1 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7 shift_unit_mode and for modes involved in scalar shifts.
9 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10 Alan Hayward <alan.hayward@arm.com>
11 David Sherwood <david.sherwood@arm.com>
13 * combine.c (force_int_to_mode): New function, split out from...
14 (force_to_mode): ...here. Keep xmode up-to-date and use it
15 instead of GET_MODE (x).
17 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18 Alan Hayward <alan.hayward@arm.com>
19 David Sherwood <david.sherwood@arm.com>
21 * optabs-query.h (extraction_insn::struct_mode): Change type to
22 opt_scalar_int_mode and update comment.
23 (extraction_insn::field_mode): Change type to scalar_int_mode.
24 (extraction_insn::pos_mode): Likewise.
25 * combine.c (make_extraction): Update accordingly.
26 * optabs-query.c (get_traditional_extraction_insn): Likewise.
27 (get_optab_extraction_insn): Likewise.
28 * recog.c (simplify_while_replacing): Likewise.
29 * expmed.c (narrow_bit_field_mem): Change the type of the mode
30 parameter to opt_scalar_int_mode.
32 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
33 Alan Hayward <alan.hayward@arm.com>
34 David Sherwood <david.sherwood@arm.com>
36 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
37 to a scalar_int_mode instead of a machine_mode.
38 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
39 (get_best_mode): Return a boolean and use a pointer argument to store
40 the selected mode. Replace the limit mode parameter with a bit limit.
41 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
42 for the values returned by bit_field_mode_iterator::next_mode.
43 (store_bit_field): Update call to get_best_mode.
44 (store_fixed_bit_field): Likewise.
45 (extract_fixed_bit_field): Likewise.
46 * expr.c (optimize_bitfield_assignment_op): Likewise.
47 * fold-const.c (optimize_bit_field_compare): Likewise.
48 (fold_truth_andor_1): Likewise.
49 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
50 Update for new type of m_mode.
51 (get_best_mode): As above.
53 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
54 Alan Hayward <alan.hayward@arm.com>
55 David Sherwood <david.sherwood@arm.com>
57 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
58 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
59 (store_bit_field): Check is_a <scalar_int_mode> before calling
60 strict_volatile_bitfield_p.
61 (extract_bit_field): Likewise.
63 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
64 Alan Hayward <alan.hayward@arm.com>
65 David Sherwood <david.sherwood@arm.com>
67 * target.def (cstore_mode): Return a scalar_int_mode.
68 * doc/tm.texi: Regenerate.
69 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
70 * targhooks.h (default_cstore_mode): Likewise.
71 * targhooks.c (default_cstore_mode): Likewise, using a forced
73 * expmed.c (emit_cstore): Expect the target of the cstore to be
76 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
77 Alan Hayward <alan.hayward@arm.com>
78 David Sherwood <david.sherwood@arm.com>
80 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
81 * builtins.c (expand_builtin_signbit): Use it.
82 * cfgexpand.c (expand_debug_expr): Likewise.
83 * dojump.c (do_jump): Likewise.
84 (do_compare_and_jump): Likewise.
85 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
86 * expmed.c (make_tree): Likewise.
87 * expr.c (expand_expr_real_2): Likewise.
88 (expand_expr_real_1): Likewise.
89 (try_casesi): Likewise.
90 * fold-const-call.c (fold_const_call_ss): Likewise.
91 * fold-const.c (unextend): Likewise.
92 (extract_muldiv_1): Likewise.
93 (fold_single_bit_test): Likewise.
94 (native_encode_int): Likewise.
95 (native_encode_string): Likewise.
96 (native_interpret_int): Likewise.
97 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
98 * internal-fn.c (expand_addsub_overflow): Likewise.
99 (expand_neg_overflow): Likewise.
100 (expand_mul_overflow): Likewise.
101 (expand_arith_overflow): Likewise.
102 * match.pd: Likewise.
103 * stor-layout.c (layout_type): Likewise.
104 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
105 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
106 * tree-ssanames.c (get_range_info): Likewise.
107 * tree-switch-conversion.c (array_value_type) Likewise.
108 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
109 (vect_recog_divmod_pattern): Likewise.
110 (vect_recog_mixed_size_cond_pattern): Likewise.
111 * tree-vrp.c (extract_range_basic): Likewise.
112 (simplify_float_conversion_using_ranges): Likewise.
113 * tree.c (int_fits_type_p): Likewise.
114 * ubsan.c (instrument_bool_enum_load): Likewise.
115 * varasm.c (mergeable_string_section): Likewise.
116 (narrowing_initializer_constant_valid_p): Likewise.
117 (output_constant): Likewise.
119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
120 Alan Hayward <alan.hayward@arm.com>
121 David Sherwood <david.sherwood@arm.com>
123 * machmode.h (NARROWEST_INT_MODE): New macro.
124 * expr.c (alignment_for_piecewise_move): Use it instead of
125 GET_CLASS_NARROWEST_MODE (MODE_INT).
126 (push_block): Likewise.
127 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
129 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
132 Alan Hayward <alan.hayward@arm.com>
133 David Sherwood <david.sherwood@arm.com>
135 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
137 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
138 * optabs.c (expand_unop): Likewise.
140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
141 Alan Hayward <alan.hayward@arm.com>
142 David Sherwood <david.sherwood@arm.com>
144 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
145 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
146 (popcount_loc_descriptor): Likewise.
147 (bswap_loc_descriptor): Likewise.
148 (rotate_loc_descriptor): Likewise.
149 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
150 calling the functions above.
152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
153 Alan Hayward <alan.hayward@arm.com>
154 David Sherwood <david.sherwood@arm.com>
156 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
158 (try_combine): Likewise.
159 (simplify_if_then_else): Likewise.
160 * cse.c (cse_insn): Likewise.
161 * dwarf2out.c (mem_loc_descriptor): Likewise.
162 * emit-rtl.c (gen_lowpart_common): Likewise.
163 * simplify-rtx.c (simplify_truncation): Likewise.
164 (simplify_binary_operation_1): Likewise.
165 (simplify_const_relational_operation): Likewise.
166 (simplify_ternary_operation): Likewise.
167 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
170 Alan Hayward <alan.hayward@arm.com>
171 David Sherwood <david.sherwood@arm.com>
173 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
174 * reload.c (push_reload): Likewise.
175 (find_reloads): Likewise.
177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
178 Alan Hayward <alan.hayward@arm.com>
179 David Sherwood <david.sherwood@arm.com>
181 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
182 (make_compound_operation_int): Likewise.
183 (change_zero_ext): Likewise.
184 * expr.c (convert_move): Likewise.
185 (convert_modes): Likewise.
186 * fwprop.c (forward_propagate_subreg): Likewise.
187 * loop-iv.c (get_biv_step_1): Likewise.
188 * optabs.c (widen_operand): Likewise.
189 * postreload.c (move2add_valid_value_p): Likewise.
190 * recog.c (simplify_while_replacing): Likewise.
191 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
192 (simplify_binary_operation_1): Likewise. Remove redundant
195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
196 Alan Hayward <alan.hayward@arm.com>
197 David Sherwood <david.sherwood@arm.com>
199 * combine.c (combine_simplify_rtx): Add checks for
200 is_a <scalar_int_mode>.
201 (simplify_if_then_else): Likewise.
202 (make_field_assignment): Likewise.
203 (simplify_comparison): Likewise.
204 * ifcvt.c (noce_try_bitop): Likewise.
205 * loop-invariant.c (canonicalize_address_mult): Likewise.
206 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
209 Alan Hayward <alan.hayward@arm.com>
210 David Sherwood <david.sherwood@arm.com>
212 * simplify-rtx.c (simplify_binary_operation_1): Use
213 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
216 Alan Hayward <alan.hayward@arm.com>
217 David Sherwood <david.sherwood@arm.com>
219 * wide-int.h (int_traits<unsigned char>) New class.
220 (int_traits<unsigned short>) Likewise.
221 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
222 Use GET_MODE_UNIT_PRECISION and remove redundant test for
224 * combine.c (set_nonzero_bits_and_sign_copies): Use
225 is_a <scalar_int_mode>.
226 (find_split_point): Likewise.
227 (combine_simplify_rtx): Likewise.
228 (simplify_logical): Likewise.
229 (expand_compound_operation): Likewise.
230 (expand_field_assignment): Likewise.
231 (make_compound_operation): Likewise.
232 (extended_count): Likewise.
233 (change_zero_ext): Likewise.
234 (simplify_comparison): Likewise.
235 * dwarf2out.c (scompare_loc_descriptor): Likewise.
236 (ucompare_loc_descriptor): Likewise.
237 (minmax_loc_descriptor): Likewise.
238 (mem_loc_descriptor): Likewise.
239 (loc_descriptor): Likewise.
240 * expmed.c (init_expmed_one_mode): Likewise.
241 * lra-constraints.c (lra_constraint_offset): Likewise.
242 * optabs.c (prepare_libcall_arg): Likewise.
243 * postreload.c (move2add_note_store): Likewise.
244 * reload.c (operands_match_p): Likewise.
245 * rtl.h (load_extend_op): Likewise.
246 * rtlhooks.c (gen_lowpart_general): Likewise.
247 * simplify-rtx.c (simplify_truncation): Likewise.
248 (simplify_unary_operation_1): Likewise.
249 (simplify_binary_operation_1): Likewise.
250 (simplify_const_binary_operation): Likewise.
251 (simplify_const_relational_operation): Likewise.
252 (simplify_subreg): Likewise.
253 * stor-layout.c (bitwise_mode_for_mode): Likewise.
254 * var-tracking.c (adjust_mems): Likewise.
255 (prepare_call_arguments): Likewise.
257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
258 Alan Hayward <alan.hayward@arm.com>
259 David Sherwood <david.sherwood@arm.com>
261 * machmode.h (is_int_mode): New fuction.
262 * combine.c (find_split_point): Use it.
263 (combine_simplify_rtx): Likewise.
264 (simplify_if_then_else): Likewise.
265 (simplify_set): Likewise.
266 (simplify_shift_const_1): Likewise.
267 (simplify_comparison): Likewise.
268 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
269 * cse.c (notreg_cost): Likewise.
270 (cse_insn): Likewise.
271 * cselib.c (cselib_lookup_1): Likewise.
272 * dojump.c (do_jump_1): Likewise.
273 (do_compare_rtx_and_jump): Likewise.
274 * dse.c (get_call_args): Likewise.
275 * dwarf2out.c (rtl_for_decl_init): Likewise.
276 (native_encode_initializer): Likewise.
277 * expmed.c (emit_store_flag_1): Likewise.
278 (emit_store_flag): Likewise.
279 * expr.c (convert_modes): Likewise.
280 (store_field): Likewise.
281 (expand_expr_real_1): Likewise.
282 * fold-const.c (fold_read_from_constant_string): Likewise.
283 * gimple-ssa-sprintf.c (get_format_string): Likewise.
284 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
285 * optabs.c (expand_binop): Likewise.
286 (expand_unop): Likewise.
287 (expand_abs_nojump): Likewise.
288 (expand_one_cmpl_abs_nojump): Likewise.
289 * simplify-rtx.c (mode_signbit_p): Likewise.
290 (val_signbit_p): Likewise.
291 (val_signbit_known_set_p): Likewise.
292 (val_signbit_known_clear_p): Likewise.
293 (simplify_relational_operation_1): Likewise.
294 * tree.c (vector_type_mode): Likewise.
296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
297 Alan Hayward <alan.hayward@arm.com>
298 David Sherwood <david.sherwood@arm.com>
300 * machmode.h (smallest_mode_for_size): Fix formatting.
301 (smallest_int_mode_for_size): New function.
302 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
303 instead of smallest_mode_for_size.
304 * combine.c (make_extraction): Likewise.
305 * config/arc/arc.c (arc_expand_movmem): Likewise.
306 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
307 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
308 * config/s390/s390.c (s390_expand_insv): Likewise.
309 * config/sparc/sparc.c (assign_int_registers): Likewise.
310 * config/spu/spu.c (spu_function_value): Likewise.
311 (spu_function_arg): Likewise.
312 * coverage.c (get_gcov_type): Likewise.
313 (get_gcov_unsigned_t): Likewise.
314 * dse.c (find_shift_sequence): Likewise.
