551 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16553 PR tree-optimization/80487
16554 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16556 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16558 PR tree-optimization/79697
16559 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16560 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16561 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16563 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16564 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16566 2017-04-28 Martin Sebor <msebor@redhat.com>
16568 PR tree-optimization/80523
16569 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16570 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16572 (maybe_warn, format_directive, parse_directive): Use new functions.
16573 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16575 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16577 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16579 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16581 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16582 target_header_dir): Set correctly.
16583 * configure: Regenerated.
16584 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16585 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16586 instead of SYSTEM_HEADER_DIR.
16588 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16590 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16591 (estimate_local_effects): Likewise.
16592 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16593 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16594 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16595 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16596 do_estimate_edge_time, estimate_edge_time): Likewise.
16597 * ipa-inline-analysis.c (estimate_node_size_and_time,
16598 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16599 (estimate_time_after_inlining): Remove.
16601 2017-04-28 Martin Liska <mliska@suse.cz>
16603 * doc/gcov.texi: Enhance documentation of gcov.
16605 2017-04-28 Martin Liska <mliska@suse.cz>
16607 * doc/gcov.texi: Sort options in alphabetic order.
16608 * doc/gcov-dump.texi: Likewise.
16609 * doc/gcov-tool.texi: Likewise.
16610 * gcov.c (print_usage): Likewise.
16611 * gcov-dump.c (print_usage): Likewise.
16612 * gcov-tool.c (print_merge_usage_message): Likewise.
16613 (print_rewrite_usage_message): Likewise.
16614 (print_overlap_usage_message): Likewise.
16616 2017-04-28 Martin Liska <mliska@suse.cz>
16618 PR gcov-profile/53915
16619 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16621 2017-04-28 Martin Liska <mliska@suse.cz>
16623 PR gcov-profile/79891
16624 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16625 is marked by compiler as living on a line.
16626 (get_cycles_count): Remove usage of the union.
16627 (output_intermediate_file): Likewise.
16628 (find_source): Fix GNU coding style.
16629 (accumulate_line_counts): Remove old non-all block mode.
16630 (output_lines): Remove usage of the union.
16631 * profile.c (output_location): Include all BBs, even if
16632 belonging to a same line (and file) as a previous BB.
16634 2017-04-28 Martin Liska <mliska@suse.cz>
16636 * gcov.c (process_args): Handle new argument 'w'.
16637 (read_graph_file): Assign ID to BBs.
16638 (output_branch_count): Display BB # if verbose flag is set.
16639 (output_lines): Likewise for arcs.
16640 (print_usage): Add '--verbose' option help.
16641 * doc/gcov.texi: Document --verbose (-w) option.
16643 2017-04-28 Martin Liska <mliska@suse.cz>
16645 * gcov.c (struct block_location_info): New struct.
16646 (process_file): Fill up the new structure.
16647 (read_graph_file): Replace usage of encoding by the newly added
16649 (add_line_counts): Likewise.
16650 (accumulate_line_counts): Remove usage of the union.
16651 (function_info::function_info): New function.
16652 (function_info::~function_info): Likewise.
16653 (process_file): Call delete instead of release_function.
16654 (release_function): Release the function.
16655 (release_structures): Call delete instead of release_function.
16656 (solve_flow_graph): Replace usage of num_blocks.
16657 (find_exception_blocks): Likewise.
16658 (output_lines): Fix GNU coding style.
16660 2017-04-28 Martin Liska <mliska@suse.cz>
16663 * coverage.c (coverage_remove_note_file): New function.
16664 * coverage.h: Declare the function.
16665 * toplev.c (finalize): Clean if an error has been seen.
16667 2017-04-28 Martin Liska <mliska@suse.cz>
16669 PR gcov-profile/80031
16670 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16671 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16672 * gcov.c (read_graph_file): Read just number of blocks.
16673 * profile.c (branch_prob): Do not stream 0 flags per a basic
16676 2017-04-28 Martin Liska <mliska@suse.cz>
16678 * gcov-dump.c (tag_*): Add new argument to declarations.
16679 (dump_gcov_file): Likewise.
16680 (tag_blocks): Add and use new argument depth.
16681 (tag_arcs): Likewise.
16682 (tag_lines): Likewise.
16683 (tag_counters): Likewise.
16684 (tag_summary): Likewise.
16685 (dump_working_sets): Use depth to do a proper indentation.
16687 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16690 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16691 * symtab.c (symtab_node::debug_symtab): Move definition here.
16693 2017-04-28 Richard Biener <rguenther@suse.de>
16695 * lto-streamer.h (LTO_major_version): Bump to 7.
16697 2017-04-28 Richard Biener <rguenther@suse.de>
16699 * tree-vrp.c (assert_info): New struct.
16700 (add_assert_info): New helper.
16701 (register_edge_assert_for_2): Refactor to add asserts to a vector
16703 (register_edge_assert_for_1): Likewise.
16704 (register_edge_assert_for): Likewise.
16705 (finish_register_edge_assert_for): New helper actually registering
16706 asserts where live on edge.
16707 (find_conditional_asserts): Adjust.
16708 (find_switch_asserts): Likewise.
16709 (evrp_dom_walker::try_find_new_range): Generalize.
16710 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16712 2017-04-27 Marek Polacek <polacek@redhat.com>
16715 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16716 arg10 and arg11 to itype.
16718 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16720 * doc/extend.texi (Object Size Checking): Improve grammar.
16722 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16725 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16726 that the logic for permitting reciprocal estimates matches that
16729 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16732 * tree.c (type_cache_hasher::equal): Only compare
16733 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16734 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16735 non-aggregate element types.
16736 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16737 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16739 2017-04-27 Richard Biener <rguenther@suse.de>
16741 PR middle-end/80533
16742 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16743 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16744 keeping a reference to a trailing array.
16746 2017-04-27 Richard Biener <rguenther@suse.de>
16748 PR middle-end/80539
16749 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16750 being in loop-closed SSA form conservatively.
16751 (chrec_fold_multiply_poly_poly): Likewise.
16753 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16755 PR middle-end/79665
16756 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16757 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16759 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16762 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16763 (aarch64_function_arg_alignment): Return unsigned int again, but still
16764 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16765 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16766 Don't emit -Wpsabi note.
16767 (aarch64_function_arg_boundary): Likewise.
16768 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16771 2017-04-26 Nathan Sidwell <nathan@acm.org>
16773 * tree.h (crc32_unsigned_n): Declare.
16774 (crc32_unsigned, crc32_unsigned): Make inline.
16775 * tree.c (crc32_unsigned_bits): Replace with ...
16776 (crc32_unsigned_n): ... this.
16777 (crc32_unsigned, crc32_byte): Remove.
16778 (crc32_string): Remove unnecessary braces.
16780 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16782 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16783 * ipa-inline-analysis.c (MAX_TIME): Remove.
16784 (account_size_time): Use sreal for time.
16785 (dump_inline_summary): Update.
16786 (estimate_function_body_sizes): Update.
16787 (estimate_edge_size_and_time): Update.
16788 (estimate_calls_size_and_time): Update.
16789 (estimate_node_size_and_time): Update.
16790 (inline_merge_summary): Update.
16791 (inline_update_overall_summary): Update.
16792 (estimate_time_after_inlining): Update.
16793 (inline_read_section): Update.
16794 (inline_write_summary): Update.
16795 * ipa-inline.c (compute_uninlined_call_time): Update.
16796 (compute_inlined_call_time): Update.
16797 (recursive_inlining): Update.
16798 (inline_small_functions): Update.
16799 (dump_overall_stats): Update.
16800 * ipa-inline.h: Include sreal.h.
16801 (size_time_entry): Turn time to sreal.
16802 (inline_summary): Turn self_time nad time to sreal.
16804 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16806 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
16808 (sreal::stream_out, sreal::stream_in): New.
16809 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
16811 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16813 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
16816 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
16819 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
16821 (dimode_scalar_chain::compute_convert_gain): Ditto.
16822 (dimode_scalar_chain::make_vector_copies): Ditto.
16823 (dimode_scalar_chain::convert_reg): Ditto.
16824 (dimode_scalar_chain::convert_insn): Ditto.
16825 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16826 (VI248_AVX512BW_1): New mode iterator.
16827 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16828 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16831 2017-04-25 Martin Sebor <msebor@redhat.com>
16833 PR tree-optimization/80497
16834 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16835 constants are representable in HOST_WIDE_INT.
16836 (parse_directive): Ditto.
16838 2017-04-25 Martin Sebor <msebor@redhat.com>
16841 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16842 (new_zero_array): Adjust signature.
16843 (dom_info::dom_init): Used unsigned rather that size_t.
16844 (dom_info::dom_info): Same.
16846 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16847 Jakub Jelinek <jakub@redhat.com>
16850 * config/arm/arm.c: Include gimple.h.
16851 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16852 returns negative, increment ncrn only if it returned positive.
16853 (arm_needs_doubleword_align): Return int instead of bool,
16854 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16855 members, but if there is any such non-FIELD_DECL
16856 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16857 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16858 returns negative, increment nregs only if it returned positive.
16859 (arm_setup_incoming_varargs): Likewise.
16860 (arm_function_arg_boundary): Emit -Wpsabi note if
16861 arm_needs_doubleword_align returns negative, return
16862 DOUBLEWORD_ALIGNMENT only if it returned positive.
16864 2017-04-25 Marek Polacek <polacek@redhat.com>
16867 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16868 first argument to type.
16870 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16873 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16874 type checks to test for compatibility instead of equality.
16876 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16877 Jakub Jelinek <jakub@redhat.com>
16880 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16882 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16883 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16884 the alignment computation, but return their maximum in warn_alignment.
16885 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16886 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16888 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16889 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16892 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16894 * config/arc/simdext.md (dmpyh): Fix typo.
16896 2017-04-25 Richard Biener <rguenther@suse.de>
16898 PR tree-optimization/80492
16899 * alias.c (compare_base_decls): Handle registers with asm
16900 specification conservatively.
16901 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16902 compare_base_decls returning dont-know properly.
16904 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16906 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16907 (legitimate_offset_address_p): New function.
16908 (arc_legitimate_address_p): Use above function.
16910 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16912 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16914 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16916 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16917 ACCH registers whenever they are available.
16919 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16921 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16922 double regs fix when not used.
16924 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16926 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16928 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16929 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16931 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16933 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16934 when emitting short ADD instructions.
16936 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16938 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16940 (cmpsi_cc_c_insn): Likewise.
16941 (cbranchsi4_scratch): Compute proper instruction length using
16942 compact_hreg_operand.
16943 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16945 2017-04-25 Richard Biener <rguenther@suse.de>
16947 PR middle-end/80509
16948 * passes.c (pass_manager::pass_manager): Initialize
16949 m_name_to_pass_map.
16951 2017-04-25 Richard Biener <rguenther@suse.de>
16953 PR tree-optimization/79201
16954 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16956 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16959 * config/s390/vector.md: Split MEM->GPR vector moves for
16960 non-s_operand addresses.
16962 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16965 * config/s390/predicates.md (reload_const_wide_int_operand): New
16967 * config/s390/s390.md ("movti"): Remove d/P alternative.
16968 ("movti_bigconst"): New pattern definition.
16970 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16973 * s390-protos.h (s390_expand_cs_hqi): Removed.
16974 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16975 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16976 modes as well as CCZ1mode and CCZmode.
16977 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16978 signature of s390_emit_compare_and_swap.
16979 (s390_expand_cs_hqi): Likewise, make static.
16980 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16981 compare-and-swap, in some cases.
16982 (s390_expand_cs): Wrapper function.
16983 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16985 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16986 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16987 patterns for small and large integers. Forbid symref memory operands.
16988 Move expander to s390.c. Require cc register.
16989 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16990 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16991 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16992 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16993 symref memory operands. Remove CC mode and call s390_match_ccmode
16995 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16997 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16999 * config/s390/s390.md (define_peephole2): New peephole to help
17000 combining the load-and-test pattern with volatile memory.
17002 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17004 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17005 with CCZmode for TARGET_Z196.
17007 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17009 PR rtl-optimization/80501
17010 * combine.c (make_compound_operation_int): Set subreg_code to SET
17011 even for AND with mask of the sign bit of mode.
17013 PR rtl-optimization/80500
17014 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17015 sum's initial value.
17017 2017-04-25 Julian Brown <julian@codesourcery.com>
17018 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17020 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17022 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17024 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17026 2017-04-25 Julian Brown <julian@codesourcery.com>
17027 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17029 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17030 (thunderx2t99_sha): New Reservation.
17032 2017-04-25 Julian Brown <julian@codesourcery.com>
17033 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17035 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17036 type for 1-element load.
17038 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17040 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17042 2017-04-24 Martin Jambor <mjambor@suse.cz>
17044 PR tree-optimization/80293
17045 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17046 char arrays not totally scalarizable if it is false.
17047 (analyze_all_variable_accesses): Pass correct value in the new
17048 parameter. Add a statistics counter.
17050 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17052 PR middle-end/79931
17053 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17055 2017-04-24 Richard Biener <rguenther@suse.de>
17057 PR tree-optimization/80494
17058 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17059 out for complex types.
17061 2017-04-24 Richard Biener <rguenther@suse.de>
17063 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17064 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17065 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17066 (DFS): Adjust and never fail.
17067 (sccvn_dom_walker::fail): Remove.
17068 (sccvn_dom_walker::before_dom_children): Adjust.
17069 (run_scc_vn): Likewise and never fail.
17070 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17071 (pass_fre::execute): Likewise.
17073 2017-04-24 Richard Biener <rguenther@suse.de>
17075 PR tree-optimization/79725
17076 * tree-ssa-sink.c (statement_sink_location): Return whether
17077 failure reason was zero uses. Move that check later.
17078 (sink_code_in_bb): Deal with zero uses by removing the stmt
17081 2017-04-24 Richard Biener <rguenther@suse.de>
17084 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17085 pointer-based references.
17087 2017-04-24 Richard Biener <rguenther@suse.de>
17090 * pass_manager.h (pass_manager::operator new): Remove.
17091 (pass_manager::operator delete): Likewise.
17092 * passes.c (pass_manager::operator new): Remove.
17093 (pass_manager::operator delete): Likewise.
17094 (pass_manager::pass_manager): Zero individual pass members.
17096 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17099 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17100 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17101 Check "XEXP (src, 1)" operand here.
17102 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17103 Check "XEXP (src, 1)" operand here.
17104 (dimode_scalar_chain::make_vector_copies): Detect count register
17105 of a shift instruction. Zero extend count register from QImode
17106 to DImode to satisfy vector shift pattern count operand predicate.
17107 Substitute vector shift count operand with a DImode copy.
17108 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17111 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17113 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17114 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17115 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17116 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17117 (UNSPEC_NOREX_MEM): Remove definition.
17119 2017-04-21 Richard Biener <rguenther@suse.de>
17121 PR tree-optimization/79547
17122 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17123 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17124 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17125 without any constraints.
17127 2017-04-21 Richard Biener <rguenther@suse.de>
17129 PR tree-optimization/78847
17130 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17132 2017-04-21 Richard Biener <rguenther@suse.de>
17134 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17135 (build_distinct_type_copy): Likewise.
17136 (build_variant_type_copy): Likewise.
17137 * tree.c (build_qualified_type): Pass down mem-stat info.
17138 (build_distinct_type_copy): Likewise.
17139 (build_variant_type_copy): Likewise.
17141 2017-04-21 Richard Biener <rguenther@suse.de>
17143 PR tree-optimization/80237
17144 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17146 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17147 for a simplified result.
17149 2016-04-21 Richard Biener <rguenther@suse.de>
17151 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17152 sth as strict as a simple_iv but a chrec without symbols and an
17153 operand defined in the loop we are peeling (and not some subloop).
17154 (propagate_constants_for_unrolling): Propagate all constants.
17156 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17159 * config/i386/i386.c (print_reg): Remove assert for disalowed
17160 regno values, call output_operand_lossage instead.
17162 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17165 * config/i386/constraints.md (Yc): New register constraint.
17166 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17167 Use Yc constraint for alternative 2 of operand 0. Remove
17168 preferred_for_speed attribute.
17170 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17172 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17173 lastprivate clauses in SIMT case.
17175 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17177 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17179 2017-04-20 Richard Biener <rguenther@suse.de>
17181 PR tree-optimization/57796
17182 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17183 as N scalar stores.
17184 (vect_model_load_cost): Cost gathers as N scalar loads.
17186 2017-04-20 Richard Biener <rguenther@suse.de>
17188 * ggc-page.c (ggc_allocated_p): Rename to ...
17189 (safe_lookup_page_table_entry): ... this and return the lookup
17191 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17193 2017-04-20 Richard Biener <rguenther@suse.de>
17195 PR tree-optimization/80453
17196 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17197 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17198 from the conditions.
17199 (vn_phi_eq): Pass them down.
17200 (vn_phi_lookup): Record them.
17201 (vn_phi_insert): Likewise.
17203 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17205 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17206 uninitialized variable warning to avoid buffer overrun.
17208 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17211 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17212 is suppressed for '{ 0 }' in C.
17214 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17216 * BASE-VER: Set to 8.0.0.
17218 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17220 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17221 priority .init_array and .fini_array section with SECTION_NOTYPE
17224 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17226 PR middle-end/80423
17227 * tree.h (build_array_type): Add typeless_storage default argument.
17228 * tree.c (type_cache_hasher::equal): Also compare
17229 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17230 (build_array_type): Add typeless_storage argument, set
17231 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17233 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17234 (build_array_type): Likewise. Add typeless_storage argument.
17236 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17237 Jakub Jelinek <jakub@redhat.com>
17239 PR tree-optimization/80426
17240 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17241 operation on symbolic operands, also compute the overflow for the
17242 invariant part when the operation degenerates into a negation.
17244 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17247 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17248 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17251 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17253 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17256 * config/avr/avr.c (tree.h): Include it.
17257 (cgraph.h): Include it.
17258 (avr_encode_section_info): Don't warn for uninitialized progmem
17259 variable if it's just an alias.
17261 2017-04-19 Richard Biener <rguenther@suse.de>
17264 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17265 when needed by AutoPGO.
17267 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17270 * doc/lto.texi: Remove an extra 'that'.
17272 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17274 PR rtl-optimization/80429
17275 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17276 are only used in debug insns.
17278 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17279 Vladimir Makarov <vmakarov@redhat.com>
17281 * config/sparc/predicates.md (input_operand): Add comment. Return
17282 true for any memory operand when LRA is in progress.
17283 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17285 2017-04-18 Jeff Law <law@redhat.com>
17288 * mips.md ({return,simple_return}_internal): Do not overwrite
17291 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17293 PR tree-optimization/80443
17294 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17295 instead of adding 1, subtract -1 and similarly instead of subtracting
17298 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17300 PR rtl-optimization/80357
17301 * haifa-sched.c (tmp_bitmap): New variable.
17302 (model_recompute): Handle duplicate use records.
17303 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17304 (free_global_sched_pressure_data): Free it.
17306 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17309 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17310 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17311 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17312 instead of SYSTEM_HEADER_DIR.
17314 2017-04-18 Jeff Law <law@redhat.com>
17316 PR middle-end/80422
17317 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17318 predecessors after walking up the insn chain.
17320 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17323 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17324 sizetype type into debug info.
17326 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17329 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17330 unneeded test for TARGET_UPPER_REGS_SF.
17331 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17333 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17336 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17337 instead of gsi_after_labels.
17339 2017-04-18 Jeff Law <law@redhat.com>
17341 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17345 2017-04-13 Jeff Law <law@redhat.com>
17346 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17347 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17349 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17352 * config/avr/avr.c (intl.h): Include it.
17353 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17355 2017-04-18 Martin Liska <mliska@suse.cz>
17357 PR gcov-profile/78783
17358 * gcov-tool.c (gcov_output_files): Validate that destination
17359 file is either removed by the tool or by a user.
17361 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17362 Guy Benyei <guybe@mellanox.com>
17364 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17365 block, and do not negate it, the stored id is already negative.
17367 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17369 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17371 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17374 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17375 masks of options that should be turned off if the VSX vector
17376 options are turned off.
17377 (OTHER_P8_VECTOR_MASKS): Likewise.
17378 (OTHER_VSX_VECTOR_MASKS): Likewise.
17379 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17380 rs6000_disable_incompatible_switches to validate no type switches
17382 (rs6000_incompatible_switch): New function to disallow turning on
17383 other vector options if -mno-vsx, -mno-power8-vector, or
17384 -mno-power9-vector are specified.
17386 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17388 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17390 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17392 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17393 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17394 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17395 (ARG_POINTER_CFA_OFFSET): Likewise.
17397 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17399 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17400 conditions to take advantage of various optimizations.
17402 2017-04-13 Jeff Law <law@redhat.com>
17404 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17405 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17406 (zero_extendsidi2_dext): Likewise.
17408 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17411 * fold-const.c (fold_ternary_loc): Revert
17412 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17415 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17417 PR rtl-optimization/80343
17418 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17419 new scratch pseudo.
17421 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17424 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17425 to ubsan_encode_value.
17427 2017-04-13 Jeff Law <law@redhat.com>
17429 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17430 appearing in DEBUG_INSNs.
17432 2017-04-13 Martin Liska <mliska@suse.cz>
17434 PR gcov-profile/80413
17435 * gcov-io.c (gcov_write_string): Copy to buffer just when
17436 allocated size is greater than zero.
17438 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17441 * dwarf2out.c (decls_for_scope): Ignore declarations of
17442 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17444 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17447 * ipa-visibility.c (non_local_p): Fix typos.
17448 (localize_node): When localizing symbol in same comdat group,
17449 dissolve the group only when we know external symbols are going
17451 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17453 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17455 PR tree-optimization/79390
17456 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17457 order does not result in usable sequence, retry with reversed operand
17463 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17464 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17465 op0 instead of fold_convert_loc (loc, type, arg0).
17467 2017-04-12 Jeff Law <law@redhat.com>
17469 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17470 has a delay slot in the generated code.
17472 * config/cris/cris.md (cris_preferred_reload_class): Return
17473 GENNONACR_REGS rather than GENERAL_REGS.
17475 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17478 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17479 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17480 signedness of the result type.
17482 2017-04-12 Richard Biener <rguenther@suse.de>
17483 Jeff Law <law@redhat.com>
17485 PR tree-optimization/80359
17486 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17487 trim stores to TARGET_MEM_REFs.
17489 2017-04-12 Richard Biener <rguenther@suse.de>
17491 PR tree-optimization/79390
17492 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17493 threading case even more.
17495 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17498 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17499 for quad_address_p for TImode, instead of just not indexed_address.
17501 2017-04-12 Richard Biener <rguenther@suse.de>
17502 Bernd Edlinger <bernd.edlinger@hotmail.de>
17504 PR middle-end/79671
17505 * alias.c (component_uses_parent_alias_set_from): Handle
17506 TYPE_TYPELESS_STORAGE.
17507 (get_alias_set): Likewise.
17508 * tree-core.h (tree_type_common): Add typeless_storage flag.
17509 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17510 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17511 for types containing members with TYPE_TYPELESS_STORAGE.
17512 (place_field): Likewise.
17513 (layout_type): Likewise for ARRAY_TYPE.
17514 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17515 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17516 TYPE_TYPELESS_STORAGE.
17517 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17519 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17522 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17523 first argument to type.
17525 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17529 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17530 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17531 (rs6000_expand_binop_builtin): Likewise.
17532 (rs6000_expand_ternop_builtin): Likewise; also add missing
17533 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17534 vshasigma built-ins.
17535 * doc/extend.texi: Document that vec_xxpermdi's third argument
17536 must be a constant.
17538 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17540 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17541 Use shift_const cost parameter when calculating gain of STV shifts.
17543 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17545 PR rtl-optimization/70478
17546 * lra-constraints.c (process_alt_operands): Check memory for
17547 disfavoring memory insn operand.
17549 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17551 PR middle-end/80100
17552 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17553 left shift in unsigned HOST_WIDE_INT type.
17555 PR rtl-optimization/80385
17556 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17557 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17560 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17561 if they have any depend clauses.
17563 2017-04-11 Martin Liska <mliska@suse.cz>
17566 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17567 * ipa-split.c (split_function): Create a local comdat symbol
17568 if caller is in a comdat group.
17570 2017-04-11 Martin Liska <mliska@suse.cz>
17573 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17576 2017-04-11 Martin Sebor <msebor@redhat.com>
17578 PR middle-end/80364
17579 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17580 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17582 (directive::set_width, directive::set_precision, format_character):
17584 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17587 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17590 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17591 conflict, set target->arch_name instead of target->cpu_name.
17593 2017-04-11 Richard Biener <rguenther@suse.de>
17595 PR tree-optimization/80374
17596 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17597 build_zero_cst, remove fold_convertible_p check again.
17599 2017-04-11 Martin Liska <mliska@suse.cz>
17602 * ubsan.c (instrument_object_size): Do not instrument register
17605 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17608 * config/i386/i386-builtin-types.def
17609 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17610 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17611 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17612 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17613 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17614 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17615 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17616 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17617 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17618 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17619 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17620 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17621 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17622 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17623 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17624 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17625 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17626 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17627 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17628 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17629 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17630 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17631 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17632 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17633 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17634 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17635 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17636 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17637 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17638 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17639 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17640 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17641 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17642 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17643 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17644 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17645 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17646 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17647 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17648 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17649 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17650 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17651 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17652 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17653 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17654 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17656 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17657 flag to second_arg_count, handle 4 argument function type _COUNT
17658 aliases, handle second_arg_count on second argument rather than last.
17660 2017-04-10 Jeff Law <law@redhat.com>
17662 PR tree-optimization/80374
17663 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17664 record anything if we can not convert integer_zero_node to the
17667 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17670 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17671 Enhance special handling given to the TARGET_P9_MINMAX option in
17672 relation to certain other options.
17674 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17676 PR tree-optimization/80153
17677 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17678 remove POINTER_PLUS_EXPR's base part directly, rather than through
17681 2017-04-10 Richard Biener <rguenther@suse.de>
17682 Bin Cheng <bin.cheng@arm.com>
17684 PR tree-optimization/80153
17685 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17686 the first element of pointer type aff_tree. Build result expr in
17688 (add_elt_to_tree): Convert to type unconditionally. Remove other
17689 fold_convert calls.
17690 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17691 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17693 2017-04-10 Richard Biener <rguenther@suse.de>
17695 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17698 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17700 PR rtl-optimization/70478
17701 * lra-constraints.c (curr_small_class_check): New.
17702 (update_and_check_small_class_inputs): New.
17703 (process_alt_operands): Update curr_small_class_check. Disfavor
17704 alternative insn memory operands. Check available regs for small
17707 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17710 * config/mips/mips.opt (-mvirt): Update description.
17711 * doc/invoke.texi (-mvirt): Likewise.
17713 2017-04-10 Richard Biener <rguenther@suse.de>
17715 PR middle-end/80362
17716 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17717 looking for NEGATE_EXPR in -A / -B to A / B folding.
17719 2017-04-10 Martin Liska <mliska@suse.cz>
17721 PR gcov-profile/80224
17722 * gcov.c (print_usage): Fix usage string.
17723 (get_gcov_intermediate_filename): Remove.
17724 (output_gcov_file): Use both for normal and intermediate format.
17725 (generate_results): Do not initialize special file for
17726 intermediate format.
17728 2017-04-10 Richard Biener <rguenther@suse.de>
17730 PR tree-optimization/80304
17731 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17734 2017-04-10 Nathan Sidwell <nathan@acm.org>
17737 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17738 (rs6000_init_builtins): Use it.
17740 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17742 * config/arm/arm.md (<mrc>): Add mode to SET source.
17743 (<mrrc>): Likewise.
17745 2017-04-10 Richard Biener <rguenther@suse.de>
17747 PR middle-end/80344
17748 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17750 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17753 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17754 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17755 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17756 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17757 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17758 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17759 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17760 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17761 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17762 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17763 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17764 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17765 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17766 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17767 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17768 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17769 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17770 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17771 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17772 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17773 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17774 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17775 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17777 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17779 PR rtl-optimization/70478
17780 * lra-constraints.c: Reverse the last patch.
17782 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17784 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17785 Add comment for WCHAR_T.
17787 2017-04-08 Martin Liska <mliska@suse.cz>
17790 2017-04-07 Martin Liska <mliska@suse.cz>
17793 * ipa-split.c (split_function): Add function part to a same comdat
17796 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17799 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
17801 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
17803 * rs6000/rs6000.c (vec_load_pendulum): Rename...
17804 (vec_pairing): ...to this.
17805 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
17806 (rs6000_sched_init): Adjust for name change.
17807 (struct rs6000_sched_context): Likewise.
17808 (rs6000_init_sched_context): Likewise.
17809 (rs6000_set_sched_context): Likewise.
17811 2017-04-07 Jakub Jelinek <jakub@redhat.com>
17817 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
17819 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
17820 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
17821 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
17823 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17825 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17827 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17829 PR rtl-optimization/70703
17830 * ira-color.c (update_conflict_hard_regno_costs): Use
17831 int64_t instead of HOST_WIDE_INT.
17833 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17835 PR rtl-optimization/70478
17836 * lra-constraints.c (process_alt_operands): Disfavor alternative
17837 insn memory operands.
17839 2017-04-07 Jeff Law <law@redhat.com>
17841 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17842 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17844 2017-04-07 Martin Liska <mliska@suse.cz>
17847 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17848 Show error message instead of an ICE.
17850 2017-04-07 Martin Liska <mliska@suse.cz>
17853 * ipa-split.c (split_function): Add function part to a same comdat
17856 2017-04-07 Richard Biener <rguenther@suse.de>
17858 PR middle-end/80341
17859 * tree.c (get_unwidened): Also handle ! for_type case for
17861 * convert.c (do_narrow): Split out from ...
17862 (convert_to_integer_1): ... here. Do not pass final truncation
17863 type to get_unwidened for TRUNC_DIV_EXPR.
17865 2017-04-07 Richard Biener <rguenther@suse.de>
17867 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17869 (aff_combination_const): Adjust.
17870 (aff_combination_scale): Likewise.
17871 (aff_combination_add_elt): Likewise.
17872 (aff_combination_add_cst): Likewise.
17873 (aff_combination_convert): Likewise.
17874 (add_elt_to_tree): Likewise. Remove unused argument.
17875 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17877 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17879 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17881 * config/arm/arm.c (arm_default_short_enums): Use
17882 ARM_DEFAULT_SHORT_ENUMS.
17883 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17885 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17888 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17889 members with redundant out-of-class redeclaration.
17891 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17894 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17895 * config/i386/i386.md (*zero_extendsidi2):
17896 Add (?*x,*x) and (?*v,*v) alternatives.
17898 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17901 * config/i386/i386.c (ix86_expand_builtin)
17902 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17903 mode from insn data. Convert operands to insn operand mode.
17904 Copy operands that don't satisfy insn predicate to a register.
17906 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17908 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17911 2017-04-06 Richard Biener <rguenther@suse.de>
17913 PR tree-optimization/80334
17914 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17915 preserve alignment of accesses.
17917 2017-04-06 Richard Biener <rguenther@suse.de>
17919 PR tree-optimization/80262
17920 * tree-sra.c (build_ref_for_offset): Preserve address-space
17922 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17923 Drop useless address-space information on MEM_REF offsets.
17925 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17927 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17929 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17931 PR rtl-optimization/70703
17932 * ira-color.c (update_conflict_hard_regno_costs): Use
17933 HOST_WIDE_INT instead of long.
17935 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17938 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17939 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17941 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17942 for x86_64 target. Handle -m3dnowa option.
17944 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17946 PR rtl-optimization/70703
17947 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17948 (update_conflict_hard_regno_costs): Use long instead of unsigned
17949 arithmetic for cost calculation.
17951 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17952 Bernd Edlinger <bernd.edlinger@hotmail.de>
17955 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17958 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17961 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17962 ptr_mode with Pmode throughout.
17963 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17964 into probe_stack_range and use DImode.
17966 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17969 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17970 call_eh_return is true.
17972 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17974 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17975 Initialize last_match_fntype_index.
17977 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17980 * tree-nvr.c: Include internal-fn.h.
17981 (pass_return_slot::execute): Ignore internal calls without
17984 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17985 Richard Biener <rguenther@suse.de>
17988 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17989 captures used multiple times, except for the last use.
17990 * generic-match-head.c: Include gimplify.h.
17992 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17994 PR tree-optimization/79390
17995 * target.h (struct noce_if_info): Declare.
17996 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17997 * target.def (noce_conversion_profitable_p): New target hook.
17998 * ifcvt.h (struct noce_if_info): New type, moved from ...
17999 * ifcvt.c (struct noce_if_info): ... here.
18000 (noce_conversion_profitable_p): Renamed to ...
18001 (default_noce_conversion_profitable_p): ... this. No longer
18003 (noce_try_store_flag_constants, noce_try_addcc,
18004 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18005 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18006 instead of noce_conversion_profitable_p.
18007 * config/i386/i386.c: Include ifcvt.h.
18008 (ix86_option_override_internal): Don't override
18009 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18010 (ix86_noce_conversion_profitable_p): New function.
18011 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18012 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18013 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18014 * doc/tm.texi: Regenerated.
18016 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18018 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18021 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18024 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18025 instructions for small multiply cores.
18027 2017-04-04 Jeff Law <law@redhat.com>
18029 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18031 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18032 that are not set by the loop over the elements.
18034 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18037 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18038 int mode, convert_modes it to mode as unsigned, otherwise use
18039 lowpart_subreg to mode rather than SImode.
18040 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18041 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18042 Use DImode instead of SImode for the shift count operand.
18043 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18046 2017-04-04 Richard Biener <rguenther@suse.de>
18048 PR middle-end/80281
18049 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18050 arithmetic done for the negate or the plus. Simplify.
18051 (A - (-B) -> A + B): Likewise.
18052 * fold-const.c (split_tree): Make sure to not negate pointers.
18054 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18056 PR rtl-optimization/60818
18057 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18058 a compare of comparisons with the thing compared if this results
18059 in a different machine mode.
18061 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18063 * alias.c (base_alias_check): Fix typo in comment.
18064 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18065 * cgraphunit.c (symbol_table::compile): Likewise.
18066 * collect2.c (maybe_run_lto_and_relink): Likewise.
18067 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18068 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18069 * config/avr/avr.c (avr_map_op_t): Likewise.
18070 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18071 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18072 * config/epiphany/epiphany.md (movcc): Likewise.
18073 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18074 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18076 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18077 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18078 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18079 * config/sh/sh.c (sh_rtx_costs): Likewise.
18080 * fold-const.c (fold_truth_andor): Likewise.
18081 * genautomata.c (collapse_flag): Likewise.
18082 * gengtype.h (struct type::u::s): Likewise.
18083 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18084 * input.c (FORMAT_AMOUNT): Likewise.
18085 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18086 (known_aggs_to_agg_replacement_list): Likewise.
18087 * ipa-inline-analysis.c: Likewise.
18088 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18089 * ipa-polymorphic-call.c
18090 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18091 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18092 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18094 * modulo-sched.c (apply_reg_moves): Likewise.
18095 * omp-expand.c (build_omp_regions_1): Likewise.
18096 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18097 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18098 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18099 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18100 * value-prof.c: Likewise.
18101 * var-tracking.c (val_reset): Likewise.
18103 2017-04-03 Richard Biener <rguenther@suse.de>
18105 PR tree-optimization/80275
18106 * fold-const.c (split_address_to_core_and_offset): Handle
18109 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18111 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18112 descriptors is at least equal to that of functions.
18114 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18116 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18118 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18121 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18122 (mov<IMOD4:mode>): New expander.
18123 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18125 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18127 PR rtl-optimization/79405
18128 * fwprop.c (propagations_left): New variable.
18129 (forward_propagate_into): Decrement it.
18130 (fwprop_init): Initialize it.
18131 (fw_prop): If the variable has reached zero, stop propagating.
18132 (fwprop_addr): Ditto.
18134 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18137 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18138 a FUNCTION_DECL, pass it as decl instead of origin to
18141 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18143 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18146 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18149 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18150 TARGET_VSX_SMALL_INTEGER.
18152 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18154 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18155 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18157 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18159 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18160 extraction from odd-numbered MSA register.
18162 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18164 PR middle-end/80173
18165 * expmed.c (store_bit_field_1): Don't attempt to create
18166 a word subreg out of hard registers wider than word if they
18167 have HARD_REGNO_NREGS of 1 for their mode.
18169 PR middle-end/80163
18170 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18171 conversions to integer types wider than word and pointer.
18174 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18175 (rtx_equal_for_cselib_p): Pass 0 to it.
18176 * cselib.c (cselib_hasher::equal): Likewise.
18177 (rtx_equal_for_cselib_1): Add depth argument. If depth
18178 is 128, don't look up VALUE locs and punt. Increment
18179 depth in recursive calls when walking VALUE locs.
18181 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18183 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18184 (make_gcov_file_name): Use the canonical path name for generating
18186 (read_line): Fix handling of files with ascii null bytes.
18188 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18190 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18191 to initialise a vector register instead
18192 of using a const_int.
18194 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18196 PR translation/80189
18197 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18198 diagnostic messages.
18200 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18203 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18204 (dfp_diex_<mode>): Update mode of operand 1.
18205 * doc/extend.texi (dxex, dxexq): Document change to return type.
18206 (diex, diexq): Document change to argument type.
18208 2017-03-30 Martin Jambor <mjambor@suse.cz>
18211 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18212 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18213 it reflects the signature changes performed at the callee side.
18214 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18215 to cgraph_build_function_type_skip_args.
18216 (build_function_decl_skip_args): Adjust call to the above function.
18218 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18221 * config/i386/sse.md
18222 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18223 register as dest whenever it is a MEM not rtx_equal_p to the
18224 corresponding dup operand, and when forcing into reg move the
18225 reg into the memory afterwards.
18226 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18227 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18228 for the force_reg mode.
18229 (avx512vl_vextractf128<mode>): Use register as dest either
18230 always when a MEM, or when it is a MEM not rtx_equal_p to the
18231 corresponding dup operand, or even not when it is a CONST_VECTOR
18232 depending on the mode and lo vs. hi.
18233 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18235 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18236 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18237 Likewise. Require that operands[2] is even.
18238 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18239 Remove extraneous parens. Require that operands[2] is a multiple
18241 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18242 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18243 disallow memory then.
18245 2017-03-30 Richard Biener <rguenther@suse.de>
18247 PR tree-optimization/77498
18248 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18249 to non-constants over backedges.
18251 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18253 PR rtl-optimization/80233
18254 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18255 as last_combined_insn. Do not test for BARRIER_P separately.
18257 2017-03-29 Andreas Schwab <schwab@suse.de>
18260 * calls.c (prepare_call_address): Convert funexp to Pmode before
18261 copying to temp reg.
18263 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18265 PR tree-optimization/80158
18266 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18267 Handle possible future case of more than one alternate
18269 (replace_rhs_if_not_dup): Likewise.
18270 (replace_one_candidate): Likewise.
18272 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18274 PR rtl-optimization/80193
18275 * ira.c (ira): Do not check allocation for LRA.
18277 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18279 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18280 (nvptx_output_simt_exit): Declare.
18281 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18282 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18283 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18284 (nvptx_declare_function_name): ...here. Emit declaration of local
18285 memory space buffer for omp_simt_enter insn.
18286 (nvptx_output_unisimt_switch): New.
18287 (nvptx_output_softstack_switch): New.
18288 (nvptx_output_simt_enter): New.
18289 (nvptx_output_simt_exit): New.
18290 * config/nvptx/nvptx.h (struct machine_function): New fields
18291 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18292 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18293 (UNSPECV_SIMT_EXIT): Ditto.
18294 (omp_simt_enter_insn): New insn.
18295 (omp_simt_enter): New expansion.
18296 (omp_simt_exit): New insn.
18297 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18299 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18300 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18301 (expand_GOMP_SIMT_EXIT): New.
18302 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18303 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18304 (GOMP_SIMT_EXIT): Ditto.
18305 * target-insns.def (omp_simt_enter): New insn.
18306 (omp_simt_exit): Ditto.
18307 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18309 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18310 (lower_rec_input_clauses): Likewise.
18311 (lower_lastprivate_clauses): Handle SIMT privatization.
18313 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18314 (ompdevlow_adjust_simt_enter): New.
18315 (find_simtpriv_var_op): New.
18316 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18317 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18319 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18320 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18321 (copy_decl_for_dup_finish): Ditto.
18323 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18325 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18328 * config/i386/i386.c (ix86_option_override_internal): Always
18329 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18331 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18333 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18335 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18337 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18338 mark new edge's irreducible flag accordign to it.
18339 (vect_do_peeling): Check loop preheader edge's irreducible flag
18340 and pass it to function slpeel_add_loop_guard.
18342 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18344 PR tree-optimization/80218
18345 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18346 Update block frequencies and counts.
18348 2017-03-28 Richard Biener <rguenther@suse.de>
18350 PR tree-optimization/78644
18351 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18352 of a simplification result we may not use it at all.
18354 2017-03-28 Richard Biener <rguenther@suse.de>
18357 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18358 without arguments, generate default definition of a SSA name.
18360 2017-03-28 Richard Biener <rguenther@suse.de>
18362 PR middle-end/80222
18363 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18364 TYPE_REF_CAN_ALIAS_ALL references.
18365 * fold-const.c (fold_indirect_ref_1): Likewise.
18367 2017-03-28 Martin Liska <mliska@suse.cz>
18370 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18371 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18373 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18374 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18376 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18377 (EXTRA_SPECS): Define.
18378 (SUBTARGET_EXTRA_SPECS): Likewise.
18379 (SUBTARGET_CPP_SPEC): Likewise.
18380 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18381 SUBTARGET_EXTRA_SPECS.
18382 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18384 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18386 * config/arc/simdext.md (vst64_insn): Update pattern.
18387 (vld32wh_insn): Likewise.
18388 (vld32wl_insn): Likewise.
18389 (vld64_insn): Likewise.
18390 (vld32_insn): Likewise.
18392 2017-03-28 Marek Polacek <polacek@redhat.com>
18395 * fold-const.c (fold_comparison): Use protected_set_expr_location
18396 instead of SET_EXPR_LOCATION.
18398 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18400 * tree.c (add_expr): Avoid name lookup warning.
18402 2017-03-27 Jeff Law <law@redhat.com>
18404 PR tree-optimization/80216
18405 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18406 function name. Limit recursion depth.
18407 (record_temporary_equivalences): Corresponding changes.
18409 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18411 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18414 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18417 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18419 * cfgcleanup.c (reg_note_cfa_p): New array.
18420 (insns_have_identical_cfa_notes): New function.
18421 (old_insns_match_p): Don't cross-jump in between /f
18422 and non-/f instructions. If both i1 and i2 are frame related,
18423 verify all CFA notes, their order and content.
18425 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18428 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18429 HImode and SImode with zero extend to DImode to one insn.
18430 (bswap<mode>2_extenddi): Likewise.
18431 (bswapsi2_extenddi): Likewise.
18432 (bswaphi2_extendsi): Likewise.
18433 (bswaphi2): Combine bswap HImode and SImode into one insn.
18434 Separate memory insns from swapping register.
18435 (bswapsi2): Likewise.
18436 (bswap<mode>2): Likewise.
18437 (bswaphi2_internal): Delete, no longer used.
18438 (bswapsi2_internal): Likewise.
18439 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18440 store, and gpr<-gpr swap insns.
18441 (bswap<mode>2_store): Likewise.
18442 (bswaphi2_reg): Register only splitter, combine with the splitter.
18443 (bswaphi2 splitter): Likewise.
18444 (bswapsi2_reg): Likewise.
18445 (bswapsi2 splitter): Likewise.
18446 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18447 the insns into load, store, and register/register insns.
18448 (bswapdi2_ldbrx): Likewise.
18449 (bswapdi2_load): Likewise.
18450 (bswapdi2_store): Likewise.
18451 (bswapdi2_reg): Likewise.
18453 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18455 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18456 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18458 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18461 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18464 special handling for target option conflicts between dform
18465 options (-mpower9-dform, -mpower9-dform-vector,
18466 -mpower9-dform-scalar) and -mno-direct-move.
18468 2017-03-27 Richard Biener <rguenther@suse.de>
18470 PR tree-optimization/80181
18471 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18473 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18475 * config/arc/predicates.md (move_double_src_operand): Replace the
18476 call to move_double_src_operand with a call to address_operand.
18478 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18480 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18481 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18482 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18484 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18486 * config/arc/predicates.md (long_immediate_loadstore_operand):
18487 Consider scaled addresses cases.
18489 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18491 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18492 restored when in interrupt.
18493 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18494 doesn't have delay slot.
18496 2017-03-27 Richard Biener <rguenther@suse.de>
18499 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18500 inlined thunk clones.
18502 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18505 * asan.c (instrument_derefs): Copy over last operand from
18506 original COMPONENT_REF to the new COMPONENT_REF with
18507 DECL_BIT_FIELD_REPRESENTATIVE.
18508 * ubsan.c (instrument_object_size): Likewise.
18510 2017-03-27 Richard Biener <rguenther@suse.de>
18512 PR tree-optimization/80170
18513 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18514 sure DR/SCEV didnt fold in constants we do not see when looking
18515 at the reference base alignment.
18517 2017-03-27 Richard Biener <rguenther@suse.de>
18519 PR middle-end/80171
18520 * gimple-fold.c (fold_ctor_reference): Properly guard against
18521 NULL return value from canonicalize_constructor_val.
18523 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18526 * config/i386/i386.c (ix86_expand_builtin)
18527 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18528 flags reg setting and flags reg using instructions.
18529 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18530 clobbering instructions to zero extend op2.
18532 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18534 * doc/install.texi (Configuration) <--with-aix-soname>:
18535 Update link to AIX ld.
18537 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18539 PR rtl-optimization/80160
18540 PR rtl-optimization/80159
18541 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18542 reg_alternate_class into account.
18544 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18547 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18548 to consider in curr_insn_transform.
18550 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18552 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18553 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18554 and emit_mode_inner.
18556 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18558 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18559 argument to the overloaded builtin variants. Use the new flag to
18560 deprecate certain builtin variants.
18561 * config/s390/s390-builtin-types.def: Add new builtin types.
18562 * config/s390/s390-builtins.h: Support new flags field for
18563 overloaded builtins.
18564 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18565 (s390_macro_to_expand): Enable vector float data type.
18566 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18567 builtins by incrementing the __VEC__ version number.
18568 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18570 (s390_resolve_overloaded_builtin): Emit error messages depending
18571 on the builtin flags.
18572 * config/s390/s390.c (s390_expand_builtin): Support additional
18573 flags argument. Change error message to match the messages
18574 emitted in s390-c.c.
18575 * config/s390/s390.md: New UNSPEC_* constants.
18576 (op_type): Add new instruction types.
18577 * config/s390/vecintrin.h: Add new builtins and test data class
18579 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18580 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18581 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18582 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18583 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18584 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18586 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18587 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18588 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18589 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18591 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18592 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18593 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18594 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18596 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18597 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18598 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18599 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18600 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18601 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18602 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18604 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18605 ("vec_scatter_element<V_HW_4:mode>_DI")
18606 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18607 ("vec_fpint<mode>", "vflls")
18608 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18609 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18610 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18611 ("*vec_cmphe<mode>_cc"): ... these.
18613 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18614 mode constant instead of magic value.
18616 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18618 * config/s390/s390.c (s390_expand_vec_compare): Support other
18619 vector floating point modes than just V2DF.
18620 (s390_expand_vcond): Likewise.
18621 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18622 (s390_cannot_change_mode_class): Prevent mode changes between TF
18623 and V1TF in vector registers.
18624 * config/s390/s390.md (DF, SF): New mode attributes.
18625 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18626 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18627 SFmode support for VRs.
18628 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18630 (VFT, VF_HW): New mode iterators.
18631 (vw, sdx): New mode attributes.
18632 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18633 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18634 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18635 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18636 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18637 also the new vector floating point modes. Renaming to ...
18639 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18640 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18641 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18642 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18643 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18644 ("vec_unordered<mode>"): ... these.
18646 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18647 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18648 ("*vec_extendv2df"): New insn definitions.
18650 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18652 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18653 ("mulditi3_2", "*muldi3_sign"): New patterns.
18654 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18655 rename the pattern definition.
18657 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18659 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18661 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18663 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18665 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18666 instruction if possible.
18667 * config/s390/vector.md (vec_halfnumelts): New mode
18669 ("*vec_vllezlf<mode>"): New pattern.
18671 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18673 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18674 ("popcountv4si2", "popcountv2di2"): Rename to ...
18675 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18676 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18678 ("popcount<mode>2_vxe"): New pattern.
18680 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18682 * common/config/s390/s390-common.c (processor_flags_table): Add
18684 * config.gcc: Add arch12.
18685 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18686 Default to arch12 for unknown CPU model numbers.
18687 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18688 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18689 PROCESSOR_max sanity check.
18690 * config/s390/s390-opts.h (enum processor_type): Add
18692 * config/s390/s390.c (processor_table): Add arch12.
18693 (s390_expand_builtin): Add check for B_VXE flag.
18694 (s390_issue_rate): Add PROCESSOR_ARCH12.
18695 (s390_get_sched_attrmask): Likewise.
18696 (s390_get_unit_mask): Likewise.
18697 (s390_sched_score): Enable z13 scheduling for arch12.
18698 (s390_sched_reorder): Likewise.
18699 (s390_sched_variable_issue): Likewise.
18700 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18702 (s390_tune_attr): Use z13 scheduling also for arch12.
18703 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18704 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18705 (TARGET_VXE_P): New macros.
18706 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18707 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18708 * config/s390/s390.opt: Add arch12 as processor_type.
18710 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18712 * config/s390/s390.md
18713 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18714 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18715 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18717 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18718 Rename expanders to ...
18720 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18721 ("fixuns_truncdddi2_emu"): ... these.
18723 ("fixuns_trunc<mode>si2_emu"): New expander.
18725 ("*fixuns_truncdfdi2_z13"): Rename to ...
18726 ("*fixuns_truncdfdi2_vx"): ... this.
18728 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18730 * config/s390/2964.md: Remove the single element vector compare
18731 instructions which are no longer used.
18732 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18734 (s390_canonicalize_comparison): Remove handling of DFmode
18736 (s390_expand_vec_compare_scalar): Remove function.
18737 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18738 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18740 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18742 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18744 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18746 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18747 will anyway by matched by mov<mode>_64dfp.
18749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18751 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18752 vlef/vstef. Add missing operand to vleif.
18754 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18756 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18757 pair for all vector types with 64 bit elements.
18758 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18759 * config/s390/vector.md (V_HW_64): ... here.
18760 (V_128_NOSINGLE): New mode iterator.
18761 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18762 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18763 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18764 ("*vec_load_pairv2di"): Change to ...
18765 ("*vec_load_pair<mode>"): ... this one.
18767 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18769 * config/s390/constraints.md: Add comments.
18770 (jKK): Reject element sizes > 8 bytes.
18771 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18773 * config/s390/s390.md: Add the s_operand checks formerly in
18774 s390_split_ok_p to various splitters where they are still
18776 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18777 for 128 bit vectors. Plus two splitters.
18779 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18781 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18784 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18787 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18788 error if the boundary argument is not constant.
18790 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18792 PR rtl-optimization/80112
18793 * loop-doloop.c (doloop_condition_get): Don't check condition
18794 if cmp isn't SET with IF_THEN_ELSE src.
18796 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18798 PR tree-optimization/80158
18799 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
18800 replacing a candidate statement, also replace it for the
18801 candidate's alternate interpretation.
18802 (replace_rhs_if_not_dup): Likewise.
18803 (replace_one_candidate): Likewise.
18805 2017-03-24 Richard Biener <rguenther@suse.de>
18807 PR tree-optimization/80167
18808 * graphite-isl-ast-to-gimple.c
18809 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
18811 (translate_isl_ast_to_gimple::get_rename): Likewise.
18813 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
18815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
18816 handling of certain combinations of target options, including the
18817 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
18818 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
18820 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18823 * config/arm/arm.md (*load_multiple): Add reload_completed to
18824 matching condition.
18826 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18827 Richard Biener <rguenth@suse.de>
18829 PR tree-optimization/79908
18830 PR tree-optimization/80136
18831 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18832 been cast away, gimplify_and_add suffices.
18834 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18836 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18838 2017-03-23 Richard Biener <rguenther@suse.de>
18840 PR tree-optimization/80032
18841 * gimplify.c (gimple_push_cleanup): Forced unconditional
18842 cleanups still have to go to the conditional_cleanups
18845 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18847 PR tree-optimization/80072
18848 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18850 (next_operand_entry_id): Change type to unsigned int.
18851 (sort_by_operand_rank): Make sure to return the right return value
18852 even if unsigned fields are bigger than INT_MAX.
18853 (struct oecount): Change cnt and id type to unsigned int.
18854 (oecount_hasher::equal): Formatting fix.
18855 (oecount_cmp): Make sure to return the right return value
18856 even if unsigned fields are bigger than INT_MAX.
18857 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18860 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18861 TREE_READONLY on result if writing it more than once.
18864 * doc/invoke.texi (-fsanitize=thread): Document that with
18865 -fnon-call-exceptions atomics are not able to throw
18869 * tsan.c: Include tree-eh.h.
18870 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18871 maybe_clean_or_replace_eh_stmt where needed.
18872 (instrument_memory_accesses): Add cfg_changed argument.
18873 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18874 if it returned true.
18875 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18877 PR rtl-optimization/63191
18878 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18879 wrapper function, moved the whole old content into ...
18880 (ix86_delegitimize_address_1): ... this. New inline function.
18881 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18882 true as last argument instead of ix86_delegitimize_address.
18884 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18886 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18887 cortexa57_branch_cost.
18889 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18891 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18893 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18896 * doc/md.texi (Constraints): Document wA constraint.
18897 * config/rs6000/constraints.md (wA): New.
18898 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18899 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18900 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18901 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18903 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18906 * gimplify.c (is_oacc_declared): New function.
18907 (oacc_default_clause): Use it to set default flags for acc declared
18908 variables inside parallel regions.
18909 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18910 declared variables.
18911 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18912 declare attribute to any decl as necessary.
18914 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18917 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18918 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18919 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18920 (arm_arch_lpae): This.
18921 * config/arm/arm.c (arm_arch7ve): Rename into ...
18922 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18923 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18926 2017-03-22 Martin Liska <mliska@suse.cz>
18929 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18930 error message instead of an ICE.
18932 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18934 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18936 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18938 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18943 2017-03-21 Martin Sebor <msebor@redhat.com>
18945 * doc/extend.texi: Use "cannot" instead of "can't."
18946 * doc/hostconfig.texi: Same.
18947 * doc/install.texi: Same.
18948 * doc/invoke.texi: Same.
18949 * doc/loop.texi: Same.
18950 * doc/md.texi: Same.
18951 * doc/objc.texi: Same.
18952 * doc/rtl.texi: Same.
18953 * doc/tm.texi: Same.
18954 * doc/tm.texi.in: Same.
18955 * doc/trouble.texi: Same.
18957 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18960 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18961 (collect_checksum_attributes): Set it.
18962 (die_checksum_ordered): Use it.
18964 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18966 PR tree-optimization/79908
18967 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18968 change: For a VA_ARG whose LHS has been cast away, use
18969 force_gimple_operand to construct the side effects.