315 * expmed.c (store_bit_field_1): Likewise.
316 * expr.c (convert_move): Likewise.
317 (store_field): Likewise.
318 * internal-fn.c (expand_arith_overflow): Likewise.
319 * optabs-query.c (get_best_extraction_insn): Likewise.
320 * optabs.c (expand_twoval_binop_libfunc): Likewise.
321 * stor-layout.c (layout_type): Likewise.
322 (initialize_sizetypes): Likewise.
323 * targhooks.c (default_get_mask_mode): Likewise.
324 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
327 Alan Hayward <alan.hayward@arm.com>
328 David Sherwood <david.sherwood@arm.com>
330 * machmode.h (opt_mode::else_blk): New function.
331 (int_mode_for_mode): Declare.
332 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
333 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
335 * cfgexpand.c (expand_debug_expr): Likewise.
336 * combine.c (gen_lowpart_or_truncate): Likewise.
337 (gen_lowpart_for_combine): Likewise.
338 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
339 * config/avr/avr.c (avr_to_int_mode): Likewise.
340 (avr_out_plus_1): Likewise.
341 (avr_out_plus): Likewise.
342 (avr_out_round): Likewise.
343 * config/i386/i386.c (ix86_split_to_parts): Likewise.
344 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
345 (s390_expand_vcond): Likewise.
346 * config/spu/spu.c (spu_split_immediate): Likewise.
347 (spu_expand_mov): Likewise.
348 * dse.c (get_stored_val): Likewise.
349 * expmed.c (store_bit_field_1): Likewise.
350 (convert_extracted_bit_field): Use int_mode_for_mode instead of
352 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
353 (extract_low_bits): Likewise.
354 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
355 handling rather than repeating the check.
356 (emit_group_store): Likewise.
357 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
358 * optabs.c (expand_absneg_bit): Likewise.
359 (expand_copysign_absneg): Likewise.
360 (expand_copysign_bit): Likewise.
361 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
362 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
363 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
364 * var-tracking.c (prepare_call_arguments): Likewise.
365 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
366 int_mode_for_mode instead of mode_for_size.
367 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
370 Alan Hayward <alan.hayward@arm.com>
371 David Sherwood <david.sherwood@arm.com>
373 * machmode.h (int_mode_for_size): New function.
374 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
375 instead of mode_for_size.
376 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
378 * combine.c (expand_field_assignment): Use int_mode_for_size
379 instead of mode_for_size.
380 (make_extraction): Likewise.
381 (simplify_shift_const_1): Likewise.
382 (simplify_comparison): Likewise.
383 * dojump.c (do_jump): Likewise.
384 * dwarf2out.c (mem_loc_descriptor): Likewise.
385 * emit-rtl.c (init_derived_machine_modes): Likewise.
386 * expmed.c (flip_storage_order): Likewise.
387 (convert_extracted_bit_field): Likewise.
388 * expr.c (copy_blkmode_from_reg): Likewise.
389 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
390 * internal-fn.c (expand_mul_overflow): Likewise.
391 * lower-subreg.c (simple_move): Likewise.
392 * optabs-libfuncs.c (init_optabs): Likewise.
393 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
394 * tree.c (vector_type_mode): Likewise.
395 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
396 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
397 * tree-vect-generic.c (expand_vector_parallel): Likewise.
398 * tree-vect-stmts.c (vectorizable_load): Likewise.
399 (vectorizable_store): Likewise.
401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
402 Alan Hayward <alan.hayward@arm.com>
403 David Sherwood <david.sherwood@arm.com>
405 * coretypes.h (pod_mode): New type.
406 (scalar_int_mode_pod): New typedef.
407 * machmode.h (pod_mode): New class.
408 (int_n_data_t::m): Change type to scalar_int_mode_pod.
409 * genmodes.c (emit_mode_int_n): Update accordingly.
410 * lower-subreg.h (target_lower_subreg): Change type to
412 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
416 Alan Hayward <alan.hayward@arm.com>
417 David Sherwood <david.sherwood@arm.com>
419 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
420 machine_mode to scalar_int_mode.
421 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
422 (rs6000_option_override_internal): Remove cast to int.
423 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
424 machine_mode to scalar_int_mode.
425 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
426 (rs6000_option_override_internal): Remove cast to int.
427 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
428 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
430 * config/s390/s390.c (s390_expand_builtin): Likewise.
431 * coretypes.h (scalar_int_mode): New type.
432 (opt_scalar_int_mode): New typedef.
433 * machmode.h (scalar_int_mode): New class.
434 (scalar_int_mode::includes_p): New function.
435 (byte_mode): Change type to scalar_int_mode.
436 (word_mode): Likewise.
437 (ptr_mode): Likewise.
438 * emit-rtl.c (byte_mode): Likewise.
439 (word_mode): Likewise.
440 (ptr_mode): Likewise.
441 (init_derived_machine_modes): Update accordingly.
442 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
443 and MODE_PARTIAL_INT.
444 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
448 Alan Hayward <alan.hayward@arm.com>
449 David Sherwood <david.sherwood@arm.com>
451 * target.def (libgcc_floating_mode_supported_p): Take a
453 * doc/tm.texi: Regenerate.
454 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
456 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
457 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
461 Alan Hayward <alan.hayward@arm.com>
462 David Sherwood <david.sherwood@arm.com>
464 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
465 * doc/tm.texi: Regenerate.
466 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
467 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
468 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
469 * targhooks.h (default_floatn_mode): Likewise.
470 * targhooks.c (default_floatn_mode): Likewise.
471 * tree.c (build_common_tree_nodes): Update accordingly.
473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
474 Alan Hayward <alan.hayward@arm.com>
475 David Sherwood <david.sherwood@arm.com>
477 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
478 (mode_iterator::iterate_p): Likewise.
479 (mode_iterator::get_wider): Likewise.
480 * expr.c (init_expr_target): Use opt_scalar_float_mode.
482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
483 Alan Hayward <alan.hayward@arm.com>
484 David Sherwood <david.sherwood@arm.com>
486 * coretypes.h (opt_scalar_float_mode): New typedef.
487 * machmode.h (float_mode_for_size): New function.
488 * emit-rtl.c (double_mode): Delete.
489 (init_emit_once): Use float_mode_for_size.
490 * stor-layout.c (layout_type): Likewise.
491 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
494 Alan Hayward <alan.hayward@arm.com>
495 David Sherwood <david.sherwood@arm.com>
497 * output.h (assemble_real): Take a scalar_float_mode.
498 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
499 * config/arm/arm.md (consttable_4): Likewise.
500 (consttable_8): Likewise.
501 (consttable_16): Likewise.
502 * config/mips/mips.md (consttable_float): Likewise.
503 * config/s390/s390.c (s390_output_pool_entry): Likewise.
504 * varasm.c (assemble_real): Take a scalar_float_mode.
505 (output_constant_pool_2): Update accordingly.
506 (output_constant): Likewise.
508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
509 Alan Hayward <alan.hayward@arm.com>
510 David Sherwood <david.sherwood@arm.com>
512 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
513 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
514 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
515 (native_encode_real): Likewise.
516 (native_interpret_real): Likewise.
517 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
518 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
521 Alan Hayward <alan.hayward@arm.com>
522 David Sherwood <david.sherwood@arm.com>
524 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
525 <scalar_float_mode>. Simplify.
526 (gen_extend_conv_libfunc): Likewise.
528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
529 Alan Hayward <alan.hayward@arm.com>
530 David Sherwood <david.sherwood@arm.com>
532 * coretypes.h (scalar_float_mode): New type.
533 * machmode.h (mode_traits::from_int): Use machine_mode if
534 USE_ENUM_MODES is defined.
535 (is_a): New function.
537 (dyn_cast): Likewise.
538 (scalar_float_mode): New class.
539 (scalar_float_mode::includes_p): New function.
540 (is_float_mode): Likewise.
541 * gdbhooks.py (MachineModePrinter): New class.
542 (build_pretty_printer): Use it for scalar_float_mode.
543 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
544 (format_helper::format_helper): Turn into a template.
545 * genmodes.c (get_mode_class): New function.
546 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
547 or machine_mode if none.
548 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
549 as_a <scalar_float_mode>.
550 * dwarf2out.c (mem_loc_descriptor): Likewise.
551 (insert_float): Likewise.
552 (add_const_value_attribute): Likewise.
553 * simplify-rtx.c (simplify_immed_subreg): Likewise.
554 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
555 (expand_unop): Update accordingly.
556 (expand_abs_nojump): Likewise.
557 (expand_copysign_absneg): Take a scalar_float_mode.
558 (expand_copysign_bit): Likewise.
559 (expand_copysign): Update accordingly.
561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * coretypes.h (opt_mode): New class.
566 * machmode.h (opt_mode): Likewise.
567 (opt_mode::else_void): New function.
568 (opt_mode::require): Likewise.
569 (opt_mode::exists): Likewise.
570 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
571 (GET_MODE_2XWIDER_MODE): Likewise.
572 (mode_iterator::get_wider): Update accordingly.
573 (mode_iterator::get_2xwider): Likewise.
574 (mode_iterator::get_known_wider): Likewise, turning into a template.
575 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
576 forcing a wider mode to exist.
577 * config/cr16/cr16.h (LONG_REG_P): Likewise.
578 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
579 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
580 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
581 * lower-subreg.c (init_lower_subreg): Likewise.
582 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
583 on the final iteration.
584 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
585 a wider mode exists before asking for a move pattern.
586 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
587 forcing a wider mode to exist.
588 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
589 returning false if no such mode exists.
590 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
591 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
592 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
593 Avoid checking for a MODE_INT if we already know the mode is not a
595 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
596 forcing a wider mode to exist.
597 (expmed_mult_highpart_optab): Likewise.
598 (expmed_mult_highpart): Likewise.
599 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
601 * lto-streamer-in.c (lto_input_mode_table): Likewise.
602 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
603 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
604 * internal-fn.c (expand_mul_overflow): Update use of
605 GET_MODE_2XWIDER_MODE.
606 * omp-low.c (omp_clause_aligned_alignment): Likewise.
607 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
609 (convert_plusminus_to_widen): Likewise.
610 * tree-switch-conversion.c (array_value_type): Likewise.
611 * var-tracking.c (emit_note_insn_var_location): Likewise.
612 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
613 Return false inside rather than outside the loop if no wider mode
615 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
616 and GET_MODE_2XWIDER_MODE
617 (can_compare_p): Use else_void.
618 * gdbhooks.py (OptMachineModePrinter): New class.
619 (build_pretty_printer): Use it for opt_mode.
621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
622 Alan Hayward <alan.hayward@arm.com>
623 David Sherwood <david.sherwood@arm.com>
625 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
626 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
629 Alan Hayward <alan.hayward@arm.com>
630 David Sherwood <david.sherwood@arm.com>
632 * machmode.h (mode_traits): New structure.
633 (get_narrowest_mode): New function.
634 (mode_iterator::start): Likewise.
635 (mode_iterator::iterate_p): Likewise.
636 (mode_iterator::get_wider): Likewise.
637 (mode_iterator::get_known_wider): Likewise.
638 (mode_iterator::get_2xwider): Likewise.
639 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
640 (FOR_EACH_MODE): Likewise.
641 (FOR_EACH_MODE_FROM): Likewise.
642 (FOR_EACH_MODE_UNTIL): Likewise.
643 (FOR_EACH_WIDER_MODE): Likewise.
644 (FOR_EACH_2XWIDER_MODE): Likewise.
645 * builtins.c (expand_builtin_strlen): Use new mode iterators.
646 * combine.c (simplify_comparison): Likewise
647 * config/i386/i386.c (type_natural_mode): Likewise.
648 * cse.c (cse_insn): Likewise.
649 * dse.c (find_shift_sequence): Likewise.
650 * emit-rtl.c (init_derived_machine_modes): Likewise.
651 (init_emit_once): Likewise.
652 * explow.c (hard_function_value): Likewise.
653 * expmed.c (extract_fixed_bit_field_1): Likewise.
654 (extract_bit_field_1): Likewise.
655 (expand_divmod): Likewise.
656 (emit_store_flag_1): Likewise.
657 * expr.c (init_expr_target): Likewise.
658 (convert_move): Likewise.
659 (alignment_for_piecewise_move): Likewise.
660 (widest_int_mode_for_size): Likewise.
661 (emit_block_move_via_movmem): Likewise.
662 (copy_blkmode_to_reg): Likewise.
663 (set_storage_via_setmem): Likewise.
664 (compress_float_constant): Likewise.
665 * omp-low.c (omp_clause_aligned_alignment): Likewise.
666 * optabs-query.c (get_best_extraction_insn): Likewise.
667 * optabs.c (expand_binop): Likewise.
668 (expand_twoval_unop): Likewise.
669 (expand_twoval_binop): Likewise.
670 (widen_leading): Likewise.
671 (widen_bswap): Likewise.
672 (expand_parity): Likewise.
673 (expand_unop): Likewise.
674 (prepare_cmp_insn): Likewise.
675 (prepare_float_lib_cmp): Likewise.
676 (expand_float): Likewise.
677 (expand_fix): Likewise.
678 (expand_sfix_optab): Likewise.
679 * postreload.c (move2add_use_add2_insn): Likewise.
680 * reg-stack.c (reg_to_stack): Likewise.
681 * reginfo.c (choose_hard_reg_mode): Likewise.
682 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
683 * stor-layout.c (mode_for_size): Likewise.
684 (smallest_mode_for_size): Likewise.
685 (mode_for_vector): Likewise.
686 (finish_bitfield_representative): Likewise.
687 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
688 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
689 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
690 * var-tracking.c (prepare_call_arguments): Likewise.
692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
693 Alan Hayward <alan.hayward@arm.com>
694 David Sherwood <david.sherwood@arm.com>
696 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
697 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
698 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
699 * machmode.h (mode_size): Move earlier in file.
700 (mode_precision): Likewise.
701 (mode_inner): Likewise.
702 (mode_nunits): Likewise.
703 (mode_unit_size): Likewise.
704 (unit_unit_precision): Likewise.
705 (mode_wider): Likewise.
706 (mode_2xwider): Likewise.
707 (machine_mode): New class.
708 (mode_to_bytes): New function.
709 (mode_to_bits): Likewise.
710 (mode_to_precision): Likewise.
711 (mode_to_inner): Likewise.
712 (mode_to_unit_size): Likewise.
713 (mode_to_unit_precision): Likewise.
714 (mode_to_nunits): Likewise.
715 (GET_MODE_SIZE): Use mode_to_bytes.
716 (GET_MODE_BITSIZE): Use mode_to_bits.
717 (GET_MODE_PRECISION): Use mode_to_precision.
718 (GET_MODE_INNER): Use mode_to_inner.
719 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
720 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
721 (GET_MODE_NUNITS): Use mode_to_nunits.
722 * system.h (ALWAYS_INLINE): New macro.
723 * config/powerpcspe/powerpcspe-c.c
724 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
725 int for arg1_mode and arg2_mode.
727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
728 Alan Hayward <alan.hayward@arm.com>
729 David Sherwood <david.sherwood@arm.com>
731 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
732 Prefix mode names with E_ in case statements.
733 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
734 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
735 (aarch64_split_simd_move): Likewise.
736 (aarch64_gen_storewb_pair): Likewise.
737 (aarch64_gen_loadwb_pair): Likewise.
738 (aarch64_gen_store_pair): Likewise.
739 (aarch64_gen_load_pair): Likewise.
740 (aarch64_get_condition_code_1): Likewise.
741 (aarch64_constant_pool_reload_icode): Likewise.
742 (get_rsqrte_type): Likewise.
743 (get_rsqrts_type): Likewise.
744 (get_recpe_type): Likewise.
745 (get_recps_type): Likewise.
746 (aarch64_gimplify_va_arg_expr): Likewise.
747 (aarch64_simd_container_mode): Likewise.
748 (aarch64_emit_load_exclusive): Likewise.
749 (aarch64_emit_store_exclusive): Likewise.
750 (aarch64_expand_compare_and_swap): Likewise.
751 (aarch64_gen_atomic_cas): Likewise.
752 (aarch64_emit_bic): Likewise.
753 (aarch64_emit_atomic_swap): Likewise.
754 (aarch64_emit_atomic_load_op): Likewise.
755 (aarch64_evpc_trn): Likewise.
756 (aarch64_evpc_uzp): Likewise.
757 (aarch64_evpc_zip): Likewise.
758 (aarch64_evpc_ext): Likewise.
759 (aarch64_evpc_rev): Likewise.
760 (aarch64_evpc_dup): Likewise.
761 (aarch64_gen_ccmp_first): Likewise.
762 (aarch64_gen_ccmp_next): Likewise.
763 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
764 (alpha_emit_xfloating_libcall): Likewise.
765 (emit_insxl): Likewise.
766 (alpha_arg_type): Likewise.
767 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
768 (arc_preferred_simd_mode): Likewise.
769 (arc_secondary_reload): Likewise.
770 (get_arc_condition_code): Likewise.
771 (arc_print_operand): Likewise.
772 (arc_legitimate_constant_p): Likewise.
773 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
774 * config/arc/arc.md (casesi_load): Likewise.
775 (casesi_compact_jump): Likewise.
776 * config/arc/predicates.md (proper_comparison_operator): Likewise.
777 (cc_use_register): Likewise.
778 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
779 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
780 (arm_init_iwmmxt_builtins): Likewise.
781 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
782 (neon_expand_vector_init): Likewise.
783 (arm_attr_length_move_neon): Likewise.
784 (maybe_get_arm_condition_code): Likewise.
785 (arm_emit_vector_const): Likewise.
786 (arm_preferred_simd_mode): Likewise.
787 (arm_output_iwmmxt_tinsr): Likewise.
788 (thumb1_output_casesi): Likewise.
789 (thumb2_output_casesi): Likewise.
790 (arm_emit_load_exclusive): Likewise.
791 (arm_emit_store_exclusive): Likewise.
792 (arm_expand_compare_and_swap): Likewise.
793 (arm_evpc_neon_vuzp): Likewise.
794 (arm_evpc_neon_vzip): Likewise.
795 (arm_evpc_neon_vrev): Likewise.
796 (arm_evpc_neon_vtrn): Likewise.
797 (arm_evpc_neon_vext): Likewise.
798 (arm_validize_comparison): Likewise.
799 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
800 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
801 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
802 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
803 (c6x_preferred_simd_mode): Likewise.
804 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
805 (epiphany_rtx_costs): Likewise.
806 * config/epiphany/predicates.md (proper_comparison_operator):
808 * config/frv/frv.c (condexec_memory_operand): Likewise.
809 (frv_emit_move): Likewise.
810 (output_move_single): Likewise.
811 (output_condmove_single): Likewise.
812 (frv_hard_regno_mode_ok): Likewise.
813 (frv_matching_accg_mode): Likewise.
814 * config/h8300/h8300.c (split_adds_subs): Likewise.
815 (h8300_rtx_costs): Likewise.
816 (h8300_print_operand): Likewise.
817 (compute_mov_length): Likewise.
818 (output_logical_op): Likewise.
819 (compute_logical_op_length): Likewise.
820 (compute_logical_op_cc): Likewise.
821 (h8300_shift_needs_scratch_p): Likewise.
822 (output_a_shift): Likewise.
823 (compute_a_shift_length): Likewise.
824 (compute_a_shift_cc): Likewise.
825 (expand_a_rotate): Likewise.
826 (output_a_rotate): Likewise.
827 * config/i386/i386.c (classify_argument): Likewise.
828 (function_arg_advance_32): Likewise.
829 (function_arg_32): Likewise.
830 (function_arg_64): Likewise.
831 (function_value_64): Likewise.
832 (ix86_gimplify_va_arg): Likewise.
833 (ix86_legitimate_constant_p): Likewise.
834 (put_condition_code): Likewise.
835 (split_double_mode): Likewise.
836 (ix86_avx256_split_vector_move_misalign): Likewise.
837 (ix86_expand_vector_logical_operator): Likewise.
838 (ix86_split_idivmod): Likewise.
839 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
840 (ix86_build_const_vector): Likewise.
841 (ix86_build_signbit_mask): Likewise.
842 (ix86_match_ccmode): Likewise.
843 (ix86_cc_modes_compatible): Likewise.
844 (ix86_expand_branch): Likewise.
845 (ix86_expand_sse_cmp): Likewise.
846 (ix86_expand_sse_movcc): Likewise.
847 (ix86_expand_int_sse_cmp): Likewise.
848 (ix86_expand_vec_perm_vpermi2): Likewise.
849 (ix86_expand_vec_perm): Likewise.
850 (ix86_expand_sse_unpack): Likewise.
851 (ix86_expand_int_addcc): Likewise.
852 (ix86_split_to_parts): Likewise.
853 (ix86_vectorize_builtin_gather): Likewise.
854 (ix86_vectorize_builtin_scatter): Likewise.
855 (avx_vpermilp_parallel): Likewise.
856 (inline_memory_move_cost): Likewise.
857 (ix86_tieable_integer_mode_p): Likewise.
858 (x86_maybe_negate_const_int): Likewise.
859 (ix86_expand_vector_init_duplicate): Likewise.
860 (ix86_expand_vector_init_one_nonzero): Likewise.
861 (ix86_expand_vector_init_one_var): Likewise.
862 (ix86_expand_vector_init_concat): Likewise.
863 (ix86_expand_vector_init_interleave): Likewise.
864 (ix86_expand_vector_init_general): Likewise.
865 (ix86_expand_vector_set): Likewise.
866 (ix86_expand_vector_extract): Likewise.
867 (emit_reduc_half): Likewise.
868 (ix86_emit_i387_round): Likewise.
869 (ix86_mangle_type): Likewise.
870 (ix86_expand_round_sse4): Likewise.
871 (expand_vec_perm_blend): Likewise.
872 (canonicalize_vector_int_perm): Likewise.
873 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
874 (expand_vec_perm_1): Likewise.
875 (expand_vec_perm_interleave3): Likewise.
876 (expand_vec_perm_even_odd_pack): Likewise.
877 (expand_vec_perm_even_odd_1): Likewise.
878 (expand_vec_perm_broadcast_1): Likewise.
879 (ix86_vectorize_vec_perm_const_ok): Likewise.
880 (ix86_expand_vecop_qihi): Likewise.
881 (ix86_expand_mul_widen_hilo): Likewise.
882 (ix86_expand_sse2_abs): Likewise.
883 (ix86_expand_pextr): Likewise.
884 (ix86_expand_pinsr): Likewise.
885 (ix86_preferred_simd_mode): Likewise.
886 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
887 * config/i386/sse.md (*andnot<mode>3): Likewise.
888 (<mask_codefor><code><mode>3<mask_name>): Likewise.
889 (*<code><mode>3): Likewise.
890 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
891 (ia64_expand_atomic_op): Likewise.
892 (ia64_arg_type): Likewise.
893 (ia64_mode_to_int): Likewise.
894 (ia64_scalar_mode_supported_p): Likewise.
895 (ia64_vector_mode_supported_p): Likewise.
896 (expand_vec_perm_broadcast): Likewise.
897 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
898 (iq2000_function_arg_advance): Likewise.
899 (iq2000_function_arg): Likewise.
900 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
901 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
902 (m68k_libcall_value): Likewise.
903 (m68k_function_value): Likewise.
904 (sched_attr_op_type): Likewise.
905 * config/mcore/mcore.c (mcore_output_move): Likewise.
906 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
908 (microblaze_function_arg): Likewise.
909 * config/mips/mips.c (mips16_build_call_stub): Likewise.
910 (mips_print_operand): Likewise.
911 (mips_mode_ok_for_mov_fmt_p): Likewise.
912 (mips_vector_mode_supported_p): Likewise.
913 (mips_preferred_simd_mode): Likewise.
914 (mips_expand_vpc_loongson_even_odd): Likewise.
915 (mips_expand_vec_unpack): Likewise.
916 (mips_expand_vi_broadcast): Likewise.
917 (mips_expand_vector_init): Likewise.
918 (mips_expand_vec_reduc): Likewise.
919 (mips_expand_msa_cmp): Likewise.
920 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
921 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
922 (cc_flags_for_mode): Likewise.
923 * config/msp430/msp430.c (msp430_print_operand): Likewise.
924 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
925 (nds32_output_casesi_pc_relative): Likewise.