18971 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18973 PR translation/80001
18974 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18975 more amenable to translation.
18976 (oacc_loop_auto_partitions): Likewise.
18978 2017-03-21 Marek Polacek <polacek@redhat.com>
18979 Martin Sebor <msebor@redhat.com>
18981 PR tree-optimization/80109
18982 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18983 on INTEGRAL_TYPE_P.
18985 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18986 Segher Boessenkool <segher@kernel.crashing.org>
18989 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18990 check reg_used_between_p between insn and one of succ or succ2
18991 depending on if succ is artificial insn not inserted into insn
18994 2017-03-21 Martin Liska <mliska@suse.cz>
18996 PR gcov-profile/80081
18997 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18998 * doc/gcc.texi: Include gcov-dump stuff.
18999 * doc/gcov-dump.texi: New file.
19001 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19003 PR rtl-optimization/79150
19004 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19005 conditional jump, if the jump is the last insn of the loop.
19007 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19008 Richard Biener <rguenth@suse.de>
19010 PR tree-optimization/79908
19011 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19012 been cast away, use force_gimple_operand to construct the side
19015 2017-03-21 Martin Liska <mliska@suse.cz>
19017 PR libfortran/79956
19018 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19021 2017-03-21 Brad Spengler <spender@grsecurity.net>
19024 * plugin.c (htab_hash_plugin): New function.
19025 (add_new_plugin): Use it and adjust.
19026 (parse_plugin_arg_opt): Adjust.
19027 (init_one_plugin): Likewise.
19029 2017-03-21 Richard Biener <rguenther@suse.de>
19031 PR tree-optimization/80032
19032 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19033 if set force the cleanup to happen unconditionally.
19034 (gimplify_target_expr): Push inserted clobbers with force_uncond
19035 to avoid them being removed by control-dependent DCE.
19037 2017-03-21 Richard Biener <rguenther@suse.de>
19039 PR tree-optimization/80122
19040 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19041 va_arg_pack_len when the inlined call stmt requires pack
19043 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19045 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19048 * tsan.c (instrument_builtin_call): If the memory model argument
19049 is not a constant, assume it is valid.
19052 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19055 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19057 PR rtl-optimization/79910
19058 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19059 if its dest is used by an insn before I2 (other than the combined
19060 insns themselves, which are properly handled already).
19062 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19065 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19067 * combine.c (record_used_regs): New static function.
19068 (try_combine): Handle situations where there is an additional
19069 instruction between I2 and I3 which needs to have a LOG_LINK
19073 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19075 * combine.c (try_combine): Delete redundant i1 test. Call
19076 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19078 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19081 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19082 alternatives 13/14.
19084 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19086 PR tree-optimization/80054
19087 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19088 the optimization if a PHI or any of its arguments is not dominated
19089 by the candidate's basis. Use gphi* rather than gimple* as
19091 (replace_profitable_candidates): Clean up a gimple* variable that
19092 should be a gphi* variable.
19094 2017-03-20 Martin Sebor <msebor@redhat.com>
19097 * doc/extend.texi (attribute constructor): Document present limitation.
19099 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19102 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19103 __POWER9_VECTOR__ #ifdef control, change template definition to
19104 use Power9-specific built-in function.
19105 (vec_any_eq): Likewise.
19106 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19107 to control outcomes from this test.
19108 (vector_ae_<mode>p): For VEC_F modes, likewise.
19110 2017-03-20 Ian Lance Taylor <iant@google.com>
19112 * config/i386/i386.c (ix86_function_regparm): Save an extra
19113 register for -fsplit-stack with DECL_STATIC_CHAIN.
19115 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19118 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19119 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19121 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19123 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19125 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19128 2017-03-20 Marek Polacek <polacek@redhat.com>
19131 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19133 2017-03-20 Richard Biener <rguenther@suse.de>
19135 PR tree-optimization/80113
19136 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19137 allocate extra SSA name for PHI def.
19138 (add_close_phis_to_outer_loops): Likewise.
19139 (add_close_phis_to_merge_points): Likewise.
19140 (copy_loop_close_phi_args): Likewise.
19141 (copy_cond_phi_nodes): Likewise.
19143 2017-03-20 Martin Liska <mliska@suse.cz>
19145 PR middle-end/79753
19146 * tree-chkp.c (chkp_build_returned_bound): Do not build
19147 returned bounds for a LHS that's not a BOUNDED_P type.
19149 2017-03-20 Martin Liska <mliska@suse.cz>
19153 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19154 COMPLEX_CST and VECTOR_CST.
19156 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19159 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19160 target operand. A new splitter adds the clobber statement in case
19161 the target operand is dead anyway.
19163 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19165 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19166 to age-old versions of binutils and glibc.
19168 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19170 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19172 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19174 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19176 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19178 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19179 requirement for binutils 2.13.
19181 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19183 * combine.c (try_combine): Delete redundant i1 test. Call
19184 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19186 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19188 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19189 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19191 <riscv64-*-elf>: Re-arrange section
19192 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19193 <riscv32-*-linux>: Likewise.
19194 <riscv64-*-elf>: Likewise
19195 <riscv64-*-linux>: Likewise.
19197 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19200 * aarch64.opt(verbose-cost-dump): Fix typo.
19202 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19205 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19206 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19208 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19210 * reload.c (find_reloads): When reloading a nonoffsettable address,
19211 use RELOAD_OTHER for it and its address reloads.
19213 PR rtl-optimization/79910
19214 * combine.c (record_used_regs): New static function.
19215 (try_combine): Handle situations where there is an additional
19216 instruction between I2 and I3 which needs to have a LOG_LINK
19219 2017-03-17 Jeff Law <law@redhat.com>
19221 PR tree-optimization/71437
19222 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19223 conditional in the hash table first.
19224 (vrp_dom_walker::before_dom_children): Extract condition from
19225 ASSERT_EXPR. Record condition, its inverion and any implied
19226 conditions as well.
19228 2017-03-17 Marek Polacek <polacek@redhat.com>
19229 Markus Trippelsdorf <markus@trippelsdorf.de>
19231 PR tree-optimization/80079
19232 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19235 2017-03-17 Richard Biener <rguenther@suse.de>
19237 PR middle-end/80075
19238 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19239 Properly verify the LHS before the RHS possibly claims to be
19241 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19244 2017-03-17 Martin Jambor <mjambor@suse.cz>
19246 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19247 (List of -O2 options): Likewise.
19248 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19251 2017-03-17 Tom de Vries <tom@codesourcery.com>
19253 * gcov-dump.c (print_usage): Print bug_report_url.
19255 2017-03-17 Richard Biener <rguenther@suse.de>
19257 PR middle-end/80050
19258 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19259 (parser::peek): Likewise.
19261 2017-03-17 Richard Biener <rguenther@suse.de>
19263 PR tree-optimization/80048
19264 * sese.c (free_sese_info): Properly release rename_map and
19265 copied_bb_map elements.
19267 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19269 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19270 Add linked-list forward and backlinks. Insert on
19271 construction, remove on destruction.
19272 (class pass_store_merging): Add m_stores_head field.
19273 (pass_store_merging::terminate_and_process_all_chains):
19274 Iterate over m_stores_head list.
19275 (pass_store_merging::terminate_all_aliasing_chains):
19277 (pass_store_merging::execute): Check for debug stmts first.
19278 Push new chains onto the m_stores_head stack.
19280 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19283 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19284 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19285 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19287 2017-03-16 Jeff Law <law@redhat.com>
19289 PR tree-optimization/71437
19290 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19291 member function. Implementation moved into after_dom_children
19292 member function and into the threader's thread_outgoing_edges
19294 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19295 some code into new thread_outgoing_edges.
19296 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19297 definition. Simplify marker handling (do it here). Assume we always
19298 have the available expression and the const/copies tables.
19299 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19301 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19302 * tree-vrp.c (equiv_stack): No longer file scoped.
19303 (vrp_dom_walker): New class.
19304 (vrp_dom_walker::before_dom_children): New member function.
19305 (vrp_dom_walker::after_dom_children): Likewise.
19306 (identify_jump_threads): Setup domwalker. Use it rather than
19307 walking edges in a random order by hand. Simplify setup/finalization.
19308 (finalize_jump_threads): Remove.
19309 (vrp_finalize): Do not call identify_jump_threads here.
19310 (execute_vrp): Do it here instead and call thread_through_all_blocks
19313 PR tree-optimization/71437
19314 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19316 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19318 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19319 (dom_opt_dom_walker::thread_across_edge): Remove
19320 handle_dominating_asserts argument. All callers changed.
19321 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19322 changes. Remove calls to lhs_of_dominating_assert. Other
19323 uses of handle_dominating_asserts turn into unconditional code
19324 (simplify_control_stmt_condition_1): Likewise.
19325 (simplify_control_stmt_condition): Likewise.
19326 (thread_through_normal_block, thread_across_edge): Likewise.
19327 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19328 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19329 object if it is not an SSA_NAME.
19330 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19331 before calling into the VRP specific simplifiers.
19332 (identify_jump_threads): Remove handle_dominating_asserts
19335 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19338 * tree-diagnostic.c (default_tree_printer): No longer static.
19339 * tree-diagnostic.h (default_tree_printer): New prototype.
19341 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19343 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19344 Change ins into fmov.
19346 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19348 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19349 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19350 Use h_con constraint for operand 1.
19351 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19352 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19354 2017-03-15 Jeff Law <law@redhat.com>
19356 PR tree-optimization/71437
19357 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19358 (record_temporary_equivalences): Use it.
19360 PR tree-optimization/71437
19361 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19362 tree-ssa-scopedtables.
19363 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19364 (record_conditions, record_cond, vuse_eq): Likewise.
19365 (record_edge_info): Adjust to API tweak of record_conditions.
19366 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19367 (record_temporary_equivalences, optimize_stmt): Likewise.
19368 (eliminate_redundant_computations): Likewise.
19369 (record_equivalences_from_stmt): Likewise.
19370 * tree-ssa-scopedtables.c: Include options.h and params.h.
19371 (vuse_eq): New function, moved from tree-ssa-dom.c
19372 (build_and_record_new_cond): Likewise.
19373 (record_conditions): Likewise. Accept vector of conditions rather
19374 than edge_equivalence structure for first argument.
19375 for the first argument.
19376 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19377 from tree-ssa-dom.c.
19378 (avail_exprs_stack::record_cond): Likewise.
19379 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19380 from tree-ssa-dom.c.
19381 (avail_exprs_stack): Add new member functions lookup_avail_expr
19383 (record_conditions): Declare.
19385 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19388 * lra-constraints.c (process_alt_operands): Increase reject for
19389 reloading an input/output operand.
19391 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19394 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19395 insns to convert from signed/unsigned char/short to IEEE 128-bit
19397 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19399 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19402 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19403 subreg of inner mode for values already in registers.
19405 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19407 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19408 iteration reg is used after the loop.
19410 2017-03-14 Martin Sebor <msebor@redhat.com>
19412 PR tree-optimization/79800
19413 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19414 precision in negative-positive range.
19415 (format_floating): Call non-const overload with adjusted precision.
19417 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19420 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19423 2017-03-14 Martin Sebor <msebor@redhat.com>
19425 PR middle-end/80020
19426 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19427 * builtins.def (aligned_alloc): Use it.
19430 * Makefile.in (GTFILES): Add calls.c.
19431 * calls.c: Include "gt-calls.h".
19433 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19435 PR rtl-optimization/79728
19436 * regs.h (struct target_regs): New field
19437 x_contains_allocatable_regs_of_mode.
19438 (contains_allocatable_regs_of_mode): New macro.
19439 * reginfo.c (init_reg_sets_1): Initialize it, and change
19440 contains_reg_of_mode so it includes global regs as well.
19441 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19442 rather than contains_regs_of_mode.
19444 2017-03-14 Martin Liska <mliska@suse.cz>
19446 * doc/invoke.texi: Document options that can't be combined with
19447 -fcheck-pointer-bounds.
19449 2017-03-14 Martin Liska <mliska@suse.cz>
19451 PR middle-end/79831
19452 * doc/invoke.texi (-Wchkp): Document the option.
19454 2017-03-14 Martin Liska <mliska@suse.cz>
19456 * Makefile.in: Install gcov-dump.
19458 2017-03-14 Martin Liska <mliska@suse.cz>
19460 * multiple_target.c (expand_target_clones): Bail out for
19461 an invalid attribute.
19463 2017-03-14 Richard Biener <rguenther@suse.de>
19465 * alias.c (struct alias_set_entry): Pack properly.
19466 * cfgloop.h (struct loop): Likewise.
19467 * cse.c (struct set): Likewise.
19468 * ipa-utils.c (struct searchc_env): Likewise.
19469 * loop-invariant.c (struct invariant): Likewise.
19470 * lra-remat.c (struct cand): Likewise.
19471 * recog.c (struct change_t): Likewise.
19472 * rtl.h (struct address_info): Likewise.
19473 * symbol-summary.h (function_summary): Likewise.
19474 * tree-loop-distribution.c (struct partition): Likewise.
19475 * tree-object-size.c (struct object_size_info): Likewise.
19476 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19477 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19478 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19479 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19480 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19481 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19482 (struct _stmt_vec_info): Likewise.
19484 2017-03-14 Martin Liska <mliska@suse.cz>
19487 * multiple_target.c (create_dispatcher_calls): Check that
19488 a target can create a function dispatcher.
19490 2017-03-14 Martin Liska <mliska@suse.cz>
19493 * multiple_target.c (expand_target_clones): Drop local.local
19494 flag for default implementation.
19496 2017-03-14 Richard Biener <rguenther@suse.de>
19498 PR tree-optimization/80030
19499 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19501 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19503 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19504 gcc_fallthrough() instead of __attribute__((fallthrough));
19506 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19508 * doc/gcc.texi: Remove "up" link to (DIR).
19509 * doc/gccint.texi: Ditto.
19511 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19513 * doc/install.texi (Specific) <avr>: Remove reference to
19516 2017-03-13 Jeff Law <law@redhat.com>
19518 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19519 attribute rather than comments.
19521 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19522 match_scratch operand is highest.
19524 2017-03-13 Martin Liska <mliska@suse.cz>
19526 PR middle-end/78339
19527 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19528 is a CHKP clone, use original declaration.
19530 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19532 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19533 (arc_conditional_register_usage): Use a different allocation order
19534 when optimizing for size.
19535 * common/config/arc/arc-common.c (arc_option_optimization_table):
19536 Section anchors default on when optimizing for size.
19538 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19540 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19542 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19544 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19545 * config/arc/arc.md (cpu_facility): Add cd variant.
19546 (*movqi_insn): Add code density variant.
19547 (*movhi_insn): Likewise.
19548 (*movqi_insn): Likewise.
19549 (*addsi3_mixed): Likewise.
19550 (subsi3_insn): Likewise.
19552 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19554 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19556 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19558 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19559 expressions with MINUS and UNARY ops.
19561 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19564 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19566 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19567 between vec_select and vector argument.
19568 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19569 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19570 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19571 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19572 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19573 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19575 2017-03-13 Richard Biener <rguenther@suse.de>
19578 * params.def (vect-max-peeling-for-alignment): Fix typo.
19580 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19582 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19583 issue that only occurred with binutils below 2.18.
19585 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19587 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19588 refer to binutils 2.11/2.12 minimum.
19590 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19592 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19593 ftp.kernel.org and simplify binutils requirement.
19595 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19597 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19599 (Optimize Options): Ditto. Also remove redundancy.
19601 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19603 PR translation/79848
19604 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19606 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19607 to G_ to avoid double translation.
19609 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19611 PR translation/79923
19612 * auto-profile.c (get_combined_location): Convert leading
19613 character of diagnostics to lower case and remove trailing period.
19614 (read_profile): Likewise for various diagnostics.
19615 * config/arm/arm.c (arm_option_override): Remove trailing period
19616 from various diagnostics.
19617 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19618 (msp430_expand_delay_cycles): Likewise.
19620 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19623 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19624 full command-line argument, rather than just "str".
19625 (aarch64_validate_march): Likewise.
19626 (aarch64_validate_mtune): Likewise.
19628 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19630 PR rtl-optimization/78911
19631 * lra-assigns.c (must_not_spill_p): New function.
19632 (spill_for): Use it.
19634 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19636 PR tree-optimization/79981
19637 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19638 ATOMIC_COMPARE_EXCHANGE ifn result.
19639 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19640 IFN_ATOMIC_COMPARE_EXCHANGE.
19642 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19645 * opts.c (parse_sanitizer_options): Add missing question mark to
19646 "did you mean" message.
19648 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19650 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19652 (VMULEUH_UNS): Likewise.
19653 (VMULOUB_UNS): Likewise.
19654 (VMULOUH_UNS): Likewise.
19655 * config/rs6000/rs6000.c (builtin_function_type): Remove
19656 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19658 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19661 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19662 x with result of extra_parsing_for_operand_code_0.
19663 (function_reader::extra_parsing_for_operand_code_0): Convert
19664 return type from void to rtx, returning x. When reading
19665 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19666 larger size containing struct block_symbol.
19668 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19670 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19671 -mfloat128-hardware without -m64.
19673 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19676 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19677 entries to the case statement that marks unsigned arguments to
19678 overloaded functions.
19680 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19682 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19683 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19685 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19688 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19689 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19691 2017-03-10 Martin Liska <mliska@suse.cz>
19695 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19696 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19699 2017-03-10 Olivier Hainque <hainque@adacore.com>
19701 * tree-switch-conversion (array_value_type): Start by resetting
19702 candidate type to it's main variant.
19704 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19706 PR rtl-optimization/79909
19707 * combine.c (try_combine): Use simplify_replace_rtx on individual
19708 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19709 of the whole CALL_INSN_FUNCTION_USAGE.
19711 PR tree-optimization/79972
19712 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19713 get_range_info on SSA_NAMEs. Formatting fixes.
19715 2017-03-10 Richard Biener <rguenther@suse.de>
19716 Jakub Jelinek <jakub@redhat.com>
19718 PR tree-optimization/77975
19719 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19720 edge to be constant.
19721 (get_val_for): For constant x return it. Formatting fix.
19722 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19723 would use the same bases as the current one.
19725 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19727 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19728 instead of vec_select for V1TImode.
19729 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19731 (VSX_LE_128): Add V1TI to this mode iterator.
19732 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19733 (*vsx_le_perm_store_<mode>): Likewise.
19734 (pre-reload splitter for VSX stores): Likewise.
19735 (post-reload splitter for VSX stores): Likewise.
19736 (*vsx_xxpermdi2_le_<mode>): Likewise.
19737 (*vsx_lxvd2x2_le_<mode>): Likewise.
19738 (*vsx_stxvd2x2_le_<mode>): Likewise.
19740 2017-03-09 Michael Eager <eager@eagercon.com>
19742 Correct failures with --enable-checking=yes,rtl.
19744 * config/microblaze/microblaze.c (microblaze_expand_shift):
19745 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19746 test for const0_rtx.
19747 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19748 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19750 2017-03-09 Richard Biener <rguenther@suse.de>
19752 PR tree-optimization/79977
19753 * graphite-scop-detection.c (scop_detection::merge_sese):
19754 Handle the case of extra exits to blocks dominating the entry.
19756 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19758 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19761 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19763 PR rtl-optimization/79949
19764 * lra-constraints.c (process_alt_operands): Check memory when
19765 trying to predict a cycle. Print about the overall increase.
19767 2017-03-09 Richard Biener <rguenther@suse.de>
19769 PR middle-end/79971
19770 * gimple-expr.c (useless_type_conversion_p): Preserve
19771 TYPE_SATURATING for fixed-point types.
19773 2017-03-09 Richard Biener <rguenther@suse.de>
19776 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19777 alignment of BLKmode params.
19779 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19782 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19783 (VALL_NO_V2Q): Likewise.
19785 * config/aarch64/aarch64-simd.md
19786 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19788 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19789 VALL_NO_V2Q mode iterator.
19790 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
19792 2017-03-09 Martin Liska <mliska@suse.cz>
19794 PR tree-optimization/79631
19795 * tree-chkp-opt.c (chkp_is_constant_addr): Call
19796 tree_int_cst_sign_bit just for INTEGER constants.
19798 2017-03-09 Martin Liska <mliska@suse.cz>
19802 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
19805 2017-03-09 Marek Polacek <polacek@redhat.com>
19808 * tree.c (inchash::add_expr): Handle TREE_VEC.
19810 2017-03-09 Martin Liska <mliska@suse.cz>
19813 (chkp_narrow_size_and_offset): New function.
19814 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
19815 (void chkp_parse_bit_field_ref): New function.
19816 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
19817 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
19819 2017-03-09 Martin Liska <mliska@suse.cz>
19822 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
19823 (chkp_find_bounds_1): Remove gcc_unreachable.
19825 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19828 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19829 BUILT_IN_SYNC*, determine the access type from the size suffix and
19830 always build a MEM_REF with that type. Handle forgotten
19831 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19834 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19835 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19836 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19837 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19838 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19839 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19840 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19841 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19842 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19843 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19844 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19845 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19846 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19847 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19848 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19849 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19850 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19851 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19852 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19853 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19854 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19855 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19856 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19857 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19858 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19859 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19860 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19861 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19862 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19863 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19864 definitions outside of __OPTIMIZE__ guarded section.
19867 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19868 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19869 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19870 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19873 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19875 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19876 ("vfenez<mode>"): Add missing constraints.
19878 2017-03-08 Martin Sebor <msebor@redhat.com>
19881 * config/nds32/nds32.c (nds32_option_override):
19882 Fix misspelled diagnostic.
19884 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19887 * gimplify.c (gimplify_omp_for): Replace index var in outer
19888 taskloop statement with an artificial variable and add
19889 OMP_CLAUSE_PRIVATE clause for it.
19891 2017-03-08 Richard Biener <rguenther@suse.de>
19893 PR tree-optimization/79955
19894 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19895 for accesses that are completely outside of the variable.
19897 2017-03-08 Andrew Haley <aph@redhat.com>
19899 PR tree-optimization/79943
19900 * tree-ssa-loop-split.c (compute_new_first_bound): When
19901 calculating the new upper bound, (END-BEG) should be added, not
19904 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19906 * config/avr/avr.md (setmemhi): Make sure match_dup
19907 operand number comes before match_scratch.
19909 2017-03-08 Richard Biener <rguenther@suse.de>
19911 PR tree-optimization/79920
19912 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19913 with ncopies == 1 to ...
19914 (vect_transform_slp_perm_load): ... here. Properly compute
19915 all element loads by iterating VF times over the group. Do
19916 not handle ncopies (computed in a broken way) in
19917 vect_create_mask_and_perm.
19919 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19922 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19923 is a uniform vector, use uniform_vector_p return value instead of
19924 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19926 2017-03-07 Marek Polacek <polacek@redhat.com>
19928 PR middle-end/79809
19929 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19930 (alloca_call_type): Likewise.
19932 2017-03-07 Martin Liska <mliska@suse.cz>
19934 * gcov.c (process_args): Put comment to correct location.
19936 2017-03-07 Martin Liska <mliska@suse.cz>
19938 PR middle-end/68270
19939 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19940 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19941 (chkp_narrow_bounds_for_field): Likewise.
19942 (chkp_parse_array_and_component_ref): Pass one more argument to
19945 2017-03-07 Richard Biener <rguenther@suse.de>
19947 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19950 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19952 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19953 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19955 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19958 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19959 to end of description.
19960 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19962 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19964 PR rtl-optimization/79901
19965 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19967 (*avx512f_<code><mode>3<mask_name>): ... this.
19968 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19969 iterator instead of VI8_AVX2_AVX512BW.
19971 PR rtl-optimization/79901
19972 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19973 min/max expander, expand it using expand_vec_cond_expr.
19976 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19979 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19982 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19983 to void * for PCH reasons.
19984 * dwarf2out.c (output_loc_operands, output_die): Cast
19985 v.val_vec.array to unsigned char *.
19987 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19990 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19993 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19995 PR rtl-optimization/79571
19996 * lra-constraints.c (process_alt_operands): Calculate static
19997 reject and subtract it from overall when only addresses will be
20000 2017-03-06 Julia Koval <julia.koval@intel.com>
20003 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20004 incoming stack boundary to 128 for 64-bit targets.
20006 2017-03-06 Richard Biener <rguenther@suse.de>
20008 PR tree-optimization/79894
20009 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20010 to NULL after folding it.
20012 2017-03-06 Richard Biener <rguenther@suse.de>
20014 PR tree-optimization/79824
20015 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20016 check disabling peeling for gaps.
20018 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20020 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20021 attributes): Document gettimeofday.
20023 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20025 * config/s390/s390.c (s390_option_override_internal): Set
20026 PARAM_MIN_VECT_LOOP_BOUND
20028 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20030 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20031 * config/s390/s390.md: Likewise.
20033 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20036 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20037 (<avx2_avx512>_perm<mode>): Rename to ...
20038 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20040 (<avx512>_perm<mode>_mask): Rename to ...
20041 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20043 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20044 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20045 instead of VI8F_256_512.
20046 (avx512f_perm<mode>): New define_expand.
20047 (avx512f_perm<mode>_mask): Likewise.
20048 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20049 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20051 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20053 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20054 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20056 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20058 2017-03-06 Martin Liska <mliska@suse.cz>
20061 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20062 when having a SSA NAME w/o VAR_DECL assigned to it.
20064 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20066 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20067 msa_dpsub_<su>_d): Fix MODE for vec_select.
20069 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20071 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20073 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20075 2017-03-06 Richard Biener <rguenther@suse.de>
20077 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20078 * plugin.c (register_plugin_info): Likewise.
20079 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20081 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20083 * config/i386/sse.md (sse_storehps, sse_storelps,
20084 avx_<castmode><avxsizesuffix>_<castmode>,
20085 avx512f_<castmode><avxsizesuffix>_<castmode>,
20086 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20087 in condition that at least one operand is not a MEM.
20089 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20091 PR middle-end/79805
20092 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20093 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20095 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20096 gimple_call_nothrow_p flag based on whether original builtin can throw.
20097 If it can, emit following stmts on the fallthrough edge.
20098 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20099 don't create new bb if inserting just debug stmts on the edge, try to
20100 insert them on the fallthru bb or just reset debug stmts.
20102 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20105 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20106 restore recog_data (including the operand rtxes inside it) around
20107 the call to get_insn_template.
20109 2017-03-03 Martin Sebor <msebor@redhat.com>
20111 PR tree-optimization/79699
20112 * context.c (context::~context): Free MPFR caches to avoid
20113 a memory leak on program exit.
20115 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20117 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20118 Use wide_int::ulow () instead of .elt (0).
20120 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20122 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20123 (*pushxf): Limit oF constraint to 32bit targets and add oC
20124 constraint for 64bit targets.
20125 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20126 (*pushdf): Change rmF constraint to rmC.
20128 2017-03-03 Martin Liska <mliska@suse.cz>
20130 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20131 Remove unused variable.
20133 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20136 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20137 is a memory operand, increase num_memory.
20138 (ix86_expand_args_builtin): Likewise.
20140 2017-03-03 Jan Hubicka <jh@suse.cz>
20143 * ipa-devirt.c (maybe_record_node): Properly handle
20144 __cxa_pure_virtual visibility.
20146 2017-03-03 Martin Liska <mliska@suse.cz>
20148 PR tree-optimization/79803
20149 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20151 (pass_loop_prefetch::execute): Disabled optimization if an
20152 assumption about L1 cache size is not met.
20154 2017-03-03 Martin Liska <mliska@suse.cz>
20156 PR rtl-optimization/79574
20157 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20158 (hash_scan_set): Likewise.
20159 (dump_hash_table): Likewise.
20160 (hoist_code): Likewise.