926 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
927 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
928 (nvptx_gen_unpack): Likewise.
929 (nvptx_gen_pack): Likewise.
930 (nvptx_gen_shuffle): Likewise.
931 (nvptx_gen_wcast): Likewise.
932 (nvptx_preferred_simd_mode): Likewise.
933 * config/pa/pa.c (pa_secondary_reload): Likewise.
934 * config/pa/predicates.md (base14_operand): Likewise.
935 * config/powerpcspe/powerpcspe-c.c
936 (altivec_resolve_overloaded_builtin): Likewise.
937 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
939 (rs6000_preferred_simd_mode): Likewise.
940 (output_vec_const_move): Likewise.
941 (rs6000_expand_vector_extract): Likewise.
942 (rs6000_split_vec_extract_var): Likewise.
943 (reg_offset_addressing_ok_p): Likewise.
944 (rs6000_legitimate_offset_address_p): Likewise.
945 (rs6000_legitimize_address): Likewise.
946 (rs6000_emit_set_const): Likewise.
947 (rs6000_const_vec): Likewise.
948 (rs6000_emit_move): Likewise.
949 (spe_build_register_parallel): Likewise.
950 (rs6000_darwin64_record_arg_recurse): Likewise.
951 (swap_selector_for_mode): Likewise.
952 (spe_init_builtins): Likewise.
953 (paired_init_builtins): Likewise.
954 (altivec_init_builtins): Likewise.
955 (do_load_for_compare): Likewise.
956 (rs6000_generate_compare): Likewise.
957 (rs6000_expand_float128_convert): Likewise.
958 (emit_load_locked): Likewise.
959 (emit_store_conditional): Likewise.
960 (rs6000_output_function_epilogue): Likewise.
961 (rs6000_handle_altivec_attribute): Likewise.
962 (rs6000_function_value): Likewise.
963 (emit_fusion_gpr_load): Likewise.
964 (emit_fusion_p9_load): Likewise.
965 (emit_fusion_p9_store): Likewise.
966 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
967 (fusion_gpr_mem_load): Likewise.
968 (fusion_addis_mem_combo_load): Likewise.
969 (fusion_addis_mem_combo_store): Likewise.
970 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
971 (fusion_gpr_mem_load): Likewise.
972 (fusion_addis_mem_combo_load): Likewise.
973 (fusion_addis_mem_combo_store): Likewise.
974 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
976 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
977 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
978 (rs6000_preferred_simd_mode): Likewise.
979 (output_vec_const_move): Likewise.
980 (rs6000_expand_vector_extract): Likewise.
981 (rs6000_split_vec_extract_var): Likewise.
982 (reg_offset_addressing_ok_p): Likewise.
983 (rs6000_legitimate_offset_address_p): Likewise.
984 (rs6000_legitimize_address): Likewise.
985 (rs6000_emit_set_const): Likewise.
986 (rs6000_const_vec): Likewise.
987 (rs6000_emit_move): Likewise.
988 (rs6000_darwin64_record_arg_recurse): Likewise.
989 (swap_selector_for_mode): Likewise.
990 (paired_init_builtins): Likewise.
991 (altivec_init_builtins): Likewise.
992 (rs6000_expand_float128_convert): Likewise.
993 (emit_load_locked): Likewise.
994 (emit_store_conditional): Likewise.
995 (rs6000_output_function_epilogue): Likewise.
996 (rs6000_handle_altivec_attribute): Likewise.
997 (rs6000_function_value): Likewise.
998 (emit_fusion_gpr_load): Likewise.
999 (emit_fusion_p9_load): Likewise.
1000 (emit_fusion_p9_store): Likewise.
1001 * config/rx/rx.c (rx_gen_move_template): Likewise.
1002 (flags_from_mode): Likewise.
1003 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1004 (s390_slb_comparison): Likewise.
1005 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1006 (s390_vector_mode_supported_p): Likewise.
1007 (s390_cc_modes_compatible): Likewise.
1008 (s390_match_ccmode_set): Likewise.
1009 (s390_canonicalize_comparison): Likewise.
1010 (s390_emit_compare_and_swap): Likewise.
1011 (s390_branch_condition_mask): Likewise.
1012 (s390_rtx_costs): Likewise.
1013 (s390_secondary_reload): Likewise.
1014 (__SECONDARY_RELOAD_CASE): Likewise.
1015 (s390_expand_cs): Likewise.
1016 (s390_preferred_simd_mode): Likewise.
1017 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1018 * config/sh/sh.c (sh_print_operand): Likewise.
1019 (dump_table): Likewise.
1020 (sh_secondary_reload): Likewise.
1021 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1022 * config/sh/sh.md (casesi_worker_1): Likewise.
1023 (casesi_worker_2): Likewise.
1024 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1025 (fcc_comparison_operator): Likewise.
1026 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1027 (emit_soft_tfmode_cvt): Likewise.
1028 (sparc_preferred_simd_mode): Likewise.
1029 (output_cbranch): Likewise.
1030 (sparc_print_operand): Likewise.
1031 (sparc_expand_vec_perm_bmask): Likewise.
1032 (vector_init_bshuffle): Likewise.
1033 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1034 (spu_vector_mode_supported_p): Likewise.
1035 (spu_expand_insv): Likewise.
1036 (spu_emit_branch_or_set): Likewise.
1037 (spu_handle_vector_attribute): Likewise.
1038 (spu_builtin_splats): Likewise.
1039 (spu_builtin_extract): Likewise.
1040 (spu_builtin_promote): Likewise.
1041 (spu_expand_sign_extend): Likewise.
1042 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1043 (tilegx_simd_int): Likewise.
1044 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1045 (tilepro_simd_int): Likewise.
1046 * config/v850/v850.c (const_double_split): Likewise.
1047 (v850_print_operand): Likewise.
1048 (ep_memory_offset): Likewise.
1049 * config/vax/vax.c (vax_rtx_costs): Likewise.
1050 (vax_output_int_move): Likewise.
1051 (vax_output_int_add): Likewise.
1052 (vax_output_int_subtract): Likewise.
1053 * config/visium/predicates.md (visium_branch_operator): Likewise.
1054 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1055 (visium_print_operand_address): Likewise.
1056 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1057 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1058 (xtensa_expand_conditional_branch): Likewise.
1059 (xtensa_copy_incoming_a7): Likewise.
1060 (xtensa_output_literal): Likewise.
1061 * dfp.c (decimal_real_maxval): Likewise.
1062 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1065 Alan Hayward <alan.hayward@arm.com>
1066 David Sherwood <david.sherwood@arm.com>
1068 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1069 (mode_nunits_inline): Likewise.
1070 (mode_inner_inline): Likewise.
1071 (mode_unit_size_inline): Likewise.
1072 (mode_unit_precision_inline): Likewise.
1073 (emit_insn_modes_h): Likewise. Also emit a #define of the
1075 (emit_mode_wider): Add an E_ prefix to mode names.
1076 (emit_mode_complex): Likewise.
1077 (emit_mode_inner): Likewise.
1078 (emit_mode_adjustments): Likewise.
1079 (emit_mode_int_n): Likewise.
1080 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1081 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1082 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1083 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1084 (CRC32_BUILTIN, ENTRY): Likewise.
1085 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1086 (aarch64_pop_regs): Likewise.
1087 (aarch64_process_components): Likewise.
1088 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1089 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1090 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1091 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1092 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1093 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1095 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1096 (pcmpistr): Likewise.
1097 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1098 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1099 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1101 * config/rl78/rl78.c (mduc_regs): Likewise.
1102 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1103 (htm_expand_builtin): Likewise.
1104 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1105 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1106 * config/xtensa/xtensa.c (print_operand): Likewise.
1107 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1108 (NUM_MODE_VECTOR_INT): Likewise.
1109 * genoutput.c (null_operand): Likewise.
1110 (output_operand_data): Likewise.
1111 * genrecog.c (print_parameter_value): Likewise.
1112 * lra.c (debug_operand_data): Likewise.
1114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1116 * dwarf2out.c (scompare_loc_descriptor_wide)
1117 (scompare_loc_descriptor_narrow): New functions, split out from...
1118 (scompare_loc_descriptor): ...here.
1119 * expmed.c (emit_store_flag_int): New function, split out from...
1120 (emit_store_flag): ...here.
1122 2017-08-30 Richard Biener <rguenther@suse.de>
1124 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1125 (dwarf2out_early_finish): Move setting of AT_pubnames from
1126 early debug output to early finish.
1128 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1130 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1131 and -mdata-region to the assembler.
1133 2017-08-30 Richard Biener <rguenther@suse.de>
1135 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1137 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1138 (dwarf2out_early_global_decl): Only generate a DIE for the
1139 abstract origin if it doesn't already exist or is a declaration DIE.
1140 (resolve_addr): Do not add the linkage name twice when
1141 generating a stub DIE for the DW_TAG_GNU_call_site target.
1143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1145 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1146 Use machine_mode rather than int for arg1_mode.
1148 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1151 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1152 that the second argument of the built-in functions to unpack
1153 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1154 switch statement instead a lot of if statements.
1155 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1156 Allow 64-bit values to be in Altivec registers as well as
1157 traditional floating point registers.
1158 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1160 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1162 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1163 MAX_REGS_PER_ADDRESS == 1.
1165 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1167 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1168 * config/i386/i386.c (ix86_option_override_internal): Simplify
1169 setting of opts->x_flag_entry.
1171 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1172 Jakub Jelinek <jakub@redhat.com>
1173 Richard Biener <rguenther@suse.de>
1175 PR tree-optimization/81503
1176 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1177 folded constant fits in the target type; reorder tests for clarity.
1179 2017-08-29 Martin Liska <mliska@suse.cz>
1181 * passes.def: Include pass_lower_switch.
1182 * stmt.c (dump_case_nodes): Remove and move to
1183 tree-switch-conversion.
1184 (case_values_threshold): Likewise.
1185 (expand_switch_as_decision_tree_p): Likewise.
1186 (emit_case_decision_tree): Likewise.
1187 (expand_case): Likewise.
1188 (balance_case_nodes): Likewise.
1189 (node_has_low_bound): Likewise.
1190 (node_has_high_bound): Likewise.
1191 (node_is_bounded): Likewise.
1192 (emit_case_nodes): Likewise.
1193 (struct simple_case_node): New struct.
1194 (add_case_node): Remove.
1195 (emit_case_dispatch_table): Use vector instead of case_list.
1196 (reset_out_edges_aux): Remove.
1197 (compute_cases_per_edge): Likewise.
1198 (expand_case): Build list of simple_case_node.
1199 (expand_sjlj_dispatch_table): Use it.
1200 * tree-switch-conversion.c (struct case_node): Moved from
1201 stmt.c and adjusted.
1202 (emit_case_nodes): Likewise.
1203 (node_has_low_bound): Likewise.
1204 (node_has_high_bound): Likewise.
1205 (node_is_bounded): Likewise.
1206 (case_values_threshold): Likewise.
1207 (reset_out_edges_aux): Likewise.
1208 (compute_cases_per_edge): Likewise.
1209 (add_case_node): Likewise.
1210 (dump_case_nodes): Likewise.
1211 (balance_case_nodes): Likewise.
1212 (expand_switch_as_decision_tree_p): Likewise.
1213 (emit_jump): Likewise.
1214 (emit_case_decision_tree): Likewise.
1215 (try_switch_expansion): Likewise.
1216 (do_jump_if_equal): Likewise.
1217 (emit_cmp_and_jump_insns): Likewise.
1218 (fix_phi_operands_for_edge): New function.
1219 (record_phi_operand_mapping): Likewise.
1220 (class pass_lower_switch): New pass.
1221 (pass_lower_switch::execute): New function.
1222 (make_pass_lower_switch): Likewise.
1223 (conditional_probability):
1224 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1225 * tree-pass.h: Add make_pass_lower_switch.
1227 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1230 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1233 2017-08-29 Richard Biener <rguenther@suse.de>
1235 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1236 we do not add a DW_AT_inline attribute twice.
1237 (gen_subprogram_die): Remove code setting DW_AT_inline on
1238 DECL_ABSTRACT_P nodes.
1240 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1242 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1243 calls to internal functions.
1244 (gimplify_modify_expr): Likewise.
1245 * tree-call-cdce.c (use_internal_fn): Likewise.
1246 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1247 (convert_to_divmod): Set the nothrow flag.
1248 * tree-if-conv.c (predicate_mem_writes): Likewise.
1249 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1250 (vectorizable_call): Likewise.
1251 (vectorizable_store): Likewise.
1252 (vectorizable_load): Likewise.
1253 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1254 (vect_recog_mask_conversion_pattern): Likewise.
1256 2017-08-29 Martin Liska <mliska@suse.cz>
1259 * gcc.c (driver_handle_option): Add new argument.
1260 * opts-common.c (handle_option): Pass
1261 target_option_override_hook.
1262 * opts-global.c (lang_handle_option): Add new option.
1263 (set_default_handlers): Add new argument.
1264 (decode_options): Likewise.
1265 * opts.c (target_handle_option): Likewise.
1266 (common_handle_option): Call target_option_override_hook.
1267 * opts.h (struct cl_option_handler_func): Add hook for
1268 target option override.
1269 (struct cl_option_handlers): Likewise.
1270 (set_default_handlers): Add new argument.
1271 (decode_options): Likewise.
1272 (common_handle_option): Likewise.
1273 (target_handle_option): Likewise.
1274 * toplev.c (toplev::main): Pass targetm.target_option.override
1277 2017-08-29 Richard Biener <rguenther@suse.de>
1278 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1280 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1281 life to the active subtree.
1283 2017-08-28 Jeff Law <law@redhat.com>
1285 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1286 derive_equivalences.
1287 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1289 (edge_info::derive_equivalences): New private member function
1291 * tree-ssa-dom.c (class edge_info): Changed from a struct
1292 to a class. Add ctor/dtor, methods and data members.
1293 (edge_info::edge_info): Renamed from allocate_edge_info.
1294 Initialize additional members.
1295 (edge_info::~edge_info): New.
1296 (free_dom_edge_info): Delete the edge info.
1297 (record_edge_info): Use new class & associated member functions.
1298 Tighten forms for testing for edge equivalences.
1299 (record_temporary_equivalences): Iterate over the simple
1300 equivalences rather than assuming there's only one per edge.
1301 (cprop_into_successor_phis): Iterate over the simple
1302 equivalences rather than assuming there's only one per edge.
1303 (optimize_stmt): Use operand_equal_p rather than pointer
1304 equality for mini-DSE code.
1306 2017-08-28 Nathan Sidwell <nathan@acm.org>
1308 * gcc.c (execute): Fold SIGPIPE handling into switch
1309 statement. Adjust internal error message.
1311 2017-08-28 Richard Biener <rguenther@suse.de>
1314 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1315 Do nothing for removed DIEs.
1317 2017-08-28 Richard Biener <rguenther@suse.de>
1319 PR tree-optimization/81977
1320 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1323 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1326 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1327 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1328 using targetm.gen_mem_thread_fence.
1330 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1333 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1334 predicate to register_operand. Reorder operands.
1335 (*btr<mode>): Ditto.
1336 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1337 (*btr<mode>_mask): Ditto.
1339 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1341 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1342 * config/rs6000/xmmintrin.h: New file.
1343 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1345 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1348 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1349 parameter and_insn and return it.
1350 (recombine_lvx_pattern): Insert a copy to ensure availability of
1351 the base register of the copied masking operation at the point of
1352 the instruction replacement.
1353 (recombine_stvx_pattern): Likewise.
1355 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1357 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1358 undocumented switches.
1359 (-mpower9-dform-vector): Likewise.
1360 (-mpower9-dform): Likewise.
1361 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1362 comments to delete references to -mpower9-dform* switches.
1363 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1364 Delete reference to -mpower9-dform* switches, test for
1365 -mpower9-vector instead.
1366 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1367 (OTHER_P9_VECTOR_MASKS): Likewise.
1368 (POWERPC_MASKS): Likewise.
1369 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1370 tests against -mpower9-dform* to -mpower9-vector. Delete code
1371 that checked for -mpower9-dform* consistancy with other options.
1372 Add test for -mpower9-misc to enable other power9 switches.
1373 (rs6000_init_hard_regno_mode_ok): Likewise.
1374 (rs6000_option_override_internal): Likewise.
1375 (rs6000_emit_prologue): Likewise.
1376 (rs6000_emit_epilogue): Likewise.
1377 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1378 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1379 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1381 (emit_fusion_p9_store): Likewise.
1382 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1383 resetting these macros if the assembler does not support ISA 3.0
1385 (TARGET_P9_DFORM_VECTOR): Likewise.
1386 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1387 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1389 2017-08-25 Alan Modra <amodra@gmail.com>
1391 PR rtl-optimization/81747
1392 * cse.c (cse_extended_basic_block): Don't attempt to record
1393 equivalences for degenerate conditional branches that branch
1394 to their fall-through.
1396 2017-08-24 Martin Sebor <msebor@redhat.com>
1399 * gimple-fold.c (size_must_be_zero_p): New function.
1400 (gimple_fold_builtin_memory_op): Call it.
1402 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1404 * config/rs6000/mm_malloc.h: New file.
1406 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1408 PR tree-optimization/81913
1409 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1410 analysis when either IVs in condition can wrap.
1412 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1414 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1415 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1417 2017-08-24 Richard Biener <rguenther@suse.de>
1420 * targhooks.c (default_target_can_inline_p): Properly
1421 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1422 is present and always compare.
1423 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1424 infer -mfpmath=sse from TARGET_SSE_P.
1425 (ix86_can_inline_p): Properly use target_option_default_node when
1426 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1428 2017-08-24 Richard Biener <rguenther@suse.de>
1431 * dwarf2out.c (output_die): Handle flag_generate_offload like
1433 (output_comp_unit): Likewise.
1434 (gen_array_type_die): Likewise.
1435 (dwarf2out_early_finish): Likewise.
1436 (note_variable_value_in_expr): Likewise.
1437 (dwarf2out_finish): Likewise. Adjust assert.
1438 * cgraphunit.c (symbol_table::compile): Move setting of
1439 flag_generate_offload earlier ...
1440 (symbol_table::finalize_compilation_unit): ... here, before
1441 early debug finalization.
1443 2017-08-24 Richard Biener <rguenther@suse.de>
1445 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1446 and ipa-fnsummary.h.
1447 (ix86_can_inline_p): When ix86_fpmath flags do not match
1448 check whether the callee uses FP math at all.
1450 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1453 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1454 instead of TYPE_PRECISION.
1456 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1457 Alan Hayward <alan.hayward@arm.com>
1458 David Sherwood <david.sherwood@arm.com>
1460 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1462 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1464 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1466 (ix86_frame::stack_realign_allocate): New field.
1467 (struct machine_frame_state): Modify comments.
1468 (machine_frame_state::sp_realigned_fp_end): New field.
1469 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1471 (sp_valid_at): Add assertion to assure no attempt to access invalid
1472 offset of a realigned stack.
1473 (fp_valid_at): Likewise.
1474 (choose_baseaddr): Modify comments.
1475 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1476 ix86_expand_prologue.
1477 (ix86_expand_prologue): Modify stack realignment and allocation.
1478 (ix86_expand_epilogue): Modify comments.
1479 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1480 avx2_runtime, avx512f, and avx512f_runtime.
1482 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1484 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1485 (mstackrealign): Do not init to -1.
1486 * config/i386/i386.c (ix86_option_override_internal):
1487 Check opts_set, not opts when setting default value of
1488 opts->x_ix86_force_align_arg_pointer.
1490 2017-08-23 Richard Biener <rguenther@suse.de>
1492 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1493 lang_hooks.decl_printable_name.
1494 * print-rtl-function.c (print_rtx_function): Likewise.
1495 * tree-pretty-print.c (dump_function_header): Likewise.
1497 2017-08-23 Richard Biener <rguenther@suse.de>
1500 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1501 -g0 at compile-time.
1503 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1506 * doc/sourcebuild.texi (Other hardware attributes):
1509 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1512 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1513 Fix single-use check.
1515 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1517 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1519 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1521 * doc/install.texi: Modify to add more details on running selected
1524 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1526 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1527 is combined with -mabi=ms.
1528 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1531 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1533 PR tree-optimization/81488
1534 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1535 and cached_basis fields.
1536 (MAX_SPREAD): New constant.
1537 (alloc_cand_and_find_basis): Initialize new fields.
1538 (clear_visited): New function.
1539 (create_phi_basis_1): Rename from create_phi_basis, set visited
1540 and cached_basis fields.
1541 (create_phi_basis): New wrapper function.
1542 (phi_add_costs_1): Rename from phi_add_costs, add spread
1543 parameter, set visited field, short-circuit when limits reached.
1544 (phi_add_costs): New wrapper function.
1545 (record_phi_increments_1): Rename from record_phi_increments, set
1547 (record_phi_increments): New wrapper function.
1548 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1549 (phi_incr_cost): New wrapper function.
1550 (all_phi_incrs_profitable_1): Rename from
1551 all_phi_incrs_profitable, set visited field.
1552 (all_phi_incrs_profitable): New wrapper function.
1554 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1555 Alan Hayward <alan.hayward@arm.com>
1556 David Sherwood <david.sherwood@arm.com>
1558 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1559 that takes the outer and inner modes.
1560 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1561 comparison as the canonical test for a paradoxical subreg.
1562 * combine.c (simplify_set): Use paradoxical_subreg_p.
1563 (make_extraction): Likewise.
1564 (force_to_mode): Likewise.
1565 (rtx_equal_for_field_assignment_p): Likewise.
1566 (gen_lowpart_for_combine): Likewise.
1567 (simplify_comparison): Likewise.
1568 * cse.c (equiv_constant): Likewise.
1569 * expmed.c (store_bit_field_1): Likewise.
1570 * final.c (alter_subreg): Likewise.
1571 * fwprop.c (propagate_rtx): Likewise.
1572 (forward_propagate_subreg): Likewise.
1573 * ira-conflicts.c (ira_build_conflicts): Likewise.
1574 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1575 * lra-constraints.c (curr_insn_transform): Likewise.
1576 (split_reg): Likewise.
1577 * lra-eliminations.c (move_plus_up): Likewise.
1578 (lra_eliminate_regs_1): Likewise.
1579 * recog.c (general_operand): Likewise.
1580 * ree.c (combine_reaching_defs): Likewise.
1581 * reload.c (push_reload): Likewise.
1582 (find_reloads): Likewise.
1583 * reload1.c (elimination_effects): Likewise.
1584 (compute_reload_subreg_offset): Likewise.
1585 (choose_reload_regs): Likewise.
1586 * rtlanal.c (subreg_lsb_1): Likewise.
1587 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1588 (simplify_subreg): Likewise.
1589 * var-tracking.c (track_loc_p): Likewise.
1590 * emit-rtl.c (byte_lowpart_offset): Likewise.
1591 (paradoxical_subreg_p): Delete out-of-line definition.
1593 2017-08-22 Jeff Law <law@redhat.com>
1595 PR tree-optimization/81741
1596 PR tree-optimization/71947
1597 * tree-ssa-dom.c: Include tree-inline.h.
1598 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1599 equivalences if one is more expensive to compute than the other.
1600 * tree-ssa-scopedtables.h (class const_or_copies): Make
1601 record_const_or_copy_raw method private.
1602 (class avail_exprs_stack): New method simplify_binary_operation.
1603 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1604 avail_exprs_stack::simplify_binary_operation as needed.
1605 (avail_exprs_stack::simplify_binary_operation): New function.
1607 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
1609 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
1610 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
1611 (DOT_SYMBOLS): Likewise.
1612 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
1613 (RELOCATABLE_NEEDS_FIXUP): Likewise.
1614 (RS6000_ABI_NAME): Likewise.
1615 (TARGET_CMODEL): Likewise.
1616 (TOC_SECTION_ASM_OP): Likewise.
1617 (SET_CMODEL): New macro.
1618 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
1620 2017-08-22 Richard Biener <rguenther@suse.de>
1622 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
1623 to allow for free-lang-data replacements similar to verify_type_variant.
1625 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * config/aarch64/aarch64.md (casesi): Use DImode rather than
1630 VOIDmode for the LABEL_REF.
1632 2017-08-22 Richard Biener <rguenther@suse.de>
1634 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
1636 2017-08-22 Richard Biener <rguenther@suse.de>
1638 * common.opt (feliminate-dwarf2-dups): Ignore.