20162 2017-03-03 Richard Biener <rguenther@suse.de>
20164 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20165 in place of uhigh ().
20166 (fixed_convert_from_real): Likewise.
20168 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20171 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20173 2017-03-03 Richard Biener <rguenther@suse.de>
20175 PR middle-end/79818
20176 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20177 TYPE_OVERFLOW_UNDEFINED check.
20179 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20181 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20183 (vector_ae_<mode>_p): Likewise.
20184 (vector_nez_<mode>_p): Likewise.
20185 (vector_ne_v2di_p): Likewise.
20186 (vector_ae_v2di_p): Likewise.
20187 (vector_ne_<mode>_p): Likewise.
20188 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20190 (vsx_tsqrt<mode>2_fe): Likewise.
20192 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20195 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20197 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20199 PR rtl-optimization/79780
20200 * cprop.c (one_cprop_pass): When second and further conditional trap
20201 in a single basic block is turned into an unconditional trap, turn it
20202 into a deleted note to avoid RTL verification failures.
20204 2017-03-02 Richard Biener <rguenther@suse.de>
20206 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20208 2017-03-02 Richard Biener <rguenther@suse.de>
20210 PR tree-optimization/79345
20212 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20213 param and abort the walk, returning -1 if it is hit.
20214 (walk_aliased_vdefs): Take a limit param and pass it on.
20215 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20216 defaulting to 0 and return a signed int.
20217 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20218 (check_defs): New helper.
20219 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20220 about uninitialized memory.
20221 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20222 bogus uninitialized warning.
20223 (fixed_convert_from_real): Likewise.
20225 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20227 PR tree-optimization/66768
20228 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20229 iv_use if base object can't be determined.
20231 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20233 PR tree-optimization/79345
20234 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20235 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20236 (get_pattern_stats): Initialize it.
20237 * genemit.c (gen_expand): Verify match_scratch numbers come after
20238 match_operand/match_dup numbers.
20239 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20240 match_scratch numbers.
20241 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20243 * config/s390/s390.md (trunctdsd2): Likewise.
20245 2017-03-02 Richard Biener <rguenther@suse.de>
20247 * wide-int.h (wide_int_storage::operator=): Implement in terms
20250 2017-03-02 Richard Biener <rguenther@suse.de>
20252 PR tree-optimization/79777
20253 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20254 the to insert expression to sth existing.
20256 2017-03-01 Martin Sebor <msebor@redhat.com>
20258 PR middle-end/79692
20259 * gimple-ssa-sprintf.c
20260 (directive::known_width_and_precision): New function.
20261 (format_integer): Use it.
20262 (get_mpfr_format_length): Consider the full range of precision
20263 when computing %g output with the # flag. Set the likely byte
20264 count to 3 rather than 1 when precision is indeterminate.
20265 (format_floating): Correct the lower bound of precision.
20267 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20269 * doc/invoke.texi: Document default code model for 64-bit Linux.
20271 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20274 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20275 udiv rather than div since input pattern is unsigned.
20277 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20279 * config/i386/i386.c (print_reg): Warn for values of
20280 unsupported size in integer register.
20282 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20285 * config/rs6000/predicates.md (current_file_function_operand): Do
20286 not allow self calls to be local if the function is replaceable.
20288 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20291 * config/rs6000/altivec.h (vec_ctz and others): Change the
20292 preprocessor macro that controls conditional compilation from
20293 _ARCH_PWR9 to __POWER9_VECTOR__.
20294 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20295 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20296 control (instead of _ARCH_PWR9 control) so that template
20297 definition uses power9-specific function.
20298 (vec_any_eq): Likewise.
20299 (vec_all_ne): Change macro definition to use a power9-specific
20300 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20301 _ARCH_PWR9 control).
20302 (vec_any_eq) Likewise.
20303 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20304 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20305 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20306 support for xvcmpnedp instruction.
20307 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20308 macro expansion so that Power9 implementation of vec_all_ne does
20309 not use the AltiVec predicate framework.
20310 (VCMPNEH_P): Likewise.
20311 (VCMPNEW_P): Likewise.
20312 (VCMPNED_P): Likewise.
20313 (VCMPNEFP_P): Likewise.
20314 (VCMPNEDP_P): Likewise.
20315 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20316 implementation of vec_any_eq to not use AltiVec predicate
20318 (VCMPAEH_P): Likewise.
20319 (VCMPAEW_P): Likewise.
20320 (VCMPAED_P): Likewise.
20321 (VCMPAEFP_P): Likewise.
20322 (VCMPAEDP_P): Likewise.
20323 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20324 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20325 not use the AltiVec predicate framework.
20326 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20327 of vec_any_eq to not use AltiVec predicate framework.
20328 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20329 support for predefined __POWER9_VECTOR__ macro to indicate that
20330 Power9 instruction selection is enabled.
20331 (altivec_overloaded_builtins): Remove extraneous
20332 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20333 function argument types RS6000_BTI_bool_V16QI and
20334 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20335 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20336 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20337 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20338 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20339 Power9 for implementations of vec_cmpne. Change the signature for
20340 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20341 (representing vec_all_ne) to remove the previously described first
20342 argument of type RS6000_BTI_INTSI, as this was an artifact of
20343 reliance on the AltiVec predicate framework, which is no longer
20344 used in the implementation of these functions. Add
20345 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20346 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20347 since, unlike the AltiVec predicate framework implementation, we
20348 do not share function descriptors between vec_alle and vec_anyeq.
20349 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20350 set of modes that receive special treatment even when
20351 TARGET_P9_VECTOR is true. The special treatment emits code that
20352 does not depend on Power9 instructions.
20353 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20354 define_expand to not rely on AltiVec predicate framework.
20355 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20357 (vector_ne_v2di_p): Change this define_expand to not rely on
20358 AltiVec predicate framework.
20359 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20361 (vector_ne_<mode>_p): Change this define_expand to not rely on
20362 AltiVec predicate framework.
20363 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20365 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20366 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20367 define_insn pattern.
20368 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20369 define_insn pattern because the xvcmpne<VSs>. instruction is not
20371 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20372 instruction is not supported.
20374 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20376 * config/nvptx/nvptx.c: Include intl.h.
20378 2017-03-01 Martin Jambor <mjambor@suse.cz>
20381 * ipa-prop.h (ipa_bits): Removed field known.
20382 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20383 to pointers. Adjusted their comments to warn about their sharing.
20384 (ipcp_transformation_summary): Change bits to a vector of pointers.
20385 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20386 (ipa_get_ipa_bits_for_value): Declare.
20387 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20388 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20389 (ipa_bits_hash_table): Likewise.
20390 (ipa_vr_ggc_hash_traits): Likewise.
20391 (ipa_vr_hash_table): Likewise.
20392 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20393 being pointers and vr_known being removed.
20394 (ipa_set_jf_unknown): Likewise.
20395 (ipa_get_ipa_bits_for_value): New function.
20396 (ipa_set_jfunc_bits): Likewise.
20397 (ipa_get_value_range): New overloaded functions.
20398 (ipa_set_jfunc_vr): Likewise.
20399 (ipa_compute_jump_functions_for_edge): Use the above functions to
20400 construct bits and vr parts of jump functions.
20401 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20402 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20404 (ipcp_grow_transformations_if_necessary): Also allocate
20405 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20407 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20408 them. Fix too long lines.
20409 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20410 vr_known being removed.
20411 (ipa_read_jump_function): Use new setter functions to construct bits
20412 and vr parts of jump functions or set them to NULL.
20413 (write_ipcp_transformation_info): Adjust for bits being pointers.
20414 (read_ipcp_transformation_info): Likewise.
20415 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20417 Include gt-ipa-prop.h.
20418 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20420 (ipcp_store_bits_results): Likewise.
20421 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20422 Do not write to existing jump functions but use a temporary instead.
20424 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20427 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20428 attempt to use its first operand as BIT_FIELD_REF base.
20430 2017-03-01 Richard Biener <rguenther@suse.de>
20432 PR middle-end/79721
20433 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20434 interpolating formula in wrapping arithmetic.
20435 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20437 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20439 PR tree-optimization/79734
20440 * tree-vect-generic.c (expand_vector_condition): Optimize
20441 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20442 Handle VEC_COND_EXPR where comparison has different inner width from
20443 type's inner width.
20445 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20447 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20448 markup, and similar issues. Remove @opindex entries for things
20449 that aren't options. Add missing -mmpy-option entries.
20451 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20453 PR tree-optimization/79737
20454 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20455 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20456 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20457 instead of byte_size. Formatting fix.
20458 (shift_bytes_in_array_right): Formatting fix.
20460 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20463 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20464 condition on optimize for the leaf function test.
20466 2017-02-28 Martin Liska <mliska@suse.cz>
20469 * read-rtl-function.c (function_reader::handle_unknown_directive):
20470 Bail out when one uses -flto.
20472 2017-02-28 Martin Liska <mliska@suse.cz>
20474 * common.opt: Replace space with tabular for options of <number>
20476 * config/i386/i386.opt: Show <number> value for
20477 -mlarge-data-threshold.
20478 * opts.c (print_filtered_help): Do not display number in hexadecimal
20481 2017-02-28 Martin Liska <mliska@suse.cz>
20483 * common.opt: Fix --help=option -Q for options which are of
20486 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20488 * config/i386/i386.c (print_reg): Error out for values
20489 of 8-bit size in invalid integer register.
20491 2017-02-28 Martin Sebor <msebor@redhat.com>
20493 PR tree-optimization/79691
20494 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20496 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20499 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20500 gcc_unreachable with output_operand_lossage.
20502 2017-02-28 Richard Biener <rguenther@suse.de>
20504 PR tree-optimization/79740
20505 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20507 (visit_nary_op): Insert the nary into the hashtable if we
20508 pattern-matched sth.
20509 * tree-ssa-pre.c (eliminate_insert): Robustify.
20511 2017-02-28 Richard Biener <rguenther@suse.de>
20513 PR middle-end/79731
20514 * fold-const.c (decode_field_reference): Reject out-of-bound
20517 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20519 * config/i386/i386.c: Include intl.h.
20520 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20521 instead of just cond ? "..." : "...".
20522 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20523 * coverage.c (read_counts_file): Likewise.
20524 * omp-offload.c: Include intl.h.
20525 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20526 of just cond ? "..." : "...".
20527 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20528 of just cond ? "..." : "...".
20530 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20533 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20535 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20536 'tune for' CPU name.
20537 * config/arm/arm-cpu-data.h: Regenerated.
20539 2017-02-28 Richard Biener <rguenther@suse.de>
20541 PR tree-optimization/79732
20542 * tree-inline.c (expand_call_inline): Do not shadow var.
20544 2017-02-28 Richard Biener <rguenther@suse.de>
20546 PR tree-optimization/79723
20547 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20548 address-space properly.
20550 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20552 * doc/optinfo.texi (Optimization groups): Fix option used for
20554 * doc/invoke.texi (-fopt-info): Document "omp".
20555 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20556 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20557 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20558 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20559 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20561 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20563 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20564 OPTGROUP_OMP with "omp".
20566 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20569 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20570 for arithmetic shift of unsigned V2DI.
20572 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20574 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20575 arc/linux.h headers.
20576 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20577 (LINK_SPEC): Likewise.
20578 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20579 (EXTRA_SPECS): Likewise.
20580 (STARTFILE_SPEC): Likewise.
20581 (ENDFILE_SPEC): Likewise.
20582 (LIB_SPEC): Likewise.
20583 (TARGET_SDATA_DEFAULT): Likewise.
20584 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20585 (MULTILIB_DEFAULTS): Likewise.
20586 (DWARF2_UNWIND_INFO): Likewise.
20587 * config/arc/big.h: New file.
20588 * config/arc/elf.h: Likewise.
20589 * config/arc/linux.h: Likewise.
20590 * config/arc/t-uClibc: Remove.
20592 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20594 PR tree-optimization/77536
20595 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20596 (tree_transform_and_unroll_loop): Use above function to compute the
20597 estimated niter of unrolled loop and use it when scaling profile.
20598 Also use count info rather than frequency if it's non-zero.
20599 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20600 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20601 (vect_transform_loop): Call above function.
20603 2017-02-27 Richard Biener <rguenther@suse.de>
20605 PR tree-optimization/45397
20606 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20607 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20608 (visit_nary_op): Add pattern matching for CSEing sign-changed
20609 or truncated operations with wider ones.
20611 2017-02-27 Richard Biener <rguenther@suse.de>
20613 PR tree-optimization/79690
20614 * tree-vect-stmts.c (vectorizable_store): Use vector type
20615 built from the DR with address-space.
20617 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20619 * doc/invoke.texi (Optimize Options): Refine the description
20620 of asan-use-after-return.
20622 2017-02-25 Alan Modra <amodra@gmail.com>
20624 PR rtl-optimization/79584
20625 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20626 base, not ad->base_term, the reg within base. Remove assertion
20627 that ad->base == ad->base_term. Replace gen_int_mode using
20628 bogus mode with const0_rtx.
20630 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20632 PR middle-end/79396
20633 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20634 FMA_EXPR like tcc_binary or tcc_unary.
20636 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20639 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20641 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20642 (output_loc_operands): Handle DW_OP_call_ref and
20643 DW_OP_GNU_variable_value.
20644 (struct variable_value_struct): New type.
20645 (struct variable_value_hasher): Likewise.
20646 (variable_value_hash): New variable.
20647 (string_types): Remove.
20648 (copy_loc_descr): New function.
20649 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20650 (prepend_loc_descr_to_each): New function.
20651 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20652 instead of add_loc_descr_to_each if the first argument is single
20653 location list and the second has multiple.
20654 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20655 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20656 when looking for variable value which doesn't have other location info.
20657 (loc_list_from_tree): Formatting fix.
20658 (gen_array_type_die): Simplify DW_AT_string_length handling.
20659 (adjust_string_types): Remove.
20660 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20661 string_types. Call resolve_variable_values.
20662 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20663 (resolve_addr_in_expr): Likewise. Add A argument.
20664 (copy_deref_exprloc): Remove deref argument. Adjust for the
20665 original expression being DW_OP_GNU_variable_value with optionally
20666 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20667 optionally after it.
20668 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20669 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20670 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20671 (variable_value_hasher::hash, variable_value_hasher::equal): New
20673 (resolve_variable_value_in_expr, resolve_variable_value,
20674 resolve_variable_values, note_variable_value_in_expr,
20675 note_variable_value): New functions.
20676 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20679 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20682 * opts.h (handle_generated_option): Add GENERATED_P argument.
20683 * opts-common.c (handle_option): Adjust function comment.
20684 (handle_generated_option): Add GENERATED_P argument, pass it to
20686 (control_warning_option): Pass false to handle_generated_option
20688 * opts.c (maybe_default_option): Pass true to handle_generated_option
20690 * optc-gen.awk: Likewise.
20692 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20694 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20695 a REG, look at the REG it is a SUBREG of.
20696 (splitter for cmpeqsi_t): Ditto.
20698 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20700 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20701 the special USEs with the pattern of the insn, not the insn itself.
20703 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20706 * doc/invoke.texi: Document -mload-store-pairs.
20708 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20709 Sandra Loosemore <sandra@codesourcery.com>
20711 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20712 argument isn't a CONST_INT.
20713 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20714 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20715 (nios2_validate_compare): Bypass alternate compare logic if *op2
20716 is not a CONST_INT.
20717 (ldstwm_operation_p): Return false if first_base is not a REG or
20718 if first_offset is not a CONST_INT.
20720 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20722 * config/cris/cris.md: Use correct operand in a define_peephole2.
20724 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20726 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20728 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20730 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20731 this_insn if it is an INSN or JUMP_INSN.
20732 (force_offsettable): Look at base, not at addr.
20733 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20734 on things that aren't necessarily CONST_INTs.
20736 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20738 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20739 -mfpmath=sse is the default also for x86-32 targets with SSE2
20740 instruction set when @option{-ffast-math} is enabled
20742 2017-02-24 Jeff Law <law@redhat.com>
20744 PR rtl-optimizatoin/79286
20745 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20748 2017-02-24 Richard Biener <rguenther@suse.de>
20750 PR tree-optimization/79389
20751 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20754 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20756 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20757 function comment to reflect reality.
20758 (loop_exits_before_overflow): Fix typo in function description.
20760 2017-02-24 Richard Biener <rguenther@suse.de>
20762 PR tree-optimization/79389
20763 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20764 properly that a threading opportunity exists. Detect conditional
20765 copy/constant propagation opportunities.
20767 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20769 * config/visium/visium.md (type): Add trap.
20770 (b): New mode attribute.
20771 (*btst): Rename into...
20772 (*btst<mode>): ...this and adjust.
20773 (*cbranchsi4_btst_insn): Rename into...
20774 (*cbranch<mode>4_btst_insn): ...this and adjust.
20775 (trap): New define_insn.
20777 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20779 PR tree-optimization/79389
20780 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20781 (noce_reversed_cond_code): New function.
20782 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20783 reversed_comparison_code. Formatting fix.
20784 (noce_try_store_flag): Test rev_cond != NULL in addition to
20785 reversed_comparison_code.
20786 (noce_try_store_flag_constants): Likewise.
20787 (noce_try_store_flag_mask): Likewise.
20788 (noce_try_addcc): Use rev_cond if non-NULL instead of
20789 reversed_comparison_code.
20790 (noce_try_cmove_arith): Likewise. Formatting fixes.
20791 (noce_try_minmax, noce_try_abs): Clear rev_cond.
20792 (noce_find_if_block): Initialize rev_cond.
20793 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
20794 instead of false as last argument never attempt to reverse it
20797 2017-02-23 Bin Cheng <bin.cheng@arm.com>
20799 PR tree-optimization/79663
20800 * tree-predcom.c (combine_chains): Process refs in reverse order
20801 only for ZERO length chains, and add explaining comment.
20803 2017-02-23 Jeff Law <law@redhat.com>
20805 PR tree-optimization/79578
20806 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
20807 in call to operand_equal_p.
20809 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
20812 * config/i386/cpuid.h: Fix another undefined behavior.
20814 2017-02-23 Richard Biener <rguenther@suse.de>
20816 PR tree-optimization/79683
20817 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
20818 vector types for data-refs.
20820 2017-02-23 Martin Liska <mliska@suse.cz>
20822 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
20824 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20826 PR middle-end/79665
20827 * internal-fn.c (get_range_pos_neg): Moved to ...
20828 * tree.c (get_range_pos_neg): ... here. No longer static.
20829 * tree.h (get_range_pos_neg): New prototype.
20830 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20831 are known to be in between 0 and signed maximum inclusive, try to
20832 expand both unsigned and signed divmod and use the cheaper one from
20835 2017-02-22 Jeff Law <law@redhat.com>
20837 PR tree-optimization/79578
20838 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20839 to compare base operands.
20841 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20844 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20845 gpc_reg_operand instead of fpr_reg_operand.
20847 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20849 * config/mips/mips.c (mips_return_in_memory): Force FP
20850 vector types to be returned in memory for o32 ABI.
20852 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20854 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20855 instead of DW_TAG_member for static data member declarations and don't
20856 set no_linkage_name for static inline data members.
20857 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20860 2017-02-22 Martin Liska <mliska@suse.cz>
20862 * doc/invoke.texi: Replace inequality signs with square brackets
20865 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20867 PR tree-optimization/68644
20868 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20870 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20873 * lra-constraints.c (simplify_operand_subreg): Handle
20874 WORD_REGISTER_OPERATIONS targets.
20876 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20879 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20880 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20881 elimination by swapping fld*.
20883 2017-02-22 Richard Biener <rguenther@suse.de>
20885 PR tree-optimization/79673
20886 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20887 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20888 irrelevant address-space qualifiers and avoiding a
20889 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20891 2017-02-22 Richard Biener <rguenther@suse.de>
20893 PR tree-optimization/79666
20894 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20895 to not symbolically negate if that may introduce undefined
20898 2017-02-22 Martin Liska <mliska@suse.cz>
20901 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20902 * data-streamer-out.c (streamer_write_gcov_count_stream):
20904 * value-prof.c (stream_out_histogram_value): Make assert more
20905 precise based on type of counter.
20907 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20910 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20911 Use nonimmediate_operand instead of memory_operand for operand 1.
20912 (float-extend standard_x87sse_constant_load splitter): Ditto.
20914 2017-02-21 Jeff Law <law@redhat.com>
20916 PR tree-optimization/79621
20917 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20918 blocks with edges to themselves.
20920 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20923 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20924 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20925 Use gimple_call_builtin_p.
20928 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20929 on temporarily removed DEBUG_INSNs.
20931 PR tree-optimization/79649
20932 * tree-loop-distribution.c (classify_partition): Give up on
20933 non-generic address space loads/stores.
20935 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20937 * doc/loop.texi (Loop manipulation): Remove nonexistent
20938 tree_ssa_loop_version from the documentation.
20939 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20941 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20944 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20945 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20946 * config/rs6000/rs6000.c: Include except.h.
20947 (rs6000_expand_split_stack_prologue): Call
20948 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20950 2017-02-21 Martin Jambor <mjambor@suse.cz>
20953 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20954 have been analyzed.
20956 2017-02-21 Martin Jambor <mjambor@suse.cz>
20958 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20959 for backward compatibility only.
20960 * doc/invoke.texi (Option Summary): Remove all references to
20961 -fipa-cp-alignment.
20963 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20967 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20969 * lra-constraints.c (curr_insn_transform): Handle
20970 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20972 2017-02-21 Martin Liska <mliska@suse.cz>
20974 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20976 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20978 PR translation/79638
20979 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20981 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20984 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20985 (arm_function_ok_for_sibcall): Return false for an indirect call by
20986 descriptor if all the argument registers are used.
20987 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20988 alignment of the function.
20990 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20992 PR tree-optimization/61441
20993 * simplify-rtx.c (simplify_const_unary_operation): For
20994 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20995 the sNaN unmodified.
20997 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20999 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21000 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21001 instead of SYSTEM_HEADER_DIR.
21003 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21004 Martin Liška <mliska@suse.cz>
21006 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21007 Fix typos and grammar, use active voice, and clarify.
21009 2017-02-20 Marek Polacek <polacek@redhat.com>
21011 PR middle-end/79537
21012 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21015 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21017 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21020 * config/i386/i386.c (ix86_expand_builtin): Handle
21021 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21022 ix86_builtins_isa[fcode].isa as a requirement of those
21023 flags and any other flag in the bitmask.
21024 (ix86_init_mmx_sse_builtins): Use 0 instead of
21025 ~OPTION_MASK_ISA_64BIT as mask.
21026 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21027 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21028 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21029 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21031 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21034 * lra-constraints.c (split_reg): Check requested split mode
21035 is supported by the register.
21037 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21039 * lra-constraints.c (simplify_operand_subreg): Remove early
21042 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21045 * lra-constraints.c (curr_insn_transform): Tighten condition
21046 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21048 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21051 * lra-constraints.c (curr_insn_transform): Handle
21052 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21054 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21057 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21059 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21061 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21064 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21067 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21069 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21072 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21074 * final.c (last_columnnum, override_columnnum): New variables.
21075 (final_start_function): Set last_columnnum, pass it to begin_prologue
21076 hook and pass 0 to dwarf2out_begin_prologue.
21077 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21078 to source_line debug hook.
21079 (notice_source_line): Compute last_columnnum and for debug_column_info
21080 return true on column changes.
21081 * debug.h (struct gcc_debug_hooks): Add column argument to
21082 source_line and begin_prologue hooks.
21083 (debug_nothing_int_charstar_int_bool): Remove prototype.
21084 (debug_nothing_int_int_charstar,
21085 debug_nothing_int_int_charstar_int_bool): New prototypes.
21086 (dwarf2out_begin_prologue): Add column argument.
21087 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21088 begin_prologue hooks.
21089 (debug_nothing_int_charstar_int_bool): Remove.
21090 (debug_nothing_int_int_charstar,
21091 debug_nothing_int_int_charstar_int_bool): New functions.
21092 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21093 through to dwarf2out_source_line.
21094 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21095 (dwarf2out_source_line): Add column argument, emit it if requested.
21096 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21098 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21099 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21100 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21101 through to dwarf2out_begin_prologue.
21102 (vmsdbgout_source_line): Add column argument, pass it through to
21103 dwarf2out_source_line.
21104 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21105 dbxout_source_line caller.
21106 (dbxout_source_line): Add column argument.
21108 * common.opt (gno-column-info, gcolumn-info): New options.
21109 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21110 (check_die): Also test for multiple DW_AT_decl_column attributes.
21111 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21112 DW_AT_decl_column if requested.
21113 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21115 (gen_variable_die): Likewise.
21116 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21117 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21120 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21121 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21122 (ix86_handle_option): Handle OPT_m3dnowa.
21123 * doc/invoke.texi (-m3dnowa): Document.
21124 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21125 -m3dnowa instead of -m3dnow -march=athlon.
21128 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21129 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21131 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21134 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21135 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21136 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21137 generator for vsx_xxpermdi_<mode>_be.
21138 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21139 force big-endian semantics.
21140 (vsx_xxpermdi_<mode>_be): New define_expand with same
21141 implementation as previous version of vsx_xxpermdi_<mode>.
21143 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21145 PR tree-optimization/79327
21146 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21147 variable, its initialization and use.
21149 2017-02-17 Julia Koval <julia.koval@intel.com>
21151 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21152 (OPTION_MASK_ISA_PKU_UNSET): New.
21153 (ix86_handle_option): Handle -mrdpid.
21154 * config/i386/cpuid.h (bit_RDPID): New.
21155 * config/i386/driver-i386.c (host_detect_local_cpu):
21156 Detect RDPID feature.
21157 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21158 * config/i386/i386-c.c (ix86_target_macros_internal):
21160 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21161 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21162 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21163 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21164 * config/i386/i386.md (define_insn "rdpid"): New.
21165 * config/i386/i386.opt Add -mrdpid.
21166 * config/i386/immintrin.h (_rdpid_u32): New.
21168 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21170 PR rtl-optimization/79541
21171 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21172 instead of transforming it into USE.
21174 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21176 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21177 If HONOR_SNANS (SFmode) force the input to a register.
21178 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21179 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21180 an frsp or similar insn.
21182 2017-02-17 Martin Liska <mliska@suse.cz>
21184 PR rtl-optimization/79577
21185 * params.def (selsched-max-sched-times): Increase minimum to 1.
21187 2017-02-17 Martin Liska <mliska@suse.cz>
21189 PR rtl-optimization/79574
21190 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21192 2017-02-17 Martin Liska <mliska@suse.cz>
21194 PR tree-optimization/79529
21195 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21196 ssa_defined_default_def_p to handle cases which are implicitly
21198 * tree-ssa.c (ssa_defined_default_def_p): New function.
21199 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21200 which are implicitly defined.
21201 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21203 2017-02-17 Richard Biener <rguenther@suse.de>
21205 PR middle-end/79576
21206 * params.def (max-ssa-name-query-depth): Limit to 10.
21208 2017-02-17 Richard Biener <rguenther@suse.de>
21210 PR tree-optimization/79552
21211 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21214 2017-02-17 Richard Biener <rguenther@suse.de>
21217 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21219 2017-02-17 Marek Polacek <polacek@redhat.com>
21221 PR middle-end/79536
21222 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21223 (fold_negate_expr): New wrapper.
21225 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21227 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21228 Correct terminology and de-emphasize pre-standard behavior.
21230 2017-02-16 Alan Modra <amodra@gmail.com>
21232 PR rtl-optimization/79286
21233 * ira.c (def_dominates_uses): New function.
21234 (update_equiv_regs): Don't create an equivalence for insns that
21235 may trap where the register def does not dominate the use.
21237 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21239 PR rtl-optimization/78127
21240 * lra.c (lra): Call lra_eliminate before finish the loop after
21243 2017-02-16 Richard Biener <rguenther@suse.de>
21245 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21247 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21248 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21249 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21250 (isl_val_int_from_wi): New function.
21251 (extract_affine_gmp): Rename to ...
21252 (extract_affine_wi): ... this, take a widest_int.
21253 (extract_affine_int): Just wrap extract_affine_wi.
21254 (add_param_constraints): Use isl_val_int_from_wi.
21255 (add_loop_constraints): Likewise, and extract_affine_wi.
21257 2017-02-15 Jeff Law <law@redhat.com>
21259 PR middle-end/79521
21260 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21261 ira_init_register_move_cost_if_necessary.
21263 2017-02-15 Martin Sebor <msebor@redhat.com>
21265 PR middle-end/32003
21266 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21267 removed in a prior commit.
21269 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21271 PR tree-optimization/79347
21272 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21273 counters during peeling.
21275 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21277 * Makefile.in (site.exp): Remove "set ISLVER".
21279 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21282 * real.c (real_from_integer): Call real_convert even for decimal.
21284 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21287 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21289 2017-02-14 Andrew Pinski <apinski@cavium.com>
21291 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21292 cores and change the partno/implementer to be correct.
21293 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21294 the 'B" as the implementer.
21295 * config/aarch64/aarch64-tune.md: Regenerate.
21297 2017-02-14 Carl Love <cel@us.ibm.com>
21299 * config/rs6000/rs6000.c: Add case statement entry to make the
21300 xvcvuxdsp built-in argument unsigned.
21301 * config/rs6000/vsx.md: Fix the source and return operand types so they
21302 match the instruction definitions from the ISA document. Fix typo
21303 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21306 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21309 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21310 member early_clobber_alts.
21311 * lra-lives.c (reg_early_clobber_p): New.
21312 (process_bb_lives): Use it.
21313 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21314 (debug_operand_data): Initialize early_clobber_alts.
21315 (setup_operand_alternative): Set up early_clobber_alts.
21316 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21317 alternatives to new_insn_reg.
21318 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21320 (lra_update_insn_regno_info): Pass the new arg.
21322 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21324 PR middle-end/79505
21325 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21326 (new_oacc_loop_raw): Don't clear already cleared fields.
21329 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21330 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21331 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21333 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21336 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21338 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21341 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21342 the extra instruction to the right place to store 128-bit constant
21345 2017-02-14 Martin Sebor <msebor@redhat.com>
21347 PR middle-end/79448
21348 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21349 warning for strings of unknown length.
21351 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21353 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21355 2017-02-14 Jeff Law <law@redhat.com>
21358 * ira-costs.c (scan_one_insn): Initialize register move costs
21359 for pseudos seen in USE/CLOBBER insns.
21361 PR tree-optimization/79095
21362 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21363 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21364 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21365 if the operands are known to be not equal, then the resulting range
21367 (intersect_ranges): If the new range is ~[0,0] and the old range is
21368 wide, then prefer ~[0,0].
21369 * tree-vrp.c (overflow_comparison_p_1): New function.
21370 (overflow_comparison_p): New function.
21371 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21372 if NAME is used in an overflow test.
21373 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21374 overflow check that can be expressed as an equality test, then adjust
21375 ops to be that equality test.
21377 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21379 * config/s390/s390-builtin-types.def: Remove flags argument.
21380 * config/s390/s390.c (s390_init_builtins): Likewise.
21382 2017-02-14 Martin Liska <mliska@suse.cz>
21384 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21385 vector. Fix trailing white spaces.
21387 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21389 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21392 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21394 PR rtl-optimization/68664
21395 * config/arm/arm.c (arm_sched_can_speculate_insn):
21396 New function. Declare prototype.
21397 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21399 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21401 PR rtl-optimization/68664
21402 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21404 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21406 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21408 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21409 max skip bytes for function, loop and jump.
21411 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21413 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21414 ABS_EXPR for gimple dump.
21416 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21419 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21421 PR tree-optimization/79408
21422 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21423 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21424 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21425 also if rhs1 is INTEGER_CST.
21427 2017-02-14 Richard Biener <rguenther@suse.de>
21429 PR middle-end/79432
21430 * tree-into-ssa.c (insert_phi_nodes): When the function can
21431 have abnormal edges rewrite SSA names with broken use-def
21432 dominance out of SSA and register them for PHI insertion.
21434 2017-02-13 Martin Sebor <msebor@redhat.com>
21436 PR middle-end/79496
21437 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21438 clearing info.nowrite flag when snprintf size argument is a range.
21440 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21442 * cprop.c (cprop_jump): Add missing space in string literal.
21443 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21444 (get_constraint_for_component_ref): Likewise.
21445 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21446 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21447 * lra-constraints.c (process_alt_operands): Likewise.
21448 * ipa-inline.c (inline_small_functions): Likewise.
21449 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21450 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21451 * trans-mem.c (diagnose_tm_1_op): Likewise.
21452 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21453 (grid_parallel_clauses_gridifiable): Likewise.
21455 * config/nvptx/mkoffload.c (process): Add space in between
21458 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21459 "MOD4_SSE_REGS" and "ALL_REGS".
21461 * spellcheck.c (test_data): Add , in between "foo" and "food".
21463 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21466 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21467 boundary crossing check and subsequent code generation agree.
21469 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21471 * config/aarch64/aarch64.c (has_memory_op): Delete.
21472 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21475 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21477 PR rtl-optimization/79388
21478 PR rtl-optimization/79450
21479 * combine.c (distribute_notes): When removing TEM_INSN for which
21480 corresponding dest has last value recorded, invalidate that last
21483 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21485 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21486 of explicit '@'. Add missing assembly comment marker on branch costs
21489 2017-02-13 Nathan Sidwell <nathan@acm.org>
21491 * gengtype-lex.l (<in_struct>): Add '/'.
21493 2017-02-13 Martin Liska <mliska@suse.cz>
21496 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21498 2017-02-13 Richard Biener <rguenther@suse.de>
21500 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21502 * configure: Re-generate.
21503 * config.in: Likewise.
21504 * graphite-dependences.c: Simplify as if
21505 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21506 * graphite-isl-ast-to-gimple.c: Likewise.
21507 * graphite-optimize-isl.c: Likewise.
21508 * graphite-poly.c: Likewise.
21509 * graphite-sese-to-poly.c: Likewise.
21510 * graphite.h: Likewise.
21511 * toplev.c: Include isl/version.h and use isl_version () for
21512 printing the ISL version.
21513 * doc/install.texi: Update ISL requirement.
21515 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21517 * doc/standards.texi (Standards): Update reference to
21520 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21522 * doc/extend.texi (Named Address Spaces): sourceware.org now
21524 * doc/install.texi (Binaries): Ditto.
21527 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21529 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21530 standard terminology "stringize"/"stringizing" throughout.
21531 * doc/cppinternals.texi: Likewise.
21533 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21535 * doc/extend.texi: Fix some spelling mistakes and typos.
21536 * doc/invoke.texi: Likewise.
21538 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21541 * params.def (inline-min-speedup) Change from 10 to 8.
21543 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21545 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21548 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21551 * ipa-inline-analysis.c (get_minimal_bb): New function.
21552 (record_modified): Use it.
21553 (remap_edge_change_prob): Handle also ancestor functions.
21555 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21557 * doc/contrib.texi (Contributors): Remove broken link into
21558 the Mauve CVS repository.
21560 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21562 PR middle-end/79454
21563 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21564 result computation whenever lhs doesn't have vector mode, not
21565 just when it has BLKmode.
21567 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21569 * doc/makefile.texi (profiledbootstrap): Refer to the
21570 installation instructions only in textual form.
21572 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21575 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21577 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21579 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21580 (Specific): Update mingw-w64 reference.
21582 (Specific): Remove broken link to Renesas RX processor.
21584 2017-02-10 Richard Biener <rguenther@suse.de>
21586 * toplev.c (process_options): Do not mention obsolete graphite
21587 options when printing sorry message about missing graphite support.
21588 Mention -floop-nest-optimize.
21590 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21592 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21593 (vtst_p16): Likewise.
21594 (vtstq_p8): Likewise.
21595 (vtstq_p16): Likewise.
21597 (vtstq_p64): Likewise.
21598 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21599 (vset_lane_p64): New.
21600 (vsetq_lane_p64): New.
21602 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21604 PR tree-optimization/79411
21605 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21606 stmt operands are SSA_NAMEs used in abnormal phis.
21607 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21610 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21613 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21616 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21618 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21620 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21622 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21626 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21627 not arbitrary TREE_CONSTANT.
21630 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21631 "omp declare target link" attribute unless is_global_var.
21632 * omp-offload.c (find_link_var_op): Likewise.
21634 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21635 Chung-Lin Tang <cltang@codesourcery.com>
21637 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21639 (gimplify_adjust_omp_clauses): Don't delete TILE.
21640 (gimplify_omp_for): Deal with TILE.
21641 * internal-fn.c (expand_GOACC_TILE): New function.
21642 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21644 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21645 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21647 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21648 avoid DIV for outermost collapse var.
21649 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21650 Remove out of date comments, fix whitespace.
21651 * omp-general.c (omp_extract_for_data): Deal with tiling.
21652 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21653 adjust OLF_DIM_BASE value.
21654 (struct omp_for_data): Add tiling field.
21655 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21656 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21657 for auto loops. Remove default auto determining, moved to
21658 oacc_loop_fixed_partitions.
21659 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21660 stmts, add e_mask field.
21661 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21662 (oacc_thread_numbers): Use oacc_dim_call.
21663 (oacc_xform_tile): New.
21664 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21665 (finish_oacc_loop): Adjust for ifns vector.
21666 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21667 add case for GOACC_TILE fns.
21668 (oacc_loop_xform_loop): Delete.
21669 (oacc_loop_process): Iterate over call list directly, and add
21670 handling for GOACC_TILE fns.
21671 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21673 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21674 vector partitioning to outer loops. Assign 2 partitions to loops
21675 when available. Add TILE handling.
21676 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21677 (execite_oacc_device_lower): Process GOACC_TILE fns,
21678 ignore unknown specs.
21679 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21680 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21681 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21683 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21685 * configure.ac (ACX_BUGURL): Update.
21686 * configure: Regenerate.
21688 2017-02-09 Richard Biener <rguenther@suse.de>
21690 PR tree-optimization/69823
21691 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21692 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21694 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21696 * config/arc/arc-c.def: Add __NPS400__ definition.
21697 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21698 (TARGET_NPS400): Define.
21700 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21702 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21704 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21705 pointer, arch_info.
21706 (arc_cpu_types): Fill the arch_info field with a pointer into the
21707 arc_arch_types table.
21708 (arc_selected_cpu): Declare.
21709 * config/arc/arc.c (arc_selected_cpu): Make global.
21710 (arc_selected_arch): Delete.
21711 (arc_base_cpu): Delete.
21712 (arc_override_options): Remove references to deleted variables,
21713 update access to arch information.
21714 (ARC_OPT): Update access to arch information.
21715 (ARC_OPTX): Likewise.
21716 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21717 (TARGET_ARC600): Update access to arch information.
21718 (TARGET_ARC601): Likewise.
21719 (TARGET_ARC700): Likewise.
21720 (TARGET_EM): Likewise.
21721 (TARGET_HS): Likewise.
21722 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21725 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21728 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21729 condition/operands for integer GE/LE/GEU/LEU operations.
21731 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21733 PR translation/79397
21734 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21737 2017-02-08 Martin Jambor <mjambor@suse.cz>
21740 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21741 whether allocation happened.
21742 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21743 nothing was allocated.
21745 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21747 PR tree-optimization/79408
21748 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21749 constant, but SSA_NAME with a known integer range, use the minimum
21750 of that range instead of op1 to determine if modulo can be replaced
21751 with its first operand.
21753 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21755 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21757 2017-02-08 Richard Biener <rguenther@suse.de>
21759 PR tree-optimization/71824
21760 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21761 Check all loops contained in the merged region.
21763 2017-02-07 Andrew Pinski <apinski@cavium.com>
21765 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21767 2017-02-07 Andrew Pinski <apinski@cavium.com>
21769 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21770 (thunderxt88): Likewise.
21771 (thunderxt81): Disable LSE and change v8.1 to v8.
21772 (thunderxt83): Likewise.
21774 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21775 Richard Biener <rguenther@suse.de>
21777 PR middle-end/79399
21778 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21779 type from int to size_t.
21780 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21782 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21784 PR rtl-optimization/79386
21785 * cprop.c (bypass_conditional_jumps): Initialize
21786 bypass_last_basic_block already before splitting bbs after
21787 unconditional traps...
21788 (bypass_conditional_jumps): ... rather than here.
21791 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
21792 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
21793 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
21794 fix -masm=intel patterns.
21796 2017-02-07 Richard Biener <rguenther@suse.de>
21798 PR tree-optimization/79256
21799 PR middle-end/79278
21800 * builtins.c (get_object_alignment_2): Use min_align_of_type
21801 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
21802 and ADJUST_FIELD_ALIGN.
21804 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
21806 * doc/tm.texi: Regenerate.
21807 * stor-layout.c (layout_decl): Adjust.
21808 (update_alignment_for_field): Likewise.
21809 (place_field): Likewise.
21810 (min_align_of_type): Likewise.
21811 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
21812 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
21813 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
21814 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
21815 * config/frv/frv.c (frv_adjust_field_align): Likewise.
21816 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
21817 * config/i386/i386.c (x86_field_alignment): Likewise.
21818 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
21819 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
21820 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
21821 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
21822 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
21823 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21827 2017-01-30 Richard Biener <rguenther@suse.de>
21829 PR tree-optimization/79256
21830 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21831 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21834 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21836 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21837 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21838 builtins to SImode and emit a zero-extend, if necessary.
21840 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21842 * docs/invoke.texi (RISC-V Options): Alphabetize.
21844 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21846 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21849 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21851 * config/riscv/riscv.c: New file.
21852 * common/config/riscv/riscv-common.c: Likewise.
21853 * config.gcc: Likewise.
21854 * config/riscv/constraints.md: Likewise.
21855 * config/riscv/elf.h: Likewise.
21856 * config/riscv/generic.md: Likewise.
21857 * config/riscv/linux.h: Likewise.
21858 * config/riscv/multilib-generator: Likewise.
21859 * config/riscv/peephole.md: Likewise.
21860 * config/riscv/pic.md: Likewise.
21861 * config/riscv/predicates.md: Likewise.
21862 * config/riscv/riscv-builtins.c: Likewise.
21863 * config/riscv/riscv-c.c: Likewise.
21864 * config/riscv/riscv-ftypes.def: Likewise.
21865 * config/riscv/riscv-modes.def: Likewise.
21866 * config/riscv/riscv-opts.h: Likewise.
21867 * config/riscv/riscv-protos.h: Likewise.
21868 * config/riscv/riscv.h: Likewise.
21869 * config/riscv/riscv.md: Likewise.
21870 * config/riscv/riscv.opt: Likewise.
21871 * config/riscv/sync.md: Likewise.
21872 * config/riscv/t-elf-multilib: Likewise.
21873 * config/riscv/t-linux: Likewise.
21874 * config/riscv/t-linux-multilib: Likewise.
21875 * config/riscv/t-riscv: Likewise.
21876 * configure.ac: Likewise.
21877 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21878 Waterman as RISC-V maintainers.
21879 * doc/install.texi: Add RISC-V entries.
21880 * doc/invoke.texi: Add RISC-V options section.
21881 * doc/md.texi: Add RISC-V constraints section.
21882 * configure: Regenerated.
21884 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21887 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21888 false values to be constant vectors with all 0 or all 1 bits set.
21889 (vcondu<mode><mode>): Likewise.
21890 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21892 (fpmask_comparison_operator): Update comment.
21893 (vecint_comparison_operator): New predicate.
21894 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21895 vector conditionals when the true and false values are constant
21896 vectors with all 0 bits or all 1 bits set.
21898 2017-02-06 Martin Sebor <msebor@redhat.com>
21900 PR tree-optimization/79376
21901 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21903 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21905 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21906 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21907 to simplify split condition.
21909 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21911 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21912 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21915 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21917 PR rtl-optimization/68664
21918 * target.def (can_speculate_insn): New hook.
21919 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21920 * doc/tm.texi: Regenerate.
21921 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21922 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21923 (rs6000_sched_can_speculate_insn): New function.
21925 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21927 PR tree-optimization/79284
21928 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21929 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21930 vectorizable_mask_load_store, vectorizable_operation,
21931 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21932 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21933 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21934 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21935 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21936 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21937 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21938 is_gimple_assign (stmt). Replace another such test with
21939 is_gimple_assign (stmt).
21941 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21944 * config/avr/avr.c (rtl-iter.h): Include it.
21945 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21946 (avr_legitimate_combined_insn): ...and implementation.
21948 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21950 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21951 * config/s390/s390.c (s390_const_operand_ok)
21952 (s390_canonicalize_comparison, s390_extract_part)
21953 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21954 (s390_contiguous_bitmask_p, s390_rtx_costs)
21955 (legitimize_pic_address): Likewise.
21956 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21957 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21958 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21959 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21960 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21962 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21964 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21965 REGNO($0) == REGNO($1).
21967 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21969 * config/s390/linux.h(SIZE_TYPE): Add comment.
21971 2017-02-06 Julian Brown <julian@codesourcery.com>
21972 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21973 Virendra Pathak <virendra.pathak@broadcom.com>
21975 * config/aarch64/aarch64-cores.def: Change the scheduler
21977 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21978 * config/aarch64/thunderx2t99.md: New file.
21980 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21982 * doc/standards.texi (Go Language): Update link to language
21985 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21987 * tree-eh.c (lower_resx): Sanitize profile.
21988 (cleanup_empty_eh_move_lp): Likewise.
21990 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21993 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21995 * cfgloopmanip.h (loop_version): Update prototype.
21996 * modulo-sched.c (sms_schedule): Update call of loop_version.
21997 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21998 * tree-parloops.c (gen_parallel_loop): Likewise.
21999 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22000 * tree-ssa-loop-split.c (split_loop): Likewise.
22001 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22002 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22004 2017-02-05 Martin Liska <mliska@suse.cz>
22007 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22009 (print_operand_address): Initialize a struct to zero.
22011 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22013 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22014 garbage collector only in textual form.
22016 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22018 * doc/extend.texi (x86 specific memory model extensions for
22019 transactional memory): Simplify a phrase.
22021 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22024 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22025 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22026 (atomic_storedi_1): Likewise.
22028 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22030 PR tree-optimization/79338
22031 * tree-parloops.c (gather_scalar_reductions): Don't call
22032 vect_analyze_loop_form for loop->inner before destroying loop's
22035 2017-02-03 Martin Sebor <msebor@redhat.com>
22037 PR tree-optimization/79327
22038 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22039 when precision has resulted in leading zeros.
22040 (format_integer): Adjust the likely counter to assume an unknown
22041 argument that may be zero is non-zero.
22043 2017-02-03 Jason Merrill <jason@redhat.com>
22046 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22047 avoid copying non-taken branch.
22049 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22051 PR tree-optimization/79340
22052 * tree-vect-loop.c (vectorizable_reduction): Release
22053 vec_defs elements after safe_splicing them into other vectors.
22056 PR tree-optimization/79327
22057 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22058 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22060 (format_integer): Use wide_int_to_tree instead of build_int_cst
22061 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22062 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22063 of shortest and longest sequence.
22065 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22067 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22068 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22070 2017-02-03 Walter Lee <walt@tilera.com>
22073 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22074 after initial stackframe link reg save.
22075 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22077 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22080 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22081 wu for stxssp alternative.
22083 2017-02-03 Martin Sebor <msebor@redhat.com>
22085 PR tree-optimization/79352
22086 * gimple-fold.c (get_range_strlen): Add argument.
22087 (get_range_strlen): Change return type to bool.
22088 (get_maxval_strlen): Pass in a dummy argument.
22089 * gimple-fold.h (get_range_strlen): Change return type to bool.
22090 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22091 * tree.h (array_at_struct_end_p): Add argument.
22092 * tree.c (array_at_struct_end_p): Handle it.
22094 2017-02-03 Martin Liska <mliska@suse.cz>
22097 * multiple_target.c (create_dispatcher_calls): Redirect edge
22098 from a caller of a dispatcher.
22099 (expand_target_clones): Make the clones local.
22100 (ipa_target_clone): Do both target clones and resolvers.
22101 (ipa_dispatcher_calls): Remove the pass.
22102 (pass_dispatcher_calls::gate): Likewise.
22103 (make_pass_dispatcher_calls): Likewise.
22104 * passes.def (pass_target_clone): Put as very first IPA early
22107 2017-02-03 Martin Liska <mliska@suse.cz>
22109 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22110 in case of a function with ifunc attribute.
22112 2017-02-03 Martin Liska <mliska@suse.cz>
22114 * cgraph.c (cgraph_node::dump): Dump function version info.
22115 * symtab.c (symtab_node::dump_base): Add missing new line.
22117 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22119 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22120 (ifcombine_ifandif): Use it.
22122 2017-02-03 Martin Liska <mliska@suse.cz>
22124 * doc/invoke.texi: Document default value for
22125 use-after-scope-direct-emission-threshold.
22127 2017-02-03 Martin Liska <mliska@suse.cz>
22129 PR tree-optimization/79339
22130 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22131 (format_floating): Likewise.
22133 2017-02-03 Martin Liska <mliska@suse.cz>
22136 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22138 (ipa_node_params_t::remove): Likewise.
22139 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22140 initialization from removed ipa_node_params_t::insert.
22141 (ipa_node_params::~ipa_node_params): Move from removed
22142 ipa_node_params_t::release.
22143 * symbol-summary.h (symbol_summary::m_released): New member.
22144 Do not release a summary twice. Do not allow to call finalizer
22145 for types of a summary that live in GGC memory.
22147 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22149 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22152 2017-02-02 Martin Sebor <msebor@redhat.com>
22154 PR middle-end/79275
22155 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22156 (format_string): Tighten up the range of output for non-constant
22157 strings and correct the expected range for wide non-constant strings.
22159 2017-02-02 Martin Sebor <msebor@redhat.com>
22161 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22163 PR middle-end/32003
22164 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22166 (-fdump-tree-@var): Add to index and document how to come up
22167 with pass-specific option and dump file names.
22168 (-fdump-passes): Clarify where to look for output.
22170 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22172 PR middle-end/77445
22173 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22174 statistics of the analyzed path; allow threading for speed when
22175 any of BBs along the path are optimized for speed.
22177 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22179 PR middle-end/78468
22180 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22181 settings of the virtual registers.
22184 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22186 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22187 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22190 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22192 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22193 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22195 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22197 * config/s390/s390.md: Add missing comments with the expanded
22199 * config/s390/vector.md: Likewise.
22200 * config/s390/vx-builtins.md: Likewise.
22202 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22205 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22206 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22207 conditions on a single line.
22209 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22211 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22212 __S390_VX__ to __VX__.
22214 2017-02-01 Andrew Pinski <apinski@cavium.com>
22216 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22217 stmt_info to record_stmt_cost.
22218 (vect_get_known_peeling_cost): Pass stmt_info if known to
22220 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22221 cpu_vector_cost field into
22222 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22223 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22224 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22225 splitting of scalar_stmt_cost and vec_stmt_cost.
22226 (thunderx_vector_cost): Likewise.
22227 (cortexa57_vector_cost): LIkewise.
22228 (exynosm1_vector_cost): Likewise.
22229 (xgene1_vector_cost): Likewise.
22230 (thunderx2t99_vector_cost): Improve after the splitting of the two
22232 (aarch64_builtin_vectorization_cost): Update for the splitting of
22233 scalar_stmt_cost and vec_stmt_cost.
22235 2017-02-01 Torvald Riegel <triegel@redhat.com>
22236 Richard Henderson <rth@redhat.com>
22238 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22239 conditional on existance of a fast atomic load.
22240 * optabs-query.c (can_atomic_load_p): New function.
22241 * optabs-query.h (can_atomic_load_p): Declare it.
22242 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22243 no fast atomic load is available for the particular size of access.
22244 (expand_atomic_compare_and_swap): Likewise.
22245 (expand_atomic_load): Likewise.
22246 (expand_atomic_store): Likewise.
22247 (expand_atomic_fetch_op): Likewise.
22248 * testsuite/lib/target-supports.exp
22249 (check_effective_target_sync_int_128): Remove x86 because it provides
22250 no fast atomic load.
22251 (check_effective_target_sync_int_128_runtime): Likewise.
22253 2017-02-01 Richard Biener <rguenther@suse.de>
22255 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22256 (graphite_transform_loops): Provide opt-info for optimized nests.
22257 * tree-parloop.c (parallelize_loops): Provide opt-info for
22258 parallelized loops.
22260 2017-02-01 Richard Biener <rguenther@suse.de>
22262 PR middle-end/79315
22263 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22264 was not set before.
22266 2017-02-01 Richard Biener <rguenther@suse.de>
22268 PR tree-optimization/71824
22269 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22270 Verify the loops are valid in the merged SESE region.
22271 (scop_detection::can_represent_loop_1): Check analyzing the
22272 evolution of the number of iterations in the region succeeds.
22274 2017-01-31 Ian Lance Taylor <iant@golang.org>
22276 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22277 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22279 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22281 PR preprocessor/79210
22282 * input.c (get_substring_ranges_for_loc): Replace line_width
22283 assertion with error-handling.
22285 2017-01-31 Richard Biener <rguenther@suse.de>
22287 PR tree-optimization/77318
22288 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22289 (create_pw_aff_from_tree): Take loop parameter.
22290 (add_condition_to_pbb): Pass loop of the condition to
22291 create_pw_aff_from_tree.
22293 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22295 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22296 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22298 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22302 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22304 (convert_int_to_float128): Likewise.
22305 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22306 (convert_int_to_float128): Likewise.
22307 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22308 (UNSPEC_IEEE128_CONVERT): Likewise.
22309 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22310 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22311 Use local variables for IBM extended format.
22312 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22313 (fix_trunc<mode>si2_fprs): Likewise.
22314 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22315 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22316 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22317 to know that we can now have integers of all sizes in vector
22319 (fix<uns>_<mode>di2_hw): Likewise.
22320 (float<uns>_<mode>si2_hw): Likewise.
22321 (fix_<mode>si2_hw): Likewise.
22322 (fixuns_<mode>si2_hw): Likewise.
22323 (float<uns>_<mode>di2_hw): Likewise.
22324 (float_<mode>di2_hw): Likewise.
22325 (float_<mode>si2_hw): Likewise.
22326 (floatuns_<mode>di2_hw): Likewise.
22327 (floatuns_<mode>si2_hw): Likewise.
22328 (xscvqp<su>wz_<mode>): Delete, no longer used.
22329 (xscvqp<su>dz_<mode>): Likewise.
22330 (xscv<su>dqp_<mode>): Likewise.
22331 (ieee128_mfvsrd_64bit): Likewise.
22332 (ieee128_mfvsrd_32bit): Likewise.
22333 (ieee128_mfvsrwz): Likewise.
22334 (ieee128_mtvsrw): Likewise.
22335 (ieee128_mtvsrd_64bit): Likewise.
22336 (ieee128_mtvsrd_32bit): Likewise.
22338 2017-01-31 Martin Liska <mliska@suse.cz>
22341 * ipa-prop.c (ipa_free_all_node_params): Call release method
22342 instead of ~sumbol_summary to not to trigger double times
22345 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22347 PR tree-optimization/71691
22348 * bitmap.h (class auto_bitmap): New.
22349 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22350 is_maybe_undefined instead of ssa_undefined_value_p.
22352 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22354 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22355 __S390_ARCH_LEVEL__ to __ARCH__.
22357 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22359 PR tree-optimization/79267
22360 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22361 if should_remove_lhs_p is true.
22363 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22366 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22367 (add_alignment_attribute): New.