1639 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
1640 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
1641 same_die_p_wrap, compute_section_prefix,
1642 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
1643 (comdat_symbol_id, comdat_symbol_number): Likewise.
1644 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
1646 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
1647 (output_die): Mark unreachable path unreachable.
1648 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
1649 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
1650 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
1651 (dwarf2out_early_finish): Likewise.
1653 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
1655 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
1657 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
1660 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
1661 not VAR_P. Filter attribute warnings with OPT_Wattributes.
1662 (avr_attribute_table) <io, io_low, address>: Initialize
1663 .decl_required with true.
1665 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
1667 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
1668 undocumented debugging options.
1669 (-mvsx-scalar-double): Likewise.
1670 (-mallow-df-permute): Likewise.
1671 (-mvectorize-builtins): Likewise.
1672 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1673 (rs6000_builtin_vectorized_function): Likewise.
1674 (rs6000_builtin_md_vectorized_function): Likewise.
1675 (rs6000_opt_vars): Likewise.
1677 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
1680 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
1681 (*btrq_imm): Rename from *btrq.
1682 (*btcq_imm): Rename from *btcq.
1683 (btsc): New code attribute.
1684 (*<btsc><mode>): New insn pattern.
1685 (*btr<mode>): Ditto.
1686 (*<btsc><mode>_mask): New insn_and_split pattern.
1687 (*btr<mode>_mask): Ditto.
1689 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1690 Alan Hayward <alan.hayward@arm.com>
1691 David Sherwood <david.sherwood@arm.com>
1693 * function.c (pad_below): Simplify padding calculation.
1695 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * target.def (function_prologue): Remove frame size argument.
1700 (function_epilogue): Likewise.
1701 * doc/tm.texi: Regenerate.
1702 * final.c (final_start_function): Update call to function_prologue.
1703 (final_end_function): Update call to function_epilogue.
1704 (default_function_pro_epilogue): Remove frame size argument.
1705 * output.h (default_function_pro_epilogue): Likewise.
1706 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
1707 (arm_output_function_prologue): Likewise.
1708 * config/frv/frv.c (frv_function_prologue): Likewise.
1709 (frv_function_epilogue): Likewise.
1710 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
1711 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
1712 (ia64_output_function_epilogue): Likewise.
1713 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
1714 (m32r_output_function_epilogue): Likewise.
1715 * config/microblaze/microblaze.c (microblaze_function_prologue)
1716 (microblaze_function_epilogue): Likewise.
1717 * config/mips/mips.c (mips_output_function_prologue): Likewise.
1718 (mips_output_function_epilogue): Likewise.
1719 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
1720 (mmix_target_asm_function_epilogue): Likewise.
1721 * config/msp430/msp430.c (msp430_start_function): Likewise.
1722 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
1723 (nds32_asm_function_epilogue): Likewise.
1724 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
1725 * config/pa/pa.c (pa_output_function_prologue): Likewise.
1726 (pa_output_function_epilogue): Likewise.
1727 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
1728 (rs6000_output_function_epilogue): Likewise.
1729 * config/rl78/rl78.c (rl78_start_function): Likewise.
1730 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
1731 (rs6000_output_function_epilogue): Likewise.
1732 * config/rx/rx.c (rx_output_function_prologue): Likewise.
1733 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
1734 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
1735 (sparc_asm_function_epilogue): Likewise.
1737 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1739 * tree.h (type_has_mode_precision_p): New function.
1740 * convert.c (convert_to_integer_1): Use it.
1741 * expr.c (expand_expr_real_2): Likewise.
1742 (expand_expr_real_1): Likewise.
1743 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
1744 * match.pd: Likewise.
1745 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
1746 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
1747 * tree-tailcall.c (process_assignment): Likewise.
1748 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1749 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
1750 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
1751 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1752 (vectorizable_assignment): Likewise.
1753 (vectorizable_shift): Likewise.
1754 (vectorizable_operation): Likewise.
1755 * tree-vrp.c (register_edge_assert_for_2): Likewise.
1757 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
1759 * match.pd: Add pow (C, x) simplification.
1761 2017-08-21 Richard Biener <rguenther@suse.de>
1763 PR tree-optimization/81900
1764 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
1765 for blocks with abnormal predecessors.
1766 (compute_antic): Do not set visited flag prematurely.
1768 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
1771 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
1773 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1775 * stor-layout.h (vector_type_mode): Move to...
1776 * tree.h (vector_type_mode): ...here.
1777 * stor-layout.c (vector_type_mode): Move to...
1778 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
1780 2017-08-21 Richard Biener <rguenther@suse.de>
1782 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
1783 register_external_die hooks.
1784 (debug_false_tree_charstarstar_uhwistar): Declare.
1785 (debug_nothing_tree_charstar_uhwi): Likewise.
1786 * debug.c (do_nothing_debug_hooks): Adjust.
1787 (debug_false_tree_charstarstar_uhwistar): New do nothing.
1788 (debug_nothing_tree_charstar_uhwi): Likewise.
1789 * dbxout.c (dbx_debug_hooks): Adjust.
1790 (xcoff_debug_hooks): Likewise.
1791 * sdbout.c (sdb_debug_hooks): Likewise.
1792 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1793 * dwarf2out.c (macinfo_label_base): New global.
1794 (dwarf2out_register_external_die): New function for the
1795 register_external_die hook.
1796 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
1797 (dwarf2_debug_hooks): Use them.
1798 (dwarf2_lineno_debug_hooks): Adjust.
1799 (struct die_struct): Add with_offset flag.
1800 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
1801 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
1802 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
1803 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
1804 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
1805 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
1806 defining section names for the early LTO debug variants.
1807 (reset_indirect_string): New helper.
1808 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
1809 (print_dw_val): Add support for offsetted symbol references.
1810 (get_ultimate_context): Split out from is_cxx.
1811 (is_cxx): Use get_ultimate_context.
1812 (is_fortran): Add decl overload.
1813 (compute_comp_unit_symbol): Split out worker from
1814 compute_section_prefix.
1815 (compute_section_prefix): Call compute_comp_unit_symbol and
1816 set comdat_type_p here.
1817 (output_die): Skip DIE symbol output for the LTO added one.
1818 Handle DIE symbol references with offset.
1819 (output_comp_unit): Guard section name mangling properly.
1820 For LTO debug sections emit a symbol at the section beginning
1821 which we use to refer to its DIEs.
1822 (add_abstract_origin_attribute): For DIEs registered via
1823 dwarf2out_register_external_die directly refer to the early
1824 DIE rather than indirectly through the shadow one we created.
1825 Remove obsolete call to dwarf2out_abstract_function for
1826 non-function/block origins.
1827 (gen_array_type_die): When generating early LTO debug do
1828 not emit DW_AT_string_length.
1829 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
1830 late when in LTO. As suggested place a gcc_unreachable for
1831 the DECL_ABSTRACT_P case.
1832 (gen_subprogram_die): Avoid another specification DIE
1833 for early built declarations/definitions for the late LTO case.
1834 (gen_variable_die): Add type references for late duplicated VLA dies
1836 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
1837 we have the abstract instance already.
1838 (process_scope_var): Adjust decl DIE contexts in LTO which
1839 first puts them in limbo.
1840 (gen_decl_die): Do not generate type DIEs late apart from
1841 types for VLAs or for decls we do not yet have a DIE. Do not
1842 call dwarf2out_abstract_function late.
1843 (dwarf2out_early_global_decl): Make sure to create DIEs
1844 for abstract instances of a decl first.
1845 (dwarf2out_late_global_decl): Adjust comment.
1846 (output_macinfo_op): With multiple macro sections use
1847 macinfo_label_base to distinguish labels.
1848 (output_macinfo): Likewise. Update macinfo_label_base.
1849 Pass in the line info label.
1850 (note_variable_value_in_expr): When generating LTO resolve
1851 all variable values here by generating DIEs as needed.
1852 (init_sections_and_labels): Add early LTO debug flag parameter
1853 and generate different sections and names if set. Add generation
1854 counter for the labels so we can have multiple of them.
1855 (reset_dies): Helper to allow DIEs to be output multiple times.
1856 (dwarf2out_finish): When outputting DIEs to the fat part of an
1857 LTO object first reset DIEs.
1858 (dwarf2out_early_finish): Output early DIEs when generating LTO.
1859 (modified_type_die): Check for decl_ultimate_origin being self
1861 (gen_type_die_with_usage): Likewise.
1862 (gen_typedef_die): Allow decl_ultimate_origin being self.
1863 (set_decl_abstract_flags): Remove.
1864 (set_block_abstract_flags): Likewise.
1865 (dwarf2out_abstract_function): Treat the early generated DIEs
1866 as the abstract copy and only add DW_AT_inline and
1867 DW_AT_artificial here and call set_decl_origin_self.
1868 If the DIE has an abstract origin don't do anything.
1869 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
1870 if we have none yet (Go fails to build one, PR78628).
1871 (variably_modified_type_p): Prevent endless recursion for Ada
1872 cyclic pointer types.
1873 * lto-streamer-in.c: Include debug.h.
1874 (dref_queue): New global.
1875 (lto_read_tree_1): Stream in DIE references.
1876 (lto_input_tree): Register DIE references.
1877 (input_function): Stream DECL_DEBUG_ARGS.
1878 * lto-streamer-out.c: Include debug.h.
1879 (lto_write_tree_1): Output DIE references.
1880 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
1881 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
1882 (output_function): Stream DECL_DEBUG_ARGS.
1883 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
1884 Stream DECL_ABSTRACT_ORIGIN.
1885 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
1886 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
1887 DECL_CONTEXT for file-scope decls.
1888 * lto-streamer.h (struct dref_entry): Declare.
1889 (dref_queue): Likewise.
1890 * cfgexpand.c (pass_expand::execute): Do not call the
1891 outlining_inline_function hook here.
1892 * lto-wrapper.c (debug_obj): New global.
1893 (tool_cleanup): Unlink it if required.
1894 (debug_objcopy): New function.
1895 (run_gcc): Handle early debug sections in the IL files by
1896 extracting them to separate files, partially linkin them and
1897 feeding the result back as result to the linker.
1898 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
1899 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
1900 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
1901 sections into a separate segment.
1902 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
1904 (darwin_asm_dwarf_section): Likewise.
1905 (darwin_asm_output_dwarf_offset): Likewise.
1906 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
1908 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1909 Alan Hayward <alan.hayward@arm.com>
1910 David Sherwood <david.sherwood@arm.com>
1912 * read-md.h (md_reader::record_potential_iterator_use): Replace
1913 pointer argument with an rtx and an index.
1914 * read-rtl.c (iterator_group::apply_iterator): Likewise.
1915 (apply_mode_iterator): Likewise.
1916 (apply_code_iterator): Likewise.
1917 (apply_int_iterator): Likewise.
1918 (apply_subst_iterator): Likewise.
1919 (record_iterator_use): Likewise.
1920 (record_attribute_use): Likewise.
1921 (md_reader::record_potential_iterator_use): Likewise. Update calls
1922 to record_iterator_use and apply_iterator.
1923 (iterator_use): Replace ptr with x and index.
1924 (attribute_use): Likewise.
1925 (apply_attribute_uses): Update calls to apply_iterator.
1926 (apply_iterators): Likewise. Update initialization of iterator_use.
1927 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
1928 and record_potential_iterator_use.
1929 (rtx_reader::read_rtx_operand): Likewise.
1931 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1932 Alan Hayward <alan.hayward@arm.com>
1933 David Sherwood <david.sherwood@arm.com>
1935 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
1938 2017-08-21 Richard Biener <rguenther@suse.de>
1941 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
1942 at struct end conservatively when comparing common bases.
1944 2017-08-21 Richard Biener <rguenther@suse.de>
1946 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
1947 (mem_ref_in_stmt): Remove.
1948 (determine_max_movement): Use ref index to get at the reference.
1949 (invariantness_dom_walker::before_dom_children): Deal with
1950 lim data already initialized.
1951 (gather_mem_refs_stmt): Initialize lim data and record ref index.
1953 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
1955 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
1956 (TARGET_ISA_ROUND): Ditto.
1957 (TARGET_ROUND): Ditto.