22368 (base_type_die): Add alignment attribute.
22369 (subrange_type_die): Likewise.
22370 (modified_type_die): Likewise.
22371 (gen_array_type_die): Likewise.
22372 (gen_descr_array_type_die: Likewise.
22373 (gen_enumeration_type_die): Likewise.
22374 (gen_subprogram_die): Likewise.
22375 (gen_variable_die): Likewise.
22376 (gen_field_die): Likewise.
22377 (gen_ptr_to_mbr_type_die): Likewise.
22378 (gen_struct_or_union_type_die): Likewise.
22379 (gen_subroutine_type_die): Likewise.
22380 (gen_typedef_die): Likewise.
22381 (base_type_cmp): Compare alignment attribute.
22383 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22386 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22387 (setb_unsigned) New pattern for setb with CCUNS.
22388 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22389 subfc./subfe sequence to avoid overflow problems. Generate a
22390 shorter sequence with cmpld/setb for power9.
22391 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22392 for generating subfc. instruction.
22393 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22394 now uses this instruction.
22396 2017-01-30 Ian Lance Taylor <iant@google.com>
22399 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22400 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22402 2017-01-30 Martin Sebor <msebor@redhat.com>
22404 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22405 Move constant to the right of a relational operator.
22406 (get_mpfr_format_length, format_character, format_string): Ditto.
22407 (should_warn_p, maybe_warn): Same.
22409 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22411 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22414 * asan.c (get_translation_unit_decl): Remove function.
22415 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22417 2017-01-30 Martin Liska <mliska@suse.cz>
22419 PR gcov-profile/79259
22420 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22421 -fprofile-generate.
22423 2017-01-30 Martin Liska <mliska@suse.cz>
22426 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22427 Initialize variables with NULL value.
22429 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22432 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22434 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22436 2017-01-30 Richard Biener <rguenther@suse.de>
22438 PR tree-optimization/79276
22439 * tree-vrp.c (process_assert_insertions): Properly adjust common
22440 when removing a duplicate.
22442 * gcc.dg/torture/pr79276.c: New testcase.
22444 2017-01-30 Richard Biener <rguenther@suse.de>
22446 PR tree-optimization/79256
22447 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22448 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22450 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22452 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22455 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22456 ("*r<noxa>sbg_<mode>_sll_bitmask")
22457 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22458 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22459 Use contiguous_bitmask_nowrap_operand.
22461 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22464 * config/rs6000/altivec.h (vec_xl): Revise #define.
22465 (vec_xst): Likewise.
22467 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22469 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22471 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22473 PR rtl-optimization/79194
22474 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22475 traps before call to bypass_conditional_jumps.
22477 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22479 PR tree-optimization/71374
22480 * lra-constraints.c (check_conflict_input_operands): New.
22481 (match_reload): Use it.
22483 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22486 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22487 account to calculate conflict_set.
22489 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22491 PR rtl-optimization/78559
22492 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22493 other_insn in combine.
22495 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22497 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22498 uint16_type_node for BT_UINT16.
22500 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22502 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22503 "RTL Tests" to menu.
22504 (GIMPLE Tests): New node.
22505 (RTL Tests): New node.
22507 2017-01-27 Richard Biener <rguenther@suse.de>
22509 PR tree-optimization/79245
22510 * tree-loop-distribution.c (distribute_loop): Apply cost
22511 modeling also to detected patterns.
22513 2017-01-27 Richard Biener <rguenther@suse.de>
22515 PR tree-optimization/71433
22516 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22517 (compare_assert_loc): New function.
22518 (process_assert_insertions): Sort and optimize assert locations
22519 to remove duplicates and push down identical assertions on
22520 edges to their destination block.
22522 2017-01-27 Richard Biener <rguenther@suse.de>
22524 PR tree-optimization/79244
22525 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22526 out SSA names even if abnormal.
22528 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22530 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22531 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22532 instead of MPFR_RNDN.
22534 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22537 * arm.c (arm_option_override): Don't call build_target_option_node
22538 until after doing all option overrides.
22539 (arm_valid_target_attribute_tree): Likewise.
22541 2017-01-27 Martin Liska <mliska@suse.cz>
22543 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22544 for {cd}tors and C++ {cd}tors.
22546 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22548 * config/s390/s390.md ("*setmem_long_and")
22549 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22551 2017-01-26 Martin Sebor <msebor@redhat.com>
22553 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22556 2017-01-26 Martin Sebor <msebor@redhat.com>
22558 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22559 HAVE_DFmode before using XFmode or DFmode.
22560 (parse_directive): Avoid using the z length modifier to avoid
22561 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22563 PR middle-end/78703
22564 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22565 to accept adjustment as an array.
22566 (get_int_range): New function.
22567 (struct directive): Make width and prec arrays.
22568 (directive::set_width, directive::set_precision): Call get_int_range.
22569 (format_integer, format_floating): Handle width and precision ranges.
22570 (format_string, parse_directive): Same.
22572 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22575 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22576 comdat_type_p set, just clone them, but keep the children in the
22580 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22581 which have direct callers with -fvar-tracking-assignments enabled
22583 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22584 inside of type units.
22586 2017-01-26 Martin Sebor <msebor@redhat.com>
22588 PR middle-end/78703
22589 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22591 (struct format_result): Replace number_chars, number_chars_min,
22592 and number_chars_max with a single member of struct result_range.
22594 (format_result::operator+=): Adjust.
22595 (struct fmtresult): Remove bounded. Handle likely and unlikely
22597 (fmtresult::adjust_for_width_or_precision): New function.
22598 (fmtresult:type_max_digits): New function.
22599 (bytes_remaining): Handle likely and unlikely counters.
22600 (min_bytes_remaining): Remove.
22601 (format_percent): Simplify.
22602 (format_integer, format_floating): Set likely and unlikely counters.
22603 (get_string_length, format_character, format_string): Same.
22604 (format_plain, should_warn_p): New function.
22605 (maybe_warn): Call should_warn_p. Update diagnostic messages
22606 and handle those for all directives, including plain strings.
22607 (format_directive): Handle likely and unlikely counters.
22608 Remove unnecessary quoting from diagnostics. Add an informational
22610 (add_bytes): Remove.
22611 (pass_sprintf_length::compute_format_length): Simplify.
22612 (try_substitute_return_value): Handle likely and unlikely counters.
22614 2017-01-26 Carl Love <cel@us.ibm.com>
22616 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22617 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22619 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22622 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22623 endianess for subregs into account.
22624 * lra-constraints.c (lra_constraints): Do risky transformations
22625 always on the first iteration.
22626 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22627 last_call_used_reg_set.
22628 (process_bb_lives): Define and use last_call_used_reg_set.
22629 * lra.c (lra): Always continue after lra_constraints on the first
22632 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22634 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22636 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22638 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22640 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22641 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22642 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22643 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22644 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22645 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22646 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22647 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22648 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22650 2017-01-26 Marek Polacek <polacek@redhat.com>
22653 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22654 for the third operand.
22656 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22658 PR middle-end/79236
22659 * omp-low.c (struct omp_context): Add simt_stmt field.
22660 (scan_omp_for): Return omp_context *.
22661 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22662 context to the _simt_ SIMD stmt.
22663 (lower_omp_for): For combined SIMD with sibling _simt_
22664 SIMD, make sure to use the same decls in _looptemp_
22665 clauses as in the sibling.
22667 2017-01-26 David Sherwood <david.sherwood@arm.com>
22669 PR middle-end/79212
22670 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22673 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22676 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22677 emit fld b; fld a; if possible.
22679 * brig-builtins.def: Update copyright years.
22680 * config/arm/arm_acle_builtins.def: Update copyright years.
22682 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22685 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22686 constraint instead of o for the stxsd instruction.
22688 2017-01-25 Carl Love <cel@us.ibm.com>
22690 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22691 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22693 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22695 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22697 2017-01-25 Richard Biener <rguenther@suse.de>
22699 PR tree-optimization/69264
22700 * target.def (vector_alignment_reachable): Improve documentation.
22701 * doc/tm.texi: Regenerate.
22702 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22704 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22705 earlier changes with respect to TYPE_USER_ALIGN.
22706 (vector_alignment_reachable_p): Likewise. Improve dumping.
22708 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22711 * config/arm/arm.md (xordi3): Force constant operand into a register
22714 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22716 * doc/invoke.texi (-fstore-merging): Correct default optimization
22717 levels at which it is enabled.
22718 (-O): Move -fstore-merging from list to...
22721 2017-01-25 Richard Biener <rguenther@suse.de>
22724 * omp-expand.c: Include debug.h.
22725 (expand_omp_taskreg): Make sure to generate early debug before
22726 outlining anything from a function.
22727 (expand_omp_target): Likewise.
22728 (grid_expand_target_grid_body): Likewise.
22730 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22733 * asan.c (get_translation_unit_decl): New function.
22734 (asan_add_global): Extract modules file name from globals
22735 TRANSLATION_UNIT_DECL name.
22737 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22740 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22741 for long calls with APCS frame and VFP.
22743 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22745 * cfg.c (original_copy_tables_initialized_p): New function.
22746 * cfg.h (original_copy_tables_initialized_p): New decl.
22747 * cfgrtl.c (relink_block_chain): Guard the call to
22748 free_original_copy_tables with a call to
22749 original_copy_tables_initialized_p.
22750 * cgraph.h (symtab_node::native_rtl_p): New decl.
22751 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22752 (symtab_node::needed_p): Don't assert for early assembly output
22753 for __RTL functions.
22754 (cgraph_node::finalize_function): Set "force_output" for __RTL
22756 (cgraph_node::analyze): Bail out early for __RTL functions.
22757 (analyze_functions): Update assertion to support __RTL functions.
22758 (cgraph_node::expand): Bail out early for __RTL functions.
22759 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22761 * function.h (struct function): Update comment for field
22763 * gimple-expr.c: Include "tree-pass.h".
22764 (gimple_has_body_p): Return false for __RTL functions.
22765 * Makefile.in (OBJS): Add run-rtl-passes.o.
22766 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22768 (gcc::pass_manager::get_clean_slate): New accessor.
22769 * passes.c: Include "insn-addr.h".
22770 (should_skip_pass_p): Add logging. Update logic for running
22771 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22772 property-provider override so it is only done for gimple passes.
22774 (skip_pass): New function.
22775 (execute_one_pass): Call skip_pass when skipping passes.
22776 * read-md.c (md_reader::read_char): Support filtering
22777 the input to a subset of line numbers.
22778 (md_reader::md_reader): Initialize fields
22779 m_first_line and m_last_line.
22780 (md_reader::read_file_fragment): New function.
22781 * read-md.h (md_reader::read_file_fragment): New decl.
22782 (md_reader::m_first_line): New field.
22783 (md_reader::m_last_line): New field.
22784 * read-rtl-function.c (function_reader::create_function): Only
22785 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22786 curr_properties. Set DECL_INITIAL to a dummy block.
22787 (read_rtl_function_body_from_file_range): New function.
22788 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22790 * run-rtl-passes.c: New file.
22791 * run-rtl-passes.h: New file.
22793 2017-01-24 Jeff Law <law@redhat.com>
22795 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
22798 2017-01-24 Bin Cheng <bin.cheng@arm.com>
22800 PR tree-optimization/79159
22801 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
22802 (record_nonwrapping_iv): Improve boundary using above function if no
22803 value range information.
22805 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
22806 Martin Jambor <mjambor@suse.cz>
22808 * brig-builtins.def: New file.
22809 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
22810 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
22811 (DEF_HSAIL_SAT_BUILTIN): Likewise.
22812 (DEF_HSAIL_INTR_BUILTIN): Likewise.
22813 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
22814 * builtin-types.def (BT_INT8): New.
22815 (BT_INT16): Likewise.
22816 (BT_UINT8): Likewise.
22817 (BT_UINT16): Likewise.
22818 (BT_FN_ULONG): Likewise.
22819 (BT_FN_UINT_INT): Likewise.
22820 (BT_FN_UINT_ULONG): Likewise.
22821 (BT_FN_UINT_LONG): Likewise.
22822 (BT_FN_UINT_PTR): Likewise.
22823 (BT_FN_ULONG_PTR): Likewise.
22824 (BT_FN_INT8_FLOAT): Likewise.
22825 (BT_FN_INT16_FLOAT): Likewise.
22826 (BT_FN_UINT32_FLOAT): Likewise.
22827 (BT_FN_UINT16_FLOAT): Likewise.
22828 (BT_FN_UINT8_FLOAT): Likewise.
22829 (BT_FN_UINT64_FLOAT): Likewise.
22830 (BT_FN_UINT16_UINT32): Likewise.
22831 (BT_FN_UINT32_UINT16): Likewise.
22832 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22833 (BT_FN_INT_PTR_INT): Likewise.
22834 (BT_FN_UINT_PTR_UINT): Likewise.
22835 (BT_FN_LONG_PTR_LONG): Likewise.
22836 (BT_FN_ULONG_PTR_ULONG): Likewise.
22837 (BT_FN_VOID_UINT64_UINT64): Likewise.
22838 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22839 (BT_FN_INT8_INT8_INT8): Likewise.
22840 (BT_FN_INT16_INT16_INT16): Likewise.
22841 (BT_FN_INT_INT_INT): Likewise.
22842 (BT_FN_UINT_FLOAT_UINT): Likewise.
22843 (BT_FN_FLOAT_UINT_UINT): Likewise.
22844 (BT_FN_ULONG_UINT_UINT): Likewise.
22845 (BT_FN_ULONG_UINT_PTR): Likewise.
22846 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22847 (BT_FN_UINT_UINT_UINT): Likewise.
22848 (BT_FN_VOID_UINT_PTR): Likewise.
22849 (BT_FN_UINT_UINT_PTR: Likewise.
22850 (BT_FN_UINT32_UINT64_PTR): Likewise.
22851 (BT_FN_INT_INT_UINT_UINT): Likewise.
22852 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22853 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22854 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22855 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22856 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22857 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22858 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22859 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22860 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22861 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22862 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22863 * doc/frontends.texi: List BRIG FE.
22864 * doc/install.texi (Testing): Add BRIG tesring requirements.
22865 * doc/invoke.texi (Overall Options): Mention BRIG.
22866 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22868 2017-01-24 Richard Biener <rguenther@suse.de>
22870 PR translation/79208
22871 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22873 2017-01-24 Martin Jambor <mjambor@suse.cz>
22876 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22877 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22878 and known_contexts.
22880 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22882 PR middle-end/79123
22883 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22884 casts from signed to unsigned really don't have a range.
22886 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22888 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22889 GMP_RNDx for compatiblity.
22891 2017-01-24 Martin Liska <mliska@suse.cz>
22894 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22895 that would prevent us to call alloca with -1 as argument.
22897 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22899 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22900 Avoid -Wformat-security warning.
22902 2017-01-23 Andrew Pinski <apinski@cavium.com>
22904 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22907 2017-01-23 Martin Sebor <msebor@redhat.com>
22909 PR middle-end/78703
22910 * gimple-ssa-sprintf.c (warn_level): New global.
22911 (format_integer): Use it here and throughout the rest of the file.
22912 Use the same switch to compute sign as base.
22913 (maybe_warn): New function.
22914 (format_directive): Factor out warnings into maybe_warn.
22915 Add debugging output. Use warn_level.
22916 (add_bytes): Use warn_level.
22917 (pass_sprintf_length::compute_format_length): Add debugging output.
22918 (try_substitute_return_value): Same.
22919 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22921 PR middle-end/78703
22922 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22923 (struct fmtresult, format_integer, format_floating): Adjust.
22924 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22925 (get_string_length, format_string,format_directive): Same.
22926 (pass_sprintf_length::compute_format_length): Same.
22927 (try_substitute_return_value): Simplify slightly.
22929 PR middle-end/78703
22930 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22931 (fmtresult::operator+=): Outlined.
22932 (struct fmtresult): Add ctors.
22933 (struct conversion_spec): Rename...
22934 (struct directive): ...to this. Add and remove data members.
22935 (directive::set_width, directive::set_precision): New functions.
22936 (format_percent): Use fmtresult ctor.
22937 (get_width_and_precision): Remove.
22938 (format_integer): Make naming changes. Avoid computing width and
22940 (format_floating): Same. Adjust indentation.
22941 (format_character, format_none): New functions.
22942 (format_string): Moved character handling to format_character.
22943 (format_directive): Remove arguments, change return type.
22944 (parse_directive): New function.
22945 (pass_sprintf_length::compute_format_length): Move directive
22946 parsing to parse_directive.
22948 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22950 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22951 (assign_assembler_name_if_needed): ... this.
22952 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22953 (assign_assembler_name_if_needed): ... this.
22954 (free_lang_data_in_cgraph): Adjust callers.
22955 * cgraphunit.c (cgraph_node::analyze): Likewise.
22956 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22959 2017-01-23 Richard Biener <rguenther@suse.de>
22961 PR tree-optimization/79088
22962 PR tree-optimization/79188
22963 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22964 resetting loop bounds after last path deletion. Reset loop
22965 bounds of the target loop, make code match the comments.
22966 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22967 Make sure loops need no fixups.
22969 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22971 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22972 exponent support with double type for first argument.
22973 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22974 type returned by __builtin_vec_extract_sig,
22975 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22976 functions from "vector int" to "vector unsigned int" or from
22977 "vector long long int" to "vector unsigned long long int".
22978 Changed type returned by __builtin_vec_extract_exp,
22979 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22980 functions from "vector int" to "vector unsigned int" or from
22981 "vector long long int" to "vector unsigned long long int".
22982 Changed return type of __builtin_vec_test_data_class,
22983 __builtin_vec_test_data_class_sp, and
22984 __builtin_vec_test_data_class_dp from "vector int" to
22985 "vector bool int" or from "vector long long int" to "vector bool
22986 long long int" and changed second argument type from "unsigned
22987 int" to "int". Added new overloaded function forms "vector float
22988 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22989 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22990 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22991 double, vector unsigned long long int)" and "vector double
22992 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22993 long int)". Changed return type of
22994 __builtin_scalar_test_data_class and
22995 __builtin_scalar_test_data_class_sp and
22996 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22997 int" and changed second argument from "unsigned int" to "int".
22998 Changed type returned by __builtin_scalar_test_neg,
22999 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23000 from "int" to "bool int". Added new overloaded function form
23001 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23002 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23003 exponent double-precision with floating point first argument.
23004 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23005 documentation of scalar_test_data_class, scalar_test_neg,
23006 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23007 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23008 vec_test_data_class built-in functions to reflect refinements in
23009 their type signatures.
23011 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23013 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23015 (aarch64_elf_asm_destructor): Likewise.
23017 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23019 PR rtl-optimization/78634
23020 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23021 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23022 * ifcvt.c (noce_try_cmove): Add missing cost check.
23024 PR rtl-optimization/71724
23025 * combine.c (if_then_else_cond): Look for situations where it is
23026 beneficial to undo the work of one of the recursive calls.
23028 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23030 PR tree-optimization/70754
23031 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23032 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23033 combined stmt before it if not NULL.
23034 (combine_chains): Process refs reversely and compute dominance point
23037 2017-01-23 Martin Liska <mliska@suse.cz>
23039 PR tree-optimization/79196
23040 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23041 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23043 (strlen_optimize_stmt): Call the renamed function.
23045 2017-01-23 Michael Matz <matz@suse.de>
23047 PR tree-optimization/78384
23048 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23050 2017-01-23 Richard Biener <rguenther@suse.de>
23052 PR tree-optimization/79186
23053 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23054 both incoming edges before moving an assert.
23056 2017-01-23 Martin Jambor <mjambor@suse.cz>
23058 * ipa-prop.c (load_from_param_1): Removed.
23059 (load_from_unmodified_param): Bits from load_from_param_1 put back
23061 (load_from_param): Removed.
23062 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23063 with stmt. Reverted back to use of load_from_unmodified_param.
23065 2017-01-23 Martin Jambor <mjambor@suse.cz>
23068 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23069 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23070 field a pointer to garbage collected vector, mark lattices and
23071 ipcp_orig_node with GTY((skip)).
23072 (ipa_get_param_count): Adjust to descriptors being a pointer.
23073 (ipa_get_param): Likewise.
23074 (ipa_get_type): Likewise.
23075 (ipa_get_param_move_cost): Likewise.
23076 (ipa_set_param_used): Likewise.
23077 (ipa_get_controlled_uses): Likewise.
23078 (ipa_set_controlled_uses): Likewise.
23079 (ipa_is_param_used): Likewise.
23080 (ipa_node_params_t): Move into garbage collector. New methods insert
23082 (ipa_node_params_sum): Annotate wth GTY(()).
23083 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23085 (ipa_load_from_parm_agg): Adjust declaration.
23086 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23087 * ipa-profile.c (ipa_profile): Likewise.
23088 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23089 (ipa_populate_param_decls): Make descriptors parameter garbage
23091 (ipa_dump_param): Adjust to descriptors being a pointer.
23092 (ipa_alloc_node_params): Likewise.
23093 (ipa_initialize_node_params): Likewise.
23094 (load_from_param_1): Make descriptors parameter garbage collected.
23095 (load_from_unmodified_param): Likewise.
23096 (load_from_param): Likewise.
23097 (ipa_load_from_parm_agg): Likewise.
23098 (ipa_node_params::~ipa_node_params): Removed.
23099 (ipa_free_all_node_params): Remove call to delete operator.
23100 (ipa_node_params_t::insert): New.
23101 (ipa_node_params_t::remove): Likewise.
23102 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23103 copy known_csts and known_contexts vectors.
23104 (ipa_read_node_info): Adjust to descriptors being a pointer.
23105 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23107 (ipcp_transform_function): Make descriptors variable garbage
23110 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23112 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23113 * config/i386/avx512dqintrin.h: Ditto.
23114 * config/i386/avx512fintrin.h: Ditto.
23115 * config/i386/i386.c: Handle new builtins.
23116 * config/i386/i386-builtin.def: Add new builtins.
23117 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23118 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23120 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23121 Martin Liska <mliska@suse.cz>
23123 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23124 * asan.c (asan_expand_poison_ifn): Support stores and use
23125 appropriate ASAN report function.
23126 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23127 * internal-fn.def (ASAN_POISON_USE): Declare.
23128 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23129 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23130 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23131 ASAN_POISON calls w/o LHS.
23132 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23133 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23134 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23135 * gimplify.c (asan_poison_variables): Add attribute
23136 use_after_scope_memory to variables that really needs to live
23138 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23139 having the attribute.
23141 2017-01-23 Martin Liska <mliska@suse.cz>
23143 * asan.c (create_asan_shadow_var): New function.
23144 (asan_expand_poison_ifn): Likewise.
23145 * asan.h (asan_expand_poison_ifn): New declaration.
23146 * internal-fn.c (expand_ASAN_POISON): Likewise.
23147 * internal-fn.def (ASAN_POISON): New builtin.
23148 * sanopt.c (pass_sanopt::execute): Expand
23149 asan_expand_poison_ifn.
23150 * tree-inline.c (copy_decl_for_dup_finish): Make function
23152 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23153 * tree-ssa.c (is_asan_mark_p): New function.
23154 (execute_update_addresses_taken): Rewrite local variables
23155 (identified just by use-after-scope as addressable) into SSA.
23157 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23159 * doc/install.texi (Specific): opensource.apple.com uses https
23160 now. Remove trailing slash.
23162 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23164 * README.Portability: Remove note on an Irix compatibility issue.
23166 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23168 * gcov.c (INCLUDE_ALGORITHM): Define.
23169 (INCLUDE_VECTOR): Define.
23170 No longer include <vector> and <algorithm> directly.
23172 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23174 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23176 * doc/invoke.texi (Code Gen Options): Ditto.
23178 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23181 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23183 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23185 rtl-optimization/79125
23186 * cprop.c (local_cprop_pass): Handle cases where we make an
23187 unconditional trap.
23189 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23193 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23194 read from, for big endian.
23196 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23198 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23199 register pauth builtins for LP64 only.
23201 2017-01-20 Marek Polacek <polacek@redhat.com>
23204 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23207 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23209 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23211 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23212 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23213 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23215 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23218 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23219 in big-endian mode when they are not a single duplicated value.
23221 2017-01-20 Richard Biener <rguenther@suse.de>
23223 * BASE-VER: Bump to 7.0.1.
23225 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23227 * omp-low.c (omplow_simd_context): New struct. Use it...
23228 (lower_rec_simd_input_clauses): ...here and...
23229 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23230 references to idx, lane, max_vf, is_simt.
23232 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23234 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23237 2017-01-20 Martin Jambor <mjambor@suse.cz>
23239 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23240 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23242 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23243 (GTFILES): Rename hsa.c to hsa-common.c.
23244 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23245 * hsa-dump.c: Likewise.
23246 * hsa-gen.c: Likewise.
23247 * hsa-regalloc.c: Likewise.
23248 * ipa-hsa.c: Likewise.
23249 * omp-expand.c: Likewise.
23250 * omp-low.c: Likewise.
23251 * toplev.c: Likewise.
23253 2017-01-20 Marek Polacek <polacek@redhat.com>
23256 * doc/invoke.texi: Document -Wduplicated-branches.
23257 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23258 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23259 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23260 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23261 return 0 only when not OEP_LEXICOGRAPHIC.
23262 (fold_build_cleanup_point_expr): Use the expression
23263 location when building CLEANUP_POINT_EXPR.
23264 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23265 * tree.c (add_expr): Handle error_mark_node.
23267 2017-01-20 Martin Liska <mliska@suse.cz>
23270 * tree-profile.c (init_ic_make_global_vars): Do not call
23272 (gimple_init_gcov_profiler): Likewise.
23274 2017-01-20 Martin Liska <mliska@suse.cz>
23277 * cgraph.h (maybe_create_reference): Remove argument and
23279 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23281 * ipa-cp.c (create_specialized_node): Likewise.
23282 * symtab.c (symtab_node::maybe_create_reference): Handle
23283 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23285 2017-01-20 Martin Liska <mliska@suse.cz>
23287 * read-rtl-function.c (function_reader::create_function): Use
23288 build_decl instread of build_decl_stat.
23290 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23292 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23293 * config/i386/avx512dqintrin.h: Ditto.
23294 * config/i386/avx512fintrin.h: Ditto.
23295 * config/i386/i386-builtin-types.def: Add new types.
23296 * config/i386/i386.c: Handle new types.
23297 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23298 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23299 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23300 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23301 (__builtin_ia32_kshiftridi): New.
23302 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23304 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23308 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23309 define to rs6000_init_stack_protect_guard.
23310 (rs6000_init_stack_protect_guard): New function.
23312 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23313 Yunqiang Su <yunqiang.su@imgtec.com>
23315 * config.gcc (supported_defaults): Add madd4.
23316 (with_madd4): Add validation.
23317 (all_defaults): Add madd4.
23318 * config/mips/mips.opt (mmadd4): New option.
23319 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23321 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23323 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23324 (ISA_HAS_FUSED_MADD4): Likewise.
23325 * doc/invoke.texi (-mmadd4): Document the new option.
23326 * doc/install.texi (--with-madd4): Document the new option.
23328 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23330 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23331 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23332 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23333 (aarch64_init_pauth_hint_builtins): New.
23334 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23335 (aarch64_expand_builtin): Expand new builtins.
23337 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23339 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23340 * combine-stack-adj.c (no_unhandled_cfa): Handle
23341 REG_CFA_TOGGLE_RA_MANGLE.
23342 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23343 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23344 info for return address signing.
23345 (aarch64_expand_epilogue): Likewise.
23347 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23349 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23350 * config/aarch64/aarch64-protos.h
23351 (aarch64_return_address_signing_enabled): New declaration.
23352 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23354 (aarch64_expand_prologue): Sign return address before it's pushed onto
23356 (aarch64_expand_epilogue): Authenticate return address fetched from
23358 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23359 (aarch64_attributes): New function attributes for "sign-return-address".
23360 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23361 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23362 ("*do_return"): Generate combined instructions according to key index.
23363 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23364 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23366 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23367 * config/aarch64/aarch64.opt (msign-return-address=): New.
23368 * doc/extend.texi (AArch64 Function Attributes): Documents
23369 "sign-return-address=".
23370 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23372 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23374 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23375 overall option summary.
23377 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23379 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23380 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23381 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23382 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23384 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23386 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23387 -mpower9-minmax by default for -mcpu=power9.
23388 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23389 128-bit floating point.
23391 2017-01-20 Alan Modra <amodra@gmail.com>
23393 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23394 optimizing for size.
23396 2017-01-20 Alan Modra <amodra@gmail.com>
23399 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23400 for strcmp and strncmp from corresponding builtin decl.
23402 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23404 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23405 instead of i386/rtems-64.h.
23406 * config/i386/rtems-64.h: Remove.
23408 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23412 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23414 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23416 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23418 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23419 Change int to HOST_WIDE_INT.
23420 * config/aarch64/aarch64-protos.h
23421 (aarch64_simd_gen_const_vector_dup): Likewise.
23422 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23424 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23426 * langhooks-def.h (lhd_type_for_size): New decl.
23427 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23428 * langhooks.c (lhd_type_for_size): New function, taken from
23431 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23433 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23434 define_bypass for CR latency.
23435 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23436 (power9-alu2): Add define_bypass for CR latency.
23438 (power9-mul): Update insn latency.
23439 (power9-mul-compare): Update insn latency, bypass latency and remove
23442 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23444 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23446 * config/aarch64/aarch64.md
23447 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23448 aarch64_nopcrelative_literal_loads.
23449 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23451 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23453 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23454 TARGET_LOONGSON_3A.
23455 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23457 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23460 * config.gcc (supported_defaults): Add lxc1-sxc1.
23461 (with_lxc1_sxc1): Add validation.
23462 (all_defaults): Add lxc1-sxc1.
23463 * config/mips/mips.opt (mlxc1-sxc1): New option.
23464 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23466 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23467 __mips_no_lxc1_sxc1.
23468 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23469 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23470 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23472 2017-01-19 Richard Biener <rguenther@suse.de>
23474 PR tree-optimization/72488
23475 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23476 sure to restore SSA info.
23477 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23479 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23481 PR rtl-optimization/79121
23482 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23483 of the inner type when shifting an extended value.
23485 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23488 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23489 interposable aliases.
23491 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23494 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23495 Use the evmergelohi instruction.
23496 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23497 (mov_sitf_e500_subreg8_2_be): Likewise.
23498 (mov_sitf_e500_subreg12_2_le): Likewise.
23499 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23500 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23501 (mov_sitf_e500_subreg8_2_le): Likewise.
23502 (mov_sitf_e500_subreg12_2_be): Likewise.
23504 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23506 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23507 attribute from vecsimple to vecperm.
23508 (altivec_vbpermq2): Likewise.
23510 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23513 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23515 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23516 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23517 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23518 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23519 case where N arg is SIZE_MAX.
23520 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23521 (cmpstrsi): Add pattern.
23523 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23525 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23526 __builtin_vec_revb builtins.
23527 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23528 built-in functions to support generation of the ISA 3.0 XXBR<x>
23529 vector byte reverse instructions.
23530 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23531 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23532 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23533 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23534 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23535 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23536 (P9V_BUILTIN_VEC_REVB): Likewise.
23537 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23538 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23539 (p9_xxbrq_v16qi): Likewise.
23540 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23541 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23542 (p9_xxbrh_v8hi): Likewise.
23543 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23544 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23545 vec_revb built-in functions.
23547 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23549 PR rtl-optimization/78952
23550 * config/i386/i386.md (any_extract): New code iterator.
23551 (*insvqi_2): Use any_extract for source operand.
23552 (*insvqi_3): Use any_shiftrt for source operand.
23554 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23556 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23558 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23560 2017-01-18 Matthias Klose <doko@ubuntu.com>
23562 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23564 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23566 * config/rs6000/altivec.h (vec_bperm): Change #define.
23567 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23568 (altivec_vbpermq2): New define_insn.
23569 (altivec_vbpermd): Likewise.
23570 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23571 function interface.
23572 (VBPERMD): Likewise.
23573 (VBPERM): New polymorphic function interface.
23574 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23575 Add entries for P9V_BUILTIN_VEC_VBPERM.
23576 * doc/extend.texi: Add interfaces for vec_bperm.
23578 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23580 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23581 first letter of error messages.
23582 (s390_resolve_overloaded_builtin): Likewise.
23583 * config/s390/s390.c (s390_expand_builtin): Likewise.
23584 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23585 (s390_valid_target_attribute_inner_p): Likewise.
23586 * config/s390/s390.md ("tabort"): Likewise.
23588 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23590 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23591 (ISA_AVOID_DIV_HILO): New macro.
23592 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23593 (ISA_HAS_DDIV): Likewise.
23595 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23597 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23599 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23601 * doc/invoke.texi (fabi-version): Spelling fix.
23603 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23606 * doc/invoke.texi (fabi-version): Mention mangling fix for
23609 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23612 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23614 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23617 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23620 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23622 (stack_protect_test): Ditto.
23623 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23624 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23626 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23627 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23628 -mstack-protector-guard-offset=.
23629 (RS/6000 and PowerPC Options): Ditto.
23631 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23633 * config/i386/i386.h (MASK_CLASS_P): New define.
23634 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23635 there are no registers from different register sets also when
23636 mask registers are used. Update function comment.
23637 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23638 to (*k/*r) and (*k/*km) alternatives.
23640 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23642 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23643 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23644 (EH_RETURN_HANDLER_RTX): New define.
23645 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23646 Force frame pointer in EH return functions.
23647 (aarch64_expand_epilogue): Add barrier for eh_return.
23648 (aarch64_final_eh_return_addr): Remove.
23649 (aarch64_eh_return_handler_rtx): New function.
23650 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23652 (aarch64_eh_return_handler_rtx): New prototype.
23654 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23656 * config/rs6000/altivec.h (vec_rlmi): New #define.
23657 (vec_vrlnm): Likewise.
23658 (vec_rlnm): Likewise.
23659 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23660 (UNSPEC_VRLNM): Likewise.
23661 (VIlong): New mode iterator.
23662 (altivec_vrl<VI_char>mi): New define_insn.
23663 (altivec_vrl<VI_char>nm): Likewise.
23664 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23666 (VRLDNM): Likewise.
23667 (RLNM): New polymorphic function entry.
23668 (VRLWMI): New monomorphic function entry.
23669 (VRLDMI): Likewise.
23670 (RLMI): New polymorphic function entry.
23671 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23672 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23673 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23676 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23679 * dwarf2out.c (field_byte_offset): Restore the
23680 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23681 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23682 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23685 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23688 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23690 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23693 * dwarf2out.c (add_data_member_location_attribute): For constant
23694 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23695 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23696 DW_AT_byte_size attributes.
23698 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23700 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23701 after forcing to constant memory when the code model is medium.
23703 2017-01-17 Julia Koval <julia.koval@intel.com>
23706 * config/i386/avx512fintrin.h
23707 (_mm512_i32gather_ps): Change __addr type to void const*.
23708 (_mm512_mask_i32gather_ps): Ditto.
23709 (_mm512_i32gather_pd): Ditto.
23710 (_mm512_mask_i32gather_pd): Ditto.
23711 (_mm512_i64gather_ps): Ditto.
23712 (_mm512_mask_i64gather_ps): Ditto.
23713 (_mm512_i64gather_pd): Ditto.
23714 (_mm512_mask_i64gather_pd): Ditto.
23715 (_mm512_i32gather_epi32): Ditto.
23716 (_mm512_mask_i32gather_epi32): Ditto.
23717 (_mm512_i32gather_epi64): Ditto.
23718 (_mm512_mask_i32gather_epi64): Ditto.
23719 (_mm512_i64gather_epi32): Ditto.
23720 (_mm512_mask_i64gather_epi32): Ditto.
23721 (_mm512_i64gather_epi64): Ditto.
23722 (_mm512_mask_i64gather_epi64): Ditto.
23723 (_mm512_i32scatter_ps): Change __addr type to void*.
23724 (_mm512_mask_i32scatter_ps): Ditto.
23725 (_mm512_i32scatter_pd): Ditto.
23726 (_mm512_mask_i32scatter_pd): Ditto.
23727 (_mm512_i64scatter_ps): Ditto.
23728 (_mm512_mask_i64scatter_ps): Ditto.
23729 (_mm512_i64scatter_pd): Ditto.
23730 (_mm512_mask_i64scatter_pd): Ditto.
23731 (_mm512_i32scatter_epi32): Ditto.
23732 (_mm512_mask_i32scatter_epi32): Ditto.
23733 (_mm512_i32scatter_epi64): Ditto.
23734 (_mm512_mask_i32scatter_epi64): Ditto.
23735 (_mm512_i64scatter_epi32): Ditto.
23736 (_mm512_mask_i64scatter_epi32): Ditto.
23737 (_mm512_i64scatter_epi64): Ditto.
23738 (_mm512_mask_i64scatter_epi64): Ditto.
23739 * config/i386/avx512pfintrin.h
23740 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23741 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23742 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23743 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23744 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23745 (_mm512_prefetch_i32scatter_ps): Ditto.
23746 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23747 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23748 (_mm512_prefetch_i64scatter_pd): Ditto.
23749 (_mm512_prefetch_i64scatter_ps): Ditto.
23750 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23751 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23752 * config/i386/avx512vlintrin.h
23753 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23754 (_mm_mmask_i32gather_ps): Ditto.
23755 (_mm256_mmask_i32gather_pd): Ditto.
23756 (_mm_mmask_i32gather_pd): Ditto.
23757 (_mm256_mmask_i64gather_ps): Ditto.
23758 (_mm_mmask_i64gather_ps): Ditto.
23759 (_mm256_mmask_i64gather_pd): Ditto.
23760 (_mm_mmask_i64gather_pd): Ditto.
23761 (_mm256_mmask_i32gather_epi32): Ditto.
23762 (_mm_mmask_i32gather_epi32): Ditto.
23763 (_mm256_mmask_i32gather_epi64): Ditto.
23764 (_mm_mmask_i32gather_epi64): Ditto.
23765 (_mm256_mmask_i64gather_epi32): Ditto.
23766 (_mm_mmask_i64gather_epi32): Ditto.
23767 (_mm256_mmask_i64gather_epi64): Ditto.
23768 (_mm_mmask_i64gather_epi64): Ditto.
23769 (_mm256_i32scatter_ps): Change __addr type to void*.
23770 (_mm256_mask_i32scatter_ps): Ditto.
23771 (_mm_i32scatter_ps): Ditto.
23772 (_mm_mask_i32scatter_ps): Ditto.
23773 (_mm256_i32scatter_pd): Ditto.
23774 (_mm256_mask_i32scatter_pd): Ditto.
23775 (_mm_i32scatter_pd): Ditto.
23776 (_mm_mask_i32scatter_pd): Ditto.
23777 (_mm256_i64scatter_ps): Ditto.
23778 (_mm256_mask_i64scatter_ps): Ditto.
23779 (_mm_i64scatter_ps): Ditto.
23780 (_mm_mask_i64scatter_ps): Ditto.
23781 (_mm256_i64scatter_pd): Ditto.
23782 (_mm256_mask_i64scatter_pd): Ditto.
23783 (_mm_i64scatter_pd): Ditto.
23784 (_mm_mask_i64scatter_pd): Ditto.
23785 (_mm256_i32scatter_epi32): Ditto.
23786 (_mm256_mask_i32scatter_epi32): Ditto.
23787 (_mm_i32scatter_epi32): Ditto.
23788 (_mm_mask_i32scatter_epi32): Ditto.
23789 (_mm256_i32scatter_epi64): Ditto.
23790 (_mm256_mask_i32scatter_epi64): Ditto.
23791 (_mm_i32scatter_epi64): Ditto.
23792 (_mm_mask_i32scatter_epi64): Ditto.
23793 (_mm256_i64scatter_epi32): Ditto.
23794 (_mm256_mask_i64scatter_epi32): Ditto.
23795 (_mm_i64scatter_epi32): Ditto.
23796 (_mm_mask_i64scatter_epi32): Ditto.
23797 (_mm256_i64scatter_epi64): Ditto.
23798 (_mm256_mask_i64scatter_epi64): Ditto.
23799 (_mm_i64scatter_epi64): Ditto.
23800 (_mm_mask_i64scatter_epi64): Ditto.
23801 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
23802 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
23803 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
23804 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
23805 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
23806 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
23807 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
23808 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
23809 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
23810 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
23811 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
23812 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
23813 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
23814 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
23815 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
23816 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
23817 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
23818 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
23819 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
23820 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
23821 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
23822 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
23823 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
23824 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23825 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23826 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23827 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23828 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23829 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23830 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23831 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23832 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23833 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23834 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23835 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23836 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23837 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23838 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23839 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23840 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23841 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23842 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23843 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23844 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23845 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23846 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23847 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23848 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23849 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23850 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23851 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23852 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23853 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23854 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23855 definitions accordingly.
23857 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23858 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23861 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23864 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23867 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23868 hard regs for inner regno.
23870 2017-01-17 Martin Liska <mliska@suse.cz>
23873 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23874 assumption and add comment.
23876 2017-01-17 Nathan Sidwell <nathan@acm.org>
23878 * ipa-visibility.c (localize_node): New function, broken out of ...
23879 (function_and_variable_visibility): ... here. Call it.
23881 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23883 PR middle-end/77445
23884 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23885 correctly set frequency of oudgoing edge.
23886 (duplicate_thread_path): Fix profile updating.
23888 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23891 * configure.ac: Add GCC_BASE_VER.
23892 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23893 version from BASE-VER file.
23894 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23895 (gcc.o): Depend on $(BASEVER).
23896 * common.opt (dumpfullversion): New option.
23897 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23898 * doc/invoke.texi: Document -dumpfullversion.
23899 * doc/install.texi: Document --with-gcc-major-version-only.
23900 * configure: Regenerated.
23902 2017-01-17 Richard Biener <rguenther@suse.de>
23904 PR tree-optimization/71433
23905 * tree-vrp.c (register_new_assert_for): Merge same asserts
23906 on all incoming edges.
23907 (process_assert_insertions_for): Handle insertions at the
23910 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23912 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23913 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23915 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23918 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23921 2017-01-17 Alan Modra <amodra@gmail.com>
23924 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23925 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23926 symbolic stack limit when pic.
23928 2017-01-16 Martin Sebor <msebor@redhat.com>
23930 PR tree-optimization/78608
23931 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23933 2017-01-16 Jeff Law <law@redhat.com>
23936 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23937 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23938 for several include directories that may be relative to sysroot.
23939 * config/i386/x-mingw32 (gplus_includedir): Define.
23940 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23941 (native_system_includedir): Likewise.
23942 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23943 override if TARGET_SYSTEM_ROOT is defined.
23944 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23946 PR tree-optimization/79090
23947 PR tree-optimization/33562
23948 PR tree-optimization/61912
23949 PR tree-optimization/77485
23950 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23951 and computed trims into the dump file.
23953 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23955 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23957 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23960 * gimplify.c (gimplify_init_constructor): If want_value and
23961 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23965 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23966 sequence. Formatting fixes.
23967 (doloop_optimize): Formatting fixes.
23970 * gcc.c (debug_level_greater_than_spec_func): New function.
23971 (static_spec_functions): Add debug-level-gt spec function.
23972 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23974 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23975 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23976 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23977 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23978 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23979 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23981 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23983 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23984 QImode fixups to general and mask registers only.
23986 2017-01-16 Carl Love <cel@us.ibm.com>
23988 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23989 for built-in functions
23990 vector signed char vec_nabs (vector signed char)
23991 vector signed short vec_nabs (vector signed short)
23992 vector signed int vec_nabs (vector signed int)
23993 vector signed long long vec_nabs (vector signed long long)
23994 vector float vec_nabs (vector float)
23995 vector double vec_nabs (vector double)
23996 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23998 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23999 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24000 * doc/extend.texi: Update the documentation file for the new built-in
24003 2017-01-16 Martin Sebor <msebor@redhat.com>
24005 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24008 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24010 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24011 UNSPEC_VSX__XXSPLTD to require special splat handling.
24013 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24016 * system.h: Poison strndup.
24018 2017-01-16 Alan Modra <amodra@gmail.com>
24021 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24024 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24026 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24028 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24030 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24031 call recog here. Assert that INSN_CODE (insn) is non-negative.
24033 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24036 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24038 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24039 in the currently scheduled RTL fragment.
24041 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24043 PR rtl-optimization/78751
24044 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24047 2017-01-14 Jeff Law <law@redhat.com>
24049 PR tree-optimization/79090
24050 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24051 variable length stores.
24052 (compute_trims): Delete dead assignment to *trim_tail.
24053 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24056 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24058 PR rtl-optimization/78626
24059 PR rtl-optimization/78727
24060 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24061 of a block, and split such blocks after everything else is finished.
24063 2017-01-14 Alan Modra <amodra@gmail.com>
24066 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24067 target legitimate_combined_insn.
24068 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24069 (rs6000_legitimate_combined_insn): New function.
24070 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24072 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24073 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24074 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24076 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24078 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24080 2017-01-13 Jeff Law <law@redhat.com>
24082 PR tree-optimization/33562
24083 PR tree-optimization/61912
24084 PR tree-optimization/77485
24085 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24087 (delete_dead_assignment): Likewise.
24088 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24089 statement to delete_dead_call and delete_dead_assignment.
24091 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24094 * substring-locations.c (format_warning_va): Strengthen case 1 so
24095 that both endpoints of the substring must be within the format
24096 range for just the substring to be printed.
24098 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24100 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24101 * config/i386/i386.c (ix86_target_string): Add missing options
24102 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24103 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24104 flags_other and ix86_target_other to flags2_other. Display unknown
24106 (ix86_valid_target_attribute_inner_p): Add missing options and
24107 reorder options by implied ISAs, as in ix86_target_string.
24109 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24111 * hash-table.h (hash_table::too_empty_p): New function.
24112 (hash_table::expand): Use it.
24113 (hash_table::traverse): Likewise.
24114 (hash_table::empty_slot): Use sizeof (value_type) instead of
24115 sizeof (PTR) to convert bytes to elements. Shrink the table
24116 if the current size is excessive for the current number of
24119 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24121 * ira-costs.c (record_reg_classes): Break from the inner loop
24122 early once alt_fail is known to be true. Update outer loop
24123 handling accordingly.
24125 2017-01-13 Jeff Law <law@redhat.com>
24127 * tree-ssa-dse.c (decrement_count): New function.
24128 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24129 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24130 when we know the partially dead statement is a mem* function.
24132 PR tree-optimization/61912
24133 PR tree-optimization/77485
24134 * tree-ssa-dse.c: Include expr.h.
24135 (maybe_trim_constructor_store): New function.
24136 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24138 PR tree-optimization/33562
24139 PR tree-optimization/61912
24140 PR tree-optimization/77485
24141 * doc/invoke.texi: Document new dse-max-object-size param.
24142 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24143 * tree-ssa-dse.c: Include params.h.
24144 (dse_store_status): New enum.
24145 (initialize_ao_ref_for_dse): New, partially extracted from
24147 (valid_ao_ref_for_dse, normalize_ref): New.
24148 (setup_live_bytes_from_ref, compute_trims): Likewise.
24149 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24150 (maybe_trim_partially_dead_store): Likewise.
24151 (maybe_trim_complex_store): Likewise.
24152 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24153 Track what bytes live from the original store. Return tri-state
24154 for dead, partially dead or live.
24155 (dse_dom_walker): Add constructor, destructor and new private members.
24156 (delete_dead_call, delete_dead_assignment): New extracted from
24158 (dse_optimize_stmt): Make a member of dse_dom_walker.
24159 Use initialize_ao_ref_for_dse.
24161 PR tree-optimization/33562
24162 PR tree-optimization/61912
24163 PR tree-optimization/77485
24164 * sbitmap.h (bitmap_count_bits): Prototype.
24165 (bitmap_clear_range, bitmap_set_range): Likewise.
24166 * sbitmap.c (bitmap_clear_range): New function.
24167 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24169 2017-01-13 Martin Liska <mliska@suse.cz>
24172 * function.c (set_cfun): Add new argument force.
24173 * function.h (set_cfun): Likewise.
24174 * ipa-inline-transform.c (inline_call): Use the function when
24175 strict alising from is dropped for function we inline to.
24177 2017-01-13 Richard Biener <rguenther@suse.de>
24179 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24180 for dumping GIMPLE INTEGER_CSTs.
24182 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24184 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24185 to 201112L since C++17.
24187 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24190 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24191 if -fsanitize=kernel-address is present.
24193 2017-01-13 Richard Biener <rguenther@suse.de>
24195 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24196 as _Literal ( type ) number in case usual suffixes do not
24197 preserve all information.
24199 2017-01-13 Richard Biener <rguenther@suse.de>
24201 PR tree-optimization/77283
24202 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24203 and ssa-iterators.h.
24204 (is_feasible_trace): Implement a cost model based on joiner
24207 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24210 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24211 char or short to __float128/_Float128 directly.
24213 2017-01-12 Martin Sebor <msebor@redhat.com>
24215 to -Wformat-overflow.
24216 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24217 (min_bytes_remaining): Same.
24218 (get_string_length): Same.
24219 (format_string): Same.
24220 (format_directive): Same.
24222 (pass_sprintf_length::handle_gimple_call): Same.
24224 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24226 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24227 info.nowrite calls with no lhs that can't throw. Return bool
24228 whether gsi_remove has been called or not.
24229 (pass_sprintf_length::handle_gimple_call): Return bool whether
24230 try_substitute_return_value called gsi_remove. Formatting fix.
24231 (pass_sprintf_length::execute): Don't use gsi_remove if
24232 handle_gimple_call returned true.
24235 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24236 be removed due to side-effects, don't remove following barrier nor
24237 turn the successor edge into fallthru edge.
24239 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24242 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24243 element-reversing loads and stores as not swappable.
24245 2017-01-12 Nathan Sidwell <nathan@acm.org>
24246 Nicolai Stange <nicstange@gmail.com>
24248 * combine.c (try_combine): Don't ignore result of overlap checking
24249 loop. Combine overlap & asm check into single loop.
24251 2017-01-12 Richard Biener <rguenther@suse.de>
24253 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24254 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24256 2017-01-12 Richard Biener <rguenther@suse.de>
24258 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24259 and TS_TARGET_OPTION directly derive from TS_BASE.
24260 * tree-core.h (tree_optimization_option): Derive from tree_base.
24261 (tree_target_option): Likewise.
24263 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24265 * config/i386/i386.c (memory_address_length): Increase len
24266 only when rip_relative_addr_p returns false.
24268 2017-01-11 Julia Koval <julia.koval@intel.com>
24270 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24271 (OPTION_MASK_ISA_SGX_SET): New.
24272 (ix86_handle_option): Handle OPT_msgx.
24273 * config.gcc: Added sgxintrin.h.
24274 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24275 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24276 * config/i386/i386.c (ix86_target_string): Add -msgx.
24278 (ix86_option_override_internal): Handle new options.
24279 (ix86_valid_target_attribute_inner_p): Add sgx.
24280 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24281 * config/i386/i386.opt: Add msgx.
24282 * config/i386/sgxintrin.h: New file.
24283 * config/i386/x86intrin.h: Add sgxintrin.h.
24285 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24288 * fold-const.c (maybe_nonzero_address): Return 1 for function
24290 (tree_single_nonzero_warnv_p): Don't handle function local objects
24294 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24297 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24300 * opts.c: Include "spellcheck.h"
24301 (struct string_fragment): New struct.
24302 (struct edit_distance_traits<const string_fragment &>): New
24304 (get_closest_sanitizer_option): New function.
24305 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24307 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24309 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24311 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24312 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24313 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24314 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24315 for initial die_offset if dwarf_split_debug_info.
24316 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24317 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24319 (output_skeleton_debug_sections): Formatting fix. Use
24320 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24321 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24323 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24325 * config/arm/cortex-a53.md: Add bypasses for
24326 cortex_a53_r2f_cvt.
24327 (cortex_a53_r2f): Only use for transfers.
24328 (cortex_a53_f2r): Likewise.
24329 (cortex_a53_r2f_cvt): Add reservation for conversions.
24330 (cortex_a53_f2r_cvt): Likewise.
24332 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24334 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24335 to all inlined functions, change static to extern.
24337 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24340 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24342 (arm_assemble_integer): Likewise.
24344 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24346 * config.gcc: Use new awk script to check CPU, FPU and architecture
24347 parameters for --with-... options.
24348 * config/arm/parsecpu.awk: New file
24349 * config/arm/arm-cpus.in: New file.
24350 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24352 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24354 * config/arm/t-arm: Update dependency rules.
24355 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24356 of processing .def files.
24357 * config/arm/genopt.sh: Deleted.
24358 * config/arm/gentune.sh: Deleted.
24359 * config/arm/arm-cores.def: Deleted.
24360 * config/arm/arm-arches.def: Deleted.
24361 * config/arm/arm-fpus.def: Deleted.
24362 * config/arm/arm-tune.md: Regenerated.
24363 * config/arm/arm-tables.opt: Regenerated.
24364 * config/arm/arm-cpu.h: New generated file.
24365 * config/arm/arm-cpu-data.h: New generated file.
24366 * config/arm/arm-cpu-cdata.h: New generated file.
24368 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24371 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24373 (input_varpool_node): Unpack dynamically_initialized bit.
24375 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24377 PR rtl-optimization/79032
24378 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24379 the alignment of the adjusted memory reference against that of MODE,
24380 instead of the alignment of the original memory reference.
24382 2017-01-11 Martin Jambor <mjambor@suse.cz>
24384 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24386 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24387 decorated functions.
24389 2017-01-11 Richard Biener <rguenther@suse.de>
24391 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24392 set range/nonnull info for PHI results. Do not set it on
24393 stmts marked for removal.
24395 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24397 * expr.c (store_field): In the bitfield case, fetch the return value
24398 from the registers before applying a single big-endian adjustment.
24399 Always do a final load for a BLKmode value not larger than a word.
24401 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24404 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24405 that we correctly handle column numbers greater than
24406 LINE_MAP_MAX_COLUMN_NUMBER.
24408 2017-01-10 Martin Sebor <msebor@redhat.com>
24410 PR middle-end/78245
24411 * gimple-ssa-sprintf.c (get_destination_size): Call
24412 {init,fini}object_sizes.
24413 * tree-object-size.c (addr_object_size): Adjust.
24414 (pass_through_call): Adjust.
24415 (pass_object_sizes::execute): Adjust.
24416 * tree-object-size.h (fini_object_sizes): Declare.
24418 2017-01-10 Martin Sebor <msebor@redhat.com>
24420 PR tree-optimization/78775
24421 * builtins.c (get_size_range): Move...
24422 * calls.c: ...to here.
24423 (alloc_max_size): Accept zero argument.
24424 (operand_signed_p): Remove.
24425 (maybe_warn_alloc_args_overflow): Call get_size_range.
24426 * calls.h (get_size_range): Declare.
24428 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24430 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24431 from TI's devices.csv file as of September 2016.
24432 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24434 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24436 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24437 * doc/invoke.texi: Likewise.
24438 * doc/md.texi: Likewise.
24439 * doc/objc.texi: Likewise.
24441 2017-01-10 Joshua Conner <joshconner@google.com>
24443 * config/arm/fuchsia-elf.h: New file.
24444 * config/fuchsia.h: New file.