1958 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
1959 * config/i386/i386.md: Ditto.
1960 * config/i386/sse.md: Ditto.
1961 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
1962 with OPTION_MASK_ISA_SSE4_1.
1964 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
1967 * doc/extend.texi (x86 Built-in Functions): Correct the name of
1968 __builtin_ia32_lzcnt_u16.
1970 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1973 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
1974 (rs6000_set_current_function): Rewrite function to use it.
1976 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
1979 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
1980 and TYPE_WARN_IF_NOT_ALIGN.
1981 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
1982 (handle_warn_if_not_align): New.
1983 (place_union_field): Call handle_warn_if_not_align.
1984 (place_field): Call handle_warn_if_not_align.
1985 Copy TYPE_WARN_IF_NOT_ALIGN.
1986 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
1987 (layout_type): Likewise.
1988 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
1990 (tree_decl_common): Add warn_if_not_align.
1991 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
1992 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
1993 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
1994 (DECL_WARN_IF_NOT_ALIGN): Likewise.
1995 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
1996 * doc/extend.texi: Document warn_if_not_aligned attribute.
1997 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
1999 2017-08-17 Martin Liska <mliska@suse.cz>
2002 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2003 (get_data_dependence): Use it as pointer type.
2004 (distribute_loop): Likewise.
2006 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2008 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2009 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2010 (p8_vmrgow_<mode>_direct): New define_insn.
2011 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2012 handle endianness for vmrgew and vmrgow permute patterns.
2014 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2016 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2017 * config/rs6000/rs6000-cpus.def: Remove comment.
2018 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2019 (POWERPC_MASKS): Likewise.
2020 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2021 use of TARGET_VSX_TIMODE.
2022 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2023 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2024 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2025 (rs6000_option_override_internal): Remove dead code.
2026 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2027 (rs6000_legitimize_reload_address): Likewise.
2028 (rs6000_legitimate_address_p): Likewise.
2029 (rs6000_opt_masks): Delete "vsx-timode".
2030 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2031 from function comment.
2032 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2033 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2034 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2036 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2037 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2038 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2041 (define_peephole2): Likewise.
2043 2017-08-17 Martin Sebor <msebor@redhat.com>
2046 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2047 past the end of an array.
2048 (test_pp_format): Add test cases.
2050 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2052 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2053 missing ECF_NOTHROW flags.
2055 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2058 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2059 operands residing in integer registers.
2060 (*vsx_le_perm_load_<mode>): Likewise.
2061 (*vsx_le_perm_store_<mode>): Likewise.
2062 (define_peephole2): Add peepholes to optimize the above.
2064 2017-08-17 Marek Polacek <polacek@redhat.com>
2067 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2068 to mimic what shorten_compare did. Change the return type to bool.
2069 (fold_cond_expr_with_comparison): Update call to
2070 operand_equal_for_comparison_p.
2071 (fold_ternary_loc): Likewise.
2073 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2075 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2077 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2079 2017-08-17 Richard Biener <rguenther@suse.de>
2081 * tree-ssa-structalias.c (solve_graph): When propagating
2082 to successors update the graphs succ edges and avoid duplicate work.
2084 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2087 * config/i386/i386.c (ix86_option_override_internal): Save target
2088 specific options after ix86_stack_protector_guard_reg was changed.
2090 2017-08-17 Richard Biener <rguenther@suse.de>
2092 PR tree-optimization/81827
2093 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2094 (new_var_info): Initialize it conservatively.
2095 (get_call_vi): Mark register vars.
2096 (new_scalar_tmp_constraint_exp): Likewise.
2097 (handle_rhs_call): Likewise.
2098 (handle_const_call): Likewise.
2099 (create_function_info_for): Likewise.
2100 (solve_constraints): Sort varinfos to separate register from
2101 non-register vars to pack points-to solution bitmaps during
2104 2017-08-17 Marek Polacek <polacek@redhat.com>
2106 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2108 2017-08-17 Richard Biener <rguenther@suse.de>
2110 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2111 to true when overflow is undefined and we saturated the result.
2113 2017-08-17 Alan Modra <amodra@gmail.com>
2116 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2117 Don't use store multiple if only one reg needs saving.
2118 (interesting_frame_related_regno): New function.
2119 (rs6000_frame_related): Don't emit frame info for regs that
2121 (rs6000_emit_epilogue): Likewise.
2123 2017-08-16 Nathan Sidwell <nathan@acm.org>
2125 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2126 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2127 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2128 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2129 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2130 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2131 (verify_type): Adjust for TYPE_BINFO move.
2132 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2133 process TYPE_BINFO directly.
2134 (hash_tree): Likewise.
2135 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2137 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2140 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2142 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2144 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2147 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2148 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2149 Add ix86_binary_operator_ok to insn constraint.
2150 (*iordi_1_bts): Ditto.
2151 (*xordi_1_btc): Ditto.
2152 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2153 Update corresponding peephole2 pattern.
2157 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2159 PR tree-optimization/81832
2160 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2161 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2163 2017-08-16 Marek Polacek <polacek@redhat.com>
2166 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2167 regarding size_zero_node.
2169 2017-08-16 Martin Liska <mliska@suse.cz>
2172 * config.gcc: Respect previously set extra_objs in case
2175 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2177 PR tree-optimization/81835
2178 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2179 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2180 not depend on the phi.
2182 2017-08-16 Alan Modra <amodra@gmail.com>
2184 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2187 2017-08-16 Alan Modra <amodra@gmail.com>
2189 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2190 (save_reg_p): ..into this. Update all callers.
2191 (first_reg_to_save): Simplify.
2193 2017-08-16 Alan Modra <amodra@gmail.com>
2195 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2198 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2202 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2203 constant count if that count is less than 32.
2205 2017-08-15 Nathan Sidwell <nathan@acm.org>
2207 * gcc.c (execute): Emit friendlier message if inferior is killed
2208 by an external cause.
2210 2017-08-15 Richard Biener <rguenther@suse.de>
2212 PR tree-optimization/81790
2213 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2214 CONSTRUCTORs from simplifying and VN.
2216 2017-08-14 Martin Sebor <msebor@redhat.com>
2218 * builtin-attrs.def: Add comments.
2220 2017-08-14 Martin Sebor <msebor@redhat.com>
2223 * doc/extend.texi (attribute nonstring): Document new attribute.
2225 2017-08-14 Martin Sebor <msebor@redhat.com>
2228 * tree-diagnostic.c (default_tree_printer): Handle %G.
2229 * gimple-pretty-print.h (percent_G_format): Declare new function.
2230 * gimple-pretty-print.c (percent_G_format): Define.
2231 * tree-pretty-print.c (percent_K_format): Add argument.
2233 2017-08-14 Martin Sebor <msebor@redhat.com>
2235 PR translation/79998
2236 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2237 Remove a stray space.
2239 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2242 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2243 (*iordi_1_bts): Ditto.
2244 (*xordi_1_btc): Ditto.
2246 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2249 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2250 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2252 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2253 quoted strings, and make more translator-friendly.
2254 (darwin_rs6000_override_options): Likewise.
2255 (rs6000_option_override_internal): Likewise.
2256 (rs6000_return_in_memory): Fix overlong line.
2257 (init_cmulative_args): Use quoted strings, and make more
2258 translator-friendly.
2259 (rs6000_pass_by_reference): Fix overlong line.
2260 (def_builtin): Use quoted strings.
2261 (altivec_expand_predicate_builtin): Use quoted strings, and make
2262 more translator-friendly.
2263 (htm_expand_builtin): Use quoted strings.
2264 (cpu_expand_builtin): Use quoted strings, and make more
2265 translator-friendly.
2266 (altivec_expand_builtin): Likewise.
2267 (paired_expand_predicate_builtin): Likewise.
2268 (rs6000_invalid_builtin): Likewise.
2269 (builtin_function_type): Use quoted strings.
2270 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2271 more translator-friendly.
2272 (rs6000_trampoline_init): Likewise.
2273 (rs6000_handle_altivec_attribute): Likewise.
2274 (rs6000_inner_target_options): Use quoted strings.
2275 (rs6000_disable_incompatible_switches): Likewise.
2276 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2277 strings, and make more translator-friendly.
2278 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2280 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2282 PR tree-optimization/81799
2283 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2284 cond_expr to simple gimple operand.
2286 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2289 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2291 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2295 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2296 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2297 TARGET_GASISR_PROLOGUES.
2298 * config/avr/avr.c (avr_option_override): Same.
2299 (avr_pass_pre_proep::execute): Same.
2301 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2304 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2305 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2307 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2309 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2310 define_insn_and_split. Split to a memory load from 0 in
2311 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2312 using PTR mode iterator.
2313 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2314 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2315 (*add_tp_<mode>): Redefine as define_insn_and_split.
2316 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2317 address space. Merge with *add_tp_x32 using PTR mode iterator.
2318 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2319 Split to an add with a memory load from 0 in
2320 DEFAULT_TLS_SEG_REG address space.
2322 2017-08-12 Andrew Pinski <apinski@cavium.com>
2324 * config/aarch64/aarch64-option-extensions.def (rdma):
2325 Fix feature string to what Linux prints out in /proc/cpuinfo.
2327 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2330 * dwarf2out.c (modified_type_die): For C typedef types that have
2331 an ultimate origin, process the ultimate origin instead of the
2333 (gen_typedef_die): Assert that input DECLs have no ultimate
2335 (gen_type_die_with_usage): For typedef variants that have an
2336 ultimate origin, just call gen_decl_die on the original DECL.
2337 (process_scope_var): Avoid creating DIEs for local typedefs and
2338 concrete static variables.
2340 2017-08-12 Alan Modra <amodra@gmail.com>
2344 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2345 match gnu-user.h startfile.
2346 (ENDFILE_LINUX_SPEC): Similarly.
2348 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2351 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2353 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2355 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2356 * config/aarch64/aarch64.md (mov<mode>): Change.
2357 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2358 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2359 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2361 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2363 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2364 for storage order barriers.
2366 2017-08-11 Martin Liska <mliska@suse.cz>
2369 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2370 variables of void type.
2372 2017-08-11 Martin Liska <mliska@suse.cz>
2374 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2375 TARGET_SUPPORTS_ALIASES.
2376 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2377 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2378 (optimize_weakref): Likewise.
2379 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2380 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2381 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2383 2017-08-11 Martin Liska <mliska@suse.cz>
2386 * config/i386/i386.c (make_resolver_func): Do complete
2387 refactoring of the function.
2389 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2392 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2393 * config/i386/i386.c (ix86_stack_protect_guard): Use
2394 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2395 * doc/invoke.texi (x86 Options): Document
2396 -mstack-protector-guard-symbol= option.
2398 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2400 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2401 * config/i386/i386.c (ix86_split_stack_guard): New function.
2402 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2403 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2404 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2405 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2406 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2407 (split_stack_space_check): Call ix86_split_stack_guard.
2409 2017-08-10 Martin Sebor <msebor@redhat.com>
2411 * print-tree.c (print_node): Print location using the established
2413 Replace spaces with colons.
2414 (debug_raw, debug): Ditto.
2416 2017-08-10 Martin Sebor <msebor@redhat.com>
2419 * pretty-print.c (pp_format): Correct the handling of %s precision.
2421 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2424 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2426 (ix86_finalize_stack_frame_flags): This. Also clear
2427 frame_pointer_needed if -fno-omit-frame-pointer is used without
2429 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2430 with ix86_finalize_stack_frame_flags.
2431 (ix86_expand_epilogue): Likewise.
2432 (ix86_expand_split_stack_prologue): Likewise.
2433 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2435 2017-08-10 Martin Liska <mliska@suse.cz>
2438 * c-attribs.c (handle_target_attribute):
2439 Report warning for an empty string argument of target attribute.
2441 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2444 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2446 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2447 or DECL_NONLOCAL labels.
2448 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2449 or DECL_NONLOCAL labels here.
2451 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2453 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2454 to indicate when early gimple folding has been disabled.
2455 (rs6000_gimple_fold_builtin): Add debug content.
2456 (rs6000_invalid_builtin): Fix whitespace.
2457 (rs6000_expand_builtin): Fix whitespace.
2458 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2460 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2463 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2464 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2466 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2468 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2469 (qdf24xx): Likewise.