24445 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24446 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24448 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24450 2016-01-10 Richard Biener <rguenther@suse.de>
24452 PR tree-optimization/79034
24453 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24454 Propagate out degenerate PHIs in the joiner.
24456 2017-01-10 Martin Liska <mliska@suse.cz>
24458 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24459 (sort_congruence_classes_by_decl_uid): Likewise.
24460 (sort_congruence_class_groups_by_decl_uid): Likewise.
24461 (sem_item_optimizer::merge_classes): Sort class, groups in these
24462 classes and members in the groups by DECL_UID of declarations.
24463 This would make merge operations stable.
24465 2017-01-10 Martin Liska <mliska@suse.cz>
24467 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24468 usage of m_classes_vec.
24469 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24470 (sem_item_optimizer::get_group_by_hash): Likewise.
24471 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24472 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24473 (sem_item_optimizer::verify_classes): Likewise.
24474 (sem_item_optimizer::process_cong_reduction): Likewise.
24475 (sem_item_optimizer::dump_cong_classes): Likewise.
24476 (sem_item_optimizer::merge_classes): Likewise.
24477 * ipa-icf.h (congruence_class_hash): Rename from
24478 congruence_class_group_hash. Remove declaration of m_classes_vec.
24480 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24482 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24483 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24484 * config.gcc: Add avx512vpopcntdqintrin.h.
24485 * config/i386/avx512vpopcntdqintrin.h: New.
24486 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24487 * config/i386/i386-builtin-types.def: Add new types.
24488 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24489 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24490 __builtin_ia32_vpopcountq_v8di_mask): New.
24491 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24492 __AVX512VPOPCNTDQ__.
24493 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24494 (PTA_AVX512VPOPCNTDQ): Define.
24495 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24496 TARGET_AVX512VPOPCNTDQ_P): Define.
24497 * config/i386/i386.opt: Add mavx512vpopcntdq.
24498 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24499 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24501 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24503 PR middle-end/77484
24504 * predict.def (PRED_CALL): Set to 67.
24506 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24508 * expr.c (store_field): In the bitfield case, if the value comes from
24509 a function call and is of an aggregate type returned in registers, do
24510 not modify the field mode; extract the value in all cases if the mode
24511 is BLKmode and the size is not larger than a word.
24513 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24516 * config/i386/cpuid.h: Fix undefined behavior.
24518 2017-01-04 Jeff Law <law@redhat.com>
24520 PR tree-optimization/79007
24521 PR tree-optimization/67955
24522 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24523 conservative for pt.null when flag_non_call_exceptions is on.
24525 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24527 PR translation/79019
24528 PR translation/79020
24529 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24530 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24531 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24533 * config/avr/avr.opt (maccumulate-args): Likewise.
24534 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24535 * common.opt (freport-bug): Likewise.
24536 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24537 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24538 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24539 translatable string.
24540 * config/i386/i386.c (function_value_32): Likewise.
24541 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24542 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24544 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24545 * common/config/msp430/msp430-common.c (msp430_handle_option):
24547 * symtab.c (symtab_node::verify_base): Likewise.
24548 * opts.c (set_debug_level): Likewise.
24549 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24550 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24551 missing whitespace to translatable strings.
24552 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24553 * config/sh/superh.h: Likewise.
24554 * config/i386/xopintrin.h: Likewise.
24555 * config/i386/znver1.md: Likewise.
24556 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24557 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24558 * double-int.h (struct double_int): Likewise.
24559 * double-int.c (div_and_round_double): Likewise.
24560 * wide-int.cc: Likewise.
24561 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24562 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24563 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24564 (crossjumps_occurred): ... this.
24565 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24568 PR tree-optimization/78899
24569 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24570 returning bool return struct loop *, NULL for failure and the new
24572 (versionable_outer_loop_p): Don't version outer loop if it has
24573 dont_vectorized bit set.
24574 (tree_if_conversion): When versioning outer loop, ensure
24575 tree_if_conversion is performed also on the inner loop of the
24576 non-vectorizable outer loop copy.
24577 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24578 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24579 prevent vectorization of it.
24580 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24581 the outer loop vectorization of the non-scalar version is attempted
24582 before vectorization of the inner loop in scalar version. If
24583 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24584 vectorization of its inner loop.
24585 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24586 has 2 inner loops, rename also on edges from bb whose single pred
24587 is outer_loop->header. Fix typo in function comment.
24589 2017-01-09 Martin Sebor <msebor@redhat.com>
24592 * asan.c (asan_emit_stack_protection): Increase local buffer size
24593 to avoid snprintf truncation warning.
24595 2017-01-09 Andrew Pinski <apinski@cavium.com>
24597 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24598 to reference thunderx2t99 for the tuning structure
24599 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24601 (thunderx2t99_extra_costs): This.
24602 * config/aarch64/aarch64-tune.md: Regenerate.
24603 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24604 (vulcan_addrcost_table): This.
24605 (vulcan_regmove_cost): Rename to ...
24606 (thunderx2t99_regmove_cost): This.
24607 (vulcan_vector_cost): Rename to ...
24608 (thunderx2t99_vector_cost): this.
24609 (vulcan_branch_cost): Rename to ...
24610 (thunderx2t99_branch_cost): This.
24611 (vulcan_tunings): Rename to ...
24612 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24613 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24615 2017-01-09 Martin Jambor <mjambor@suse.cz>
24619 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24620 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24621 (propagate_vr_accross_jump_function): Use the above function for all
24622 value range computations for pass-through jump functions and type
24623 converasion from explicit value range values.
24624 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24625 parameters from TYPE_ARG_TYPES.
24626 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24627 (ipa_write_node_info): Stream type of the actual argument.
24628 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24630 2017-01-09 Martin Liska <mliska@suse.cz>
24633 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24634 (lookup_compiler): Do not show error message with have_E.
24636 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24638 PR tree-optimization/78938
24639 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24640 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24641 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24642 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24645 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24647 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24650 2017-01-09 Richard Biener <rguenther@suse.de>
24652 PR tree-optimization/78997
24653 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24654 name condition properly.
24656 2017-01-09 Richard Biener <rguenther@suse.de>
24659 * dwarf2out.c (is_cxx): New overload with context.
24660 (is_naming_typedef_decl): Use it.
24662 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24664 * invoke.texi (Option Summary): Correct spacing in option lists
24665 and add line breaks to fix over-long lines.
24667 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24669 PR middle-end/17660
24671 * extend.texi (Common Variable Attributes): Add xref to GCC
24672 Internals manual to explain mode attribute keywords.
24674 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24677 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24678 and Preprocessor Options.
24679 (Options for Linking): Document -pthread here....
24680 (RS/6000 and PowerPC Options): ...not here.
24681 (Solaris 2 Options): ...or here.
24682 * doc/cppopts.texi: Document -pthread.
24684 2017-01-08 Martin Sebor <msebor@redhat.com>
24686 PR middle-end/77708
24687 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24688 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24689 New member functions.
24690 (format_directive): Used them.
24692 (pass_sprintf_length::handle_gimple_call): Same.
24693 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24694 to avoid truncation for any argument.
24695 (extract_affine_mul): Same.
24696 * tree.c (get_file_function_name): Same.
24698 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24700 PR middle-end/77484
24701 * predict.def (PRED_INDIR_CALL): Set to 86.
24703 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24705 PR preprocessor/54124
24706 * doc/cppopts.texi: Reformat -d subtable to list the full name
24707 of the options. Add cross-reference to the docs for the general
24708 compiler -d options.
24709 * doc/invoke.texi (Developer Options): Add cross-reference to the
24710 preprocessor-specific -d option documentation.
24712 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24714 PR preprocessor/13498
24715 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24716 redudant material, and reflect new command-line options.
24717 (System Headers): Likewise.
24719 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24721 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24722 -isystem, and -idirafter. Copy-edit.
24723 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24724 default for -ftrack-macro-expansion. Delete obsolete and
24725 badly-formatted implementation details about -fdebug-cpp output.
24726 * doc/cppwarnopts.texi: Copy-edit.
24728 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24731 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24732 that the transition from a max line width >= 1<<10 to narrower
24733 lines works correctly.
24735 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24737 * doc/options.texi (PerFunction): New.
24738 * opt-functions.awk (switch_flags): Map both Optimization and
24739 PerFunction to CL_OPTIMIZATION.
24740 * opth-gen.awk: Test for PerFunction flag along with
24742 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24743 it only when the latter is present. Skip those that don't in
24744 the hash function generator.
24745 * common.opt (fvar-tracking): Mark as PerFunction instead of
24747 (fvar-tracking-assignments): Likewise.
24748 (fvar-tracking-assignments-toggle): Likewise.
24749 (fvar-tracking-uninit): Likewise.
24751 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24753 PR translation/79018
24754 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24757 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24760 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24761 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24762 TARGET_LONG_JUMP_TABLE_OFFSETS.
24763 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24764 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24765 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24766 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24767 * config/m68k/m68k.md (tablejump expander): Likewise.
24768 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24769 TARGET_LONG_JUMP_TABLE_OFFSETS.
24770 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24771 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24773 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24774 David Holsgrove <david.holsgrove@xilinx.com>
24776 * common/config/microblaze/microblaze-common.c
24777 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24778 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24780 * config/microblaze/microblaze.c (microblaze_must_save_register)
24781 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24783 (microblaze_eh_return): New function.
24784 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24785 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24786 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24787 * config/microblaze/microblaze.md (eh_return): New pattern.
24789 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24791 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
24792 GCC_DIAGNOSTIC_STRINGIFY): Define.
24794 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
24796 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24798 * config/arm/arm.md (<mcrr>): New.
24800 * config/arm/arm.c (arm_arch5te): New.
24801 (arm_option_override): Set arm_arch5te.
24802 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
24804 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
24805 (arm_mcrr_qualifiers): ... this. New.
24806 (MRRC_QUALIFIERS): Define to...
24807 (arm_mrrc_qualifiers): ... this. New.
24808 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
24810 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
24811 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
24812 (MRRCI, mrrc, MRRC): New.
24813 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
24814 VUNSPEC_MRRC2): New.
24816 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24818 * config/arm/arm.md (<mcr>): New.
24820 * config/arm/arm.c (arm_coproc_builtin_available): Add
24821 support for mcr, mrc, mcr2 and mrc2.
24822 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
24823 (arm_mcr_qualifiers): ... this. New.
24824 (MRC_QUALIFIERS): Define to ...
24825 (arm_mrc_qualifiers): ... this. New.
24826 (MCR_QUALIFIERS): Define to ...
24827 (arm_mcr_qualifiers): ... this. New.
24828 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24830 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24831 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24832 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24833 VUNSPEC_MRC2): New.
24835 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24837 * config/arm/arm.md (*ldc): New.
24841 * config/arm/arm.c (arm_coproc_builtin_available): Add
24842 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24843 (arm_coproc_ldc_stc_legitimate_address): New.
24844 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24845 'qualifier_const_pointer'.
24846 (LDC_QUALIFIERS): Define to...
24847 (arm_ldc_qualifiers): ... this. New.
24848 (STC_QUALIFIERS): Define to...
24849 (arm_stc_qualifiers): ... this. New.
24850 * config/arm/arm-protos.h
24851 (arm_coproc_ldc_stc_legitimate_address): New.
24852 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24853 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24854 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24855 stc2, stcl, stc2l): New.
24856 * config/arm/constraints.md (Uz): New.
24857 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24858 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24859 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24860 VUNSPEC_STC2L): New.
24862 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24864 * config/arm/arm.md (<cdp>): New.
24865 * config/arm/arm.c (neon_const_bounds): Rename this ...
24866 (arm_const_bounds): ... this.
24867 (arm_coproc_builtin_available): New.
24868 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24869 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24870 (CDP_QUALIFIERS): Define to...
24871 (arm_cdp_qualifiers): ... this. New.
24873 (arm_expand_builtin_args): Add case for 6 arguments.
24874 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24875 (arm_const_bounds): ... this.
24876 (arm_coproc_builtin_available): New.
24877 * config/arm/arm_acle.h (__arm_cdp): New.
24879 * config/arm/arm_acle_builtins.def (cdp): New.
24881 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24882 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24883 'arm_const_bounds'.
24884 * config/arm/types.md (coproc): New.
24885 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24886 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24887 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24888 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24890 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24892 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24893 (UBINOP_QUALIFIERS): New.
24895 (acle_builtin_data): New. Change comment.
24896 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24897 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24898 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24899 arm_acle_builtins.def.
24900 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24901 (arm_init_acle_builtins): New.
24902 (CRC32_BUILTIN): Remove.
24903 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24904 crc32cb, crc32ch and crc32cw.
24905 (arm_init_crc32_builtins): Remove.
24906 (arm_init_builtins): Use arm_init_acle_builtins rather
24907 than arm_init_crc32_builtins.
24908 (arm_expand_acle_builtin): New.
24909 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24910 * config/arm/arm_acle_builtins.def: New.
24912 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24914 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24915 (arm_builtin_datum): ... this.
24916 (arm_init_neon_builtin): Rename to ...
24917 (arm_init_builtin): ... this. Add a new parameters PREFIX
24918 and USE_SIG_IN_NAME.
24919 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24920 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24921 'arm_builtin_datum'.
24922 (arm_init_vfp_builtins): Likewise.
24923 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24924 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24925 (arm_expand_neon_args): Rename to ...
24926 (arm_expand_builtin_args): ... this. Rename builtin_arg
24927 enum values and differentiate between ARG_BUILTIN_MEMORY
24928 and ARG_BUILTIN_NEON_MEMORY.
24929 (arm_expand_neon_builtin_1): Rename to ...
24930 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24931 values, arm_expand_builtin_args and add bool parameter NEON.
24932 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24933 (arm_expand_vfp_builtin): Likewise.
24934 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24936 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24938 PR middle-end/77484
24939 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24940 * predict.c (tree_estimate_probability_bb): Reverse direction of
24941 polymorphic call predictor.
24943 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24945 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24946 (determine_pass_name_match): ...this new function and...
24947 (should_skip_pass_p): ...this new function.
24949 2017-01-06 Nathan Sidwell <nathan@acm.org>
24951 * ipa-visibility.c (function_and_variable_visibility): Reformat
24952 comments and long lines. Remove extrneous if.
24953 * symtab.c (symtab_node::make_decl_local): Fix code format.
24954 (symtab_node::set_section_for_node): Fix comment typo.
24956 2017-01-06 Martin Liska <mliska@suse.cz>
24959 * lra-constraints.c: Rename invariant to lra_invariant.
24960 * predict.c (set_even_probabilities): Initialize e to NULL.
24962 2017-01-05 Martin Sebor <msebor@redhat.com>
24964 PR tree-optimization/78910
24965 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24966 (format_integer): Correct off-by-one error in the handling
24967 of precision with negative numbers in signed conversions..
24969 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24971 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24973 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24975 PR tree-optimization/71016
24976 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24977 factor_out_conditional_conversion. Formatting fix.
24978 (factor_out_conditional_conversion): Add cond_stmt argument.
24979 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24980 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24983 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24985 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24986 read-rtl-function.o, and selftest-rtl.o.
24987 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24988 (selftest::aarch64_test_loading_full_dump): New function.
24989 (selftest::aarch64_run_selftests): New function.
24990 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24991 selftest::aarch64_run_selftests.
24992 * config/i386/i386.c
24993 (selftest::ix86_test_loading_dump_fragment_1): New function.
24994 (selftest::ix86_test_loading_call_insn): New function.
24995 (selftest::ix86_test_loading_full_dump): New function.
24996 (selftest::ix86_test_loading_unspec): New function.
24997 (selftest::ix86_run_selftests): Call the new functions.
24998 * emit-rtl.c (maybe_set_max_label_num): New function.
24999 * emit-rtl.h (maybe_set_max_label_num): New decl.
25000 * function.c (instantiate_decls): Guard call to
25001 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25002 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25004 * gensupport.c (gen_reader::gen_reader): Pass "false"
25005 for new "compact" param of rtx_reader.
25006 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25007 rather than an empty string for NULL strings.
25008 * read-md.c: Potentially include config.h rather than bconfig.h.
25009 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25010 (have_error): New global, copied from errors.c.
25011 (md_reader::read_name): Rename to...
25012 (md_reader::read_name_1): ...this, adding "out_loc" param,
25013 and converting "missing name or number" to returning false, rather
25015 (md_reader::read_name): Reimplement in terms of read_name_1.
25016 (md_reader::read_name_or_nil): New function.
25017 (md_reader::read_string): Handle "(nil)" by returning NULL.
25018 (md_reader::md_reader): Add new param "compact".
25019 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25020 (md_reader::read_file): New method.
25021 * read-md.h (md_reader::md_reader): Add new param "compact".
25022 (md_reader::read_file): New method.
25023 (md_reader::is_compact): New accessor.
25024 (md_reader::read_name): Convert return type from void to file_location.
25025 (md_reader::read_name_or_nil): New decl.
25026 (md_reader::read_name_1): New decl.
25027 (md_reader::m_compact): New field.
25028 (noop_reader::noop_reader): Pass "false" for new "compact" param
25030 (rtx_reader::rtx_reader): Add new "compact" param.
25031 (rtx_reader::read_rtx_operand): Make virtual and convert return
25032 type from void to rtx.
25033 (rtx_reader::read_until): New decl.
25034 (rtx_reader::handle_any_trailing_information): New virtual function.
25035 (rtx_reader::postprocess): New virtual function.
25036 (rtx_reader::finalize_string): New virtual function.
25037 (rtx_reader::m_in_call_function_usage): New field.
25038 (rtx_reader::m_reuse_rtx_by_id): New field.
25039 * read-rtl-function.c: New file.
25040 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25041 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25042 (selftest::verify_three_block_rtl_cfg): New decl.
25043 * read-rtl-function.h: New file.
25044 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25045 For host, include function.h, memmodel.h, and emit-rtl.h.
25046 (one_time_initialization): New function.
25047 (struct compact_insn_name): New struct.
25048 (compact_insn_names): New array.
25049 (find_code): Handle insn codes in compact dumps.
25050 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25051 (bind_subst_iter_and_attr): Likewise.
25052 (add_condition_to_string): Likewise.
25053 (add_condition_to_rtx): Likewise.
25054 (apply_attribute_uses): Likewise.
25055 (add_current_iterators): Likewise.
25056 (apply_iterators): Likewise.
25057 (initialize_iterators): Guard usage of apply_subst_iterator with
25058 #ifdef GENERATOR_FILE.
25059 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25060 (md_reader::read_mapping): Likewise.
25061 (add_define_attr_for_define_subst): Likewise.
25062 (add_define_subst_attr): Likewise.
25063 (read_subst_mapping): Likewise.
25064 (check_code_iterator): Likewise.
25065 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25067 (one_time_initialization): New function.
25068 (rtx_reader::read_until): New method.
25069 (read_flags): New function.
25070 (parse_reg_note_name): New function.
25071 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25072 Handle reuse_rtx ids.
25073 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25074 Add parsing support for RTL dumps, mirroring the special-cases in
25075 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25076 values, and calling handle_any_trailing_information.
25077 (rtx_reader::read_rtx_operand): Convert return type from void
25078 to rtx, returning return_rtx. Handle case 'e'. Call
25079 finalize_string on XSTR and XTMPL fields.
25080 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25081 "(nil)" values were omitted. Call the postprocess vfunc on the
25083 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25084 class ctor. Initialize m_in_call_function_usage. Call
25085 one_time_initialization.
25086 * rtl-tests.c (selftest::test_uncond_jump): Call
25087 set_new_first_and_last_insn.
25088 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25089 * selftest-rtl.c: New file.
25090 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25091 (selftest::get_insn_by_uid): New decl.
25092 * selftest-run-tests.c (selftest::run_tests): Call
25093 read_rtl_function_c_tests.
25094 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25095 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25098 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25100 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25101 operands in a special way. Assert that pos+len <= mode precision.
25103 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25105 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25106 3 argument Alias with unlimited for the negative form.
25107 (fno-vect-cost-model): Removed.
25109 2017-01-05 Martin Liska <mliska@suse.cz>
25111 * hsa-gen.c (gen_hsa_divmod): New function.
25112 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25114 2017-01-05 Martin Liska <mliska@suse.cz>
25117 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25120 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25122 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25123 small constant length operands.
25125 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25127 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25128 between loop iterations.
25130 2017-01-05 Martin Liska <mliska@suse.cz>
25133 * gimplify.c (gimplify_decl_expr): Compare to
25134 asan_poisoned_variables instread of checking flags.
25135 (gimplify_target_expr): Likewise.
25136 (gimplify_expr): Likewise.
25137 (gimplify_function_tree): Conditionally initialize
25138 asan_poisoned_variables.
25140 2017-01-04 Jeff Law <law@redhat.com>
25142 PR tree-optimizatin/78812
25143 * rtl.h (contains_mem_rtx_p): Prototype.
25144 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25145 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25146 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25147 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25148 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25150 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25152 * input.c (assert_char_at_range): Default-initialize actual_range.
25154 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25156 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25157 to match the caller.
25159 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25161 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25162 insns after final jump in test to emit dummy move.
25164 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25166 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25167 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25169 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25171 * multiple_target.c (create_dispatcher_calls): Init e_next.
25172 * tree-ssa-loop-split.c (split_loop): Init border.
25173 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25176 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25181 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25182 (altivec_register_operand): Do not return true if the operand
25183 contains a SUBREG mixing SImode and SFmode.
25184 (vsx_register_operand): Likewise.
25185 (vsx_reg_sfsubreg_ok): New predicate.
25186 (vfloat_operand): Do not return true if the operand contains a
25187 SUBREG mixing SImode and SFmode.
25188 (vint_operand): Likewise.
25189 (vlogical_operand): Likewise.
25190 (gpc_reg_operand): Likewise.
25191 (int_reg_operand): Likewise.
25192 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25193 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25194 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25196 (rs6000_emit_move_si_sf_subreg): New helper function.
25197 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25198 fixup SUBREGs involving SImode and SFmode.
25199 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25200 numbers for the new peephole2 optimization.
25201 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25202 the GLIBC math library that do AND/IOR/XOR operations on single
25203 precision floating point.
25204 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25205 target macros to say whether we need to avoid SUBREGs mixing
25207 (TARGET_ALLOW_SF_SUBREG): Likewise.
25208 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25209 (UNSPEC_SI_FROM_SF): Likewise.
25210 (iorxor): Change spacing.
25211 (and_ior_xor): New iterator for AND, IOR, and XOR.
25212 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25213 (movdi_from_sf_zero_ext): Likewise.
25214 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25215 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25216 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25217 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25218 (fms<mode>4): Likewise.
25219 (fnma<mode>4): Likewise.
25220 (fnms<mode>4): Likewise.
25221 (nfma<mode>4): Likewise.
25222 (nfms<mode>4): Likewise.
25224 2017-01-04 Marek Polacek <polacek@redhat.com>
25227 * doc/invoke.texi: Document -Wpointer-compare.
25229 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25231 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25234 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25235 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25238 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25241 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25242 documentation of the powerpc_popcntb_ok attribute.
25243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25244 code to issue warning messages if a requested CPU configuration is
25245 not supported by the binary (assembler and loader) toolchain.
25246 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25247 made to define a built-in function that has been disabled.
25248 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25249 made to define a built-in function that has been disabled.
25250 (altivec_init_builtins): Add comment explaining why definition
25251 of the DST built-in functions is not preceded by an assertion
25252 check. Add assertions to prevent ICE if attempts are made to
25253 define an altivec predicate or an abs* built-in function that has
25255 (htm_init_builtins): Add comment explaining why definition of the
25256 htm built-in functions is not preceded by an assertion check.
25258 2017-01-04 Jeff Law <law@redhat.com>
25260 PR tree-optimizatin/67955
25261 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25262 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25263 the points-to solution does not include pt_null. Use DECL_PT_UID
25266 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25268 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25269 Use gen_int_mode instead of gen_lopwart for const_int operands.
25271 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25273 PR tree-optimization/71563
25274 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25275 out of range value - has low bits known to be zero.
25277 2017-01-04 Alan Modra <amodra@gmail.com>
25279 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25280 * configure: Regenerate.
25281 * config.in: Regenerate.
25283 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25286 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25287 a substring of the message, but strcmp with the whole message. Ifdef
25288 ENABLE_NLS, translate the message first using dgettext.
25290 2017-01-03 Jeff Law <law@redhat.com>
25292 PR tree-optimizatin/78856
25293 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25294 (mark_threaded_blocks): Remove code to truncate thread paths that
25295 cross multiple loop headers. Instead invalidate the cached loop
25296 iteration information and handle case of a thread path walking
25297 into an irreducible region.
25299 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25302 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25303 assertions. Add support for doing the signbit if the IEEE 128-bit
25304 floating point value is in a GPR.
25305 * config/rs6000/rs6000.md (Fsignbit): Delete.
25306 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25307 Update the length attribute if the value is in a GPR.
25308 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25309 the sign or zero extension instruction, since the value is always 0/1.
25310 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25313 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25314 extracting SImode to a GPR register so that we can generate a
25315 store, limit the vector to be in a traditional Altivec register
25316 for the vextuwrx instruction.
25318 2017-01-03 Ian Lance Taylor <iant@google.com>
25320 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25322 2017-01-03 Martin Sebor <msebor@redhat.com>
25324 PR tree-optimization/78696
25325 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25326 precision. Use MPFR for %f for greater fidelity. Correct handling
25328 (pass_sprintf_length::compute_format_length): Set width and precision
25329 specified by asrerisk to void_node for vararg functions.
25330 (try_substitute_return_value): Adjust dump output.
25332 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25334 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25336 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25338 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25339 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25340 -mlra/-mno-lra was passed to the compiler.
25342 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25344 PR rtl-optimization/65618
25345 * emit-rtl.c (try_split): Move initialization of "before" and
25346 "after" to just before the call to emit_insn_after_setloc.
25348 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25350 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25352 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25354 * dwarf2out.c (gen_enumeration_type_die): When
25355 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25357 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25359 PR tree-optimization/78965
25360 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25361 Change first argument from const call_info & to call_info &. For %n
25362 set info.nowrite to false.
25364 PR middle-end/78901
25365 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25366 possibly throwing calls.
25368 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25369 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25370 and fns handling, rather than in a separate case SSA_NAME.
25372 2017-01-02 Jeff Law <law@redhat.com>
25374 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25375 fixes for first_period and second_period variables.
25377 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25380 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25381 (*insvqi_1): New insn pattern.
25382 (*insvqi_1_mem_rex64): Ditto.
25383 (*insvqi_2): Ditto.
25384 (*insvqi_3): Rename from *insvqi.
25386 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25388 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25390 * doc/cfg.texi (Edges): Remove reference to Java.
25391 (Maintaining the CFG): Ditto.
25393 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25395 PR middle-end/77674
25396 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25397 transparent aliases.
25399 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25401 PR middle-end/77484
25402 * predict.def (PRED_CALL): Update hitrate.
25403 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25404 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25405 into direct/indirect/polymorphic variants.
25407 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25409 Update copyright years.
25411 * gcc.c (process_command): Update copyright notice dates.
25412 * gcov-dump.c (print_version): Ditto.
25413 * gcov.c (print_version): Ditto.
25414 * gcov-tool.c (print_version): Ditto.
25415 * gengtype.c (create_file): Ditto.
25416 * doc/cpp.texi: Bump @copying's copyright year.
25417 * doc/cppinternals.texi: Ditto.
25418 * doc/gcc.texi: Ditto.
25419 * doc/gccint.texi: Ditto.
25420 * doc/gcov.texi: Ditto.
25421 * doc/install.texi: Ditto.
25422 * doc/invoke.texi: Ditto.
25424 Copyright (C) 2017 Free Software Foundation, Inc.
25426 Copying and distribution of this file, with or without modification,
25427 are permitted in any medium without royalty provided the copyright
25428 notice and this notice are preserved.