2470 * config/aarch64/aarch64.md: Include falkor.md.
2471 * config/aarch64/falkor.md: New.
2473 2017-08-09 Marek Polacek <polacek@redhat.com>
2476 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2477 * diagnostic.c (emit_diagnostic): Add a comment.
2478 (emit_diagnostic_valist): New function.
2480 2017-08-09 Marek Polacek <polacek@redhat.com>
2483 * input.c (make_location): New overload.
2484 * input.h (make_location): Declare.
2486 2017-08-08 Alan Modra <amodra@gmail.com>
2487 H.J. Lu <hongjiu.lu@intel.com>
2490 * gcc.c (NO_PIE_SPEC): Delete.
2491 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2493 (LINK_PIE_SPEC): ..to here.
2494 (LINK_COMMAND_SPEC): Support -no-pie.
2495 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2496 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2498 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2499 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2500 (ENDFILE_CRTEND_SPEC): Similarly.
2502 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2505 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2506 (mstack-protector-guard-offset=): Ditto.
2507 * config/i386/i386.c (ix86_option_override): Handle
2508 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2510 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2511 ix86_stack_protect_guard_offset variables.
2512 (TARGET_STACK_PROTECT_GUARD): Always define.
2513 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2514 and -mstack-protector-guard-offset= options.
2516 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2518 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2519 boundary case for the last candidate.
2521 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2523 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2524 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2526 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2529 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2530 * config/aarch64/aarch64-builtins.c
2531 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2532 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2533 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2535 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2536 Andrew Pinski <pinskia@gmail.com>
2539 * internal-fn.def (XORSIGN): New.
2540 * optabs.def (xorsign_optab): New.
2541 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2542 (convert_expand_mult_copysign): New.
2543 (pass_optimize_widening_mul::execute): Call
2544 convert_expand_mult_copysign.
2546 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2548 PR tree-optimization/81354
2549 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2550 Insert on edges rather than explicitly creating landing pads.
2551 (analyze_candidates_and_replace): Commit edge inserts.
2553 2017-08-08 Richard Biener <rguenther@suse.de>
2556 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2557 (expand_simple_operations): Also look through ADDR_EXPRs with
2558 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2560 2017-08-08 Richard Biener <rguenther@suse.de>
2562 PR tree-optimization/81723
2563 * tree-vect-slp.c (struct bst_traits): New hash traits.
2564 (bst_fail): New global.
2565 (vect_build_slp_tree_2): New worker, split out from ...
2566 (vect_build_slp_tree): ... this now wrapping it with using
2567 bst_fail set to cache SLP tree build fails. Properly handle
2569 (vect_analyze_slp_instance): Allocate and free bst_fail.
2571 2017-08-08 Martin Liska <mliska@suse.cz>
2574 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2575 LABEL_DECLs that can be from a different function.
2577 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2579 PR tree-optimization/81744
2580 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2581 loop's number of iterations.
2583 2017-08-08 Martin Liska <mliska@suse.cz>
2585 * asan.c: Include header files.
2586 * attribs.c (build_decl_attribute_variant): New function moved
2588 (build_type_attribute_qual_variant): Likewise.
2589 (cmp_attrib_identifiers): Likewise.
2590 (simple_cst_list_equal): Likewise.
2591 (omp_declare_simd_clauses_equal): Likewise.
2592 (attribute_value_equal): Likewise.
2593 (comp_type_attributes): Likewise.
2594 (build_type_attribute_variant): Likewise.
2595 (lookup_ident_attribute): Likewise.
2596 (remove_attribute): Likewise.
2597 (merge_attributes): Likewise.
2598 (merge_type_attributes): Likewise.
2599 (merge_decl_attributes): Likewise.
2600 (merge_dllimport_decl_attributes): Likewise.
2601 (handle_dll_attribute): Likewise.
2602 (attribute_list_equal): Likewise.
2603 (attribute_list_contained): Likewise.
2604 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2605 (lookup_attribute_by_prefix): Likewise.
2606 * bb-reorder.c: Include header files.
2607 * builtins.c: Likewise.
2608 * calls.c: Likewise.
2609 * cfgexpand.c: Likewise.
2610 * cgraph.c: Likewise.
2611 * cgraphunit.c: Likewise.
2612 * convert.c: Likewise.
2613 * dwarf2out.c: Likewise.
2614 * final.c: Likewise.
2615 * fold-const.c: Likewise.
2616 * function.c: Likewise.
2617 * gimple-expr.c: Likewise.
2618 * gimple-fold.c: Likewise.
2619 * gimple-pretty-print.c: Likewise.
2620 * gimple.c: Likewise.
2621 * gimplify.c: Likewise.
2622 * hsa-common.c: Likewise.
2623 * hsa-gen.c: Likewise.
2624 * internal-fn.c: Likewise.
2625 * ipa-chkp.c: Likewise.
2626 * ipa-cp.c: Likewise.
2627 * ipa-devirt.c: Likewise.
2628 * ipa-fnsummary.c: Likewise.
2629 * ipa-inline.c: Likewise.
2630 * ipa-visibility.c: Likewise.
2632 * lto-cgraph.c: Likewise.
2633 * omp-expand.c: Likewise.
2634 * omp-general.c: Likewise.
2635 * omp-low.c: Likewise.
2636 * omp-offload.c: Likewise.
2637 * omp-simd-clone.c: Likewise.
2638 * opts-global.c: Likewise.
2639 * passes.c: Likewise.
2640 * predict.c: Likewise.
2641 * sancov.c: Likewise.
2642 * sanopt.c: Likewise.
2643 * symtab.c: Likewise.
2644 * toplev.c: Likewise.
2645 * trans-mem.c: Likewise.
2646 * tree-chkp.c: Likewise.
2647 * tree-eh.c: Likewise.
2648 * tree-into-ssa.c: Likewise.
2649 * tree-object-size.c: Likewise.
2650 * tree-parloops.c: Likewise.
2651 * tree-profile.c: Likewise.
2652 * tree-ssa-ccp.c: Likewise.
2653 * tree-ssa-live.c: Likewise.
2654 * tree-ssa-loop.c: Likewise.
2655 * tree-ssa-sccvn.c: Likewise.
2656 * tree-ssa-structalias.c: Likewise.
2657 * tree-ssa.c: Likewise.
2658 * tree-streamer-in.c: Likewise.
2659 * tree-vectorizer.c: Likewise.
2660 * tree-vrp.c: Likewise.
2662 * ubsan.c: Likewise.
2663 * varasm.c: Likewise.
2664 * varpool.c: Likewise.
2665 * tree.c: Remove functions moved to attribs.[ch].
2667 * config/aarch64/aarch64.c: Add attrs.h header file.
2668 * config/alpha/alpha.c: Likewise.
2669 * config/arc/arc.c: Likewise.
2670 * config/arm/arm.c: Likewise.
2671 * config/avr/avr.c: Likewise.
2672 * config/bfin/bfin.c: Likewise.
2673 * config/c6x/c6x.c: Likewise.
2674 * config/cr16/cr16.c: Likewise.
2675 * config/cris/cris.c: Likewise.
2676 * config/darwin.c: Likewise.
2677 * config/epiphany/epiphany.c: Likewise.
2678 * config/fr30/fr30.c: Likewise.
2679 * config/frv/frv.c: Likewise.
2680 * config/ft32/ft32.c: Likewise.
2681 * config/h8300/h8300.c: Likewise.
2682 * config/i386/winnt.c: Likewise.
2683 * config/ia64/ia64.c: Likewise.
2684 * config/iq2000/iq2000.c: Likewise.
2685 * config/lm32/lm32.c: Likewise.
2686 * config/m32c/m32c.c: Likewise.
2687 * config/m32r/m32r.c: Likewise.
2688 * config/m68k/m68k.c: Likewise.
2689 * config/mcore/mcore.c: Likewise.
2690 * config/microblaze/microblaze.c: Likewise.
2691 * config/mips/mips.c: Likewise.
2692 * config/mmix/mmix.c: Likewise.
2693 * config/mn10300/mn10300.c: Likewise.
2694 * config/moxie/moxie.c: Likewise.
2695 * config/msp430/msp430.c: Likewise.
2696 * config/nds32/nds32-isr.c: Likewise.
2697 * config/nds32/nds32.c: Likewise.
2698 * config/nios2/nios2.c: Likewise.
2699 * config/nvptx/nvptx.c: Likewise.
2700 * config/pa/pa.c: Likewise.
2701 * config/pdp11/pdp11.c: Likewise.
2702 * config/powerpcspe/powerpcspe.c: Likewise.
2703 * config/riscv/riscv.c: Likewise.
2704 * config/rl78/rl78.c: Likewise.
2705 * config/rx/rx.c: Likewise.
2706 * config/s390/s390.c: Likewise.
2707 * config/sh/sh.c: Likewise.
2708 * config/sol2.c: Likewise.
2709 * config/sparc/sparc.c: Likewise.
2710 * config/spu/spu.c: Likewise.
2711 * config/stormy16/stormy16.c: Likewise.
2712 * config/tilegx/tilegx.c: Likewise.
2713 * config/tilepro/tilepro.c: Likewise.
2714 * config/v850/v850.c: Likewise.
2715 * config/vax/vax.c: Likewise.
2716 * config/visium/visium.c: Likewise.
2717 * config/xtensa/xtensa.c: Likewise.
2719 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2722 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
2723 constraints since the -mupper-regs-* switches have been
2725 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
2726 into a vector from a double word element that was extracted from
2727 another vector, and eliminate extra XXPERMDI instructions.
2728 (vsx_concat_<mode>_2): Likewise.
2729 (vsx_concat_<mode>_3): Likewise.
2730 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
2731 concat to allow optimizing inserts from previous extracts.
2733 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
2735 * config/i386/i386.c (ix86_stack_protect_guard): Generate
2736 memory reference to a SSP offset in TLS address space.
2737 (ix86_print_operand) <case '@'>: Remove.
2738 (ix86_print_operand_punct_valid_p): Remove '@' code.
2739 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
2741 (stack_tls_protect_set_<mode>): Remove.
2742 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
2743 (stack_tls_protect_test_<mode>): Remove.
2744 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
2746 2017-08-07 Olivier Hainque <hainque@adacore.com>
2749 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
2751 2017-08-07 Douglas Rupp <rupp@adacore.com>
2753 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
2754 variable was referenced as multidir in command.
2756 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2759 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
2762 2017-08-07 Martin Liska <mliska@suse.cz>
2764 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
2765 * config/rl78/rl78.c: Add include of attribs.h.
2766 * config/sh/sh.c: Likewise.
2767 * config/v850/v850.c: Likewise.
2769 2017-08-07 Tom de Vries <tom@codesourcery.com>
2772 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
2774 2017-08-07 Martin Liska <mliska@suse.cz>
2776 * config/mips/mips.c: Include attribs.h.
2778 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
2781 * doc/invoke.texi: Document change in behvaior for -Ofast for
2784 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
2786 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
2788 (aarch64_pop_regs): Likewise.
2789 (aarch64_gen_load_pair): Likewise.
2790 (aarch64_save_callee_saves): Likewise.
2791 (aarch64_restore_callee_saves): Likewise.
2793 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2795 * config/i386/i386.c: Revert the last change.
2797 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2800 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2802 (ix86_finalize_stack_frame_flags): This. Also clear
2803 frame_pointer_needed if -fno-omit-frame-pointer is used without
2805 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2806 with ix86_finalize_stack_frame_flags.
2807 (ix86_expand_epilogue): Likewise.
2808 (ix86_expand_split_stack_prologue): Likewise.
2810 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2813 * config/i386/i386.c (get_builtin_code_for_version): Set priority
2814 to P_AES for Westmere.
2816 2017-08-07 Jonathan Yong <10walls@gmail.com>
2818 * config/i386/mingw.opt (fset-stack-executable): Removed.
2819 * config/i386/cygming.opt (fset-stack-executable): Moved
2821 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
2823 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
2825 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
2827 2017-08-07 Marek Polacek <polacek@redhat.com>
2830 * fold-const.c (fold_indirect_ref_1): Check type_domain.
2832 2017-08-07 Martin Liska <mliska@suse.cz>
2834 * attribs.h (canonicalize_attr_name): New function.
2835 (cmp_attribs): Move from c-format.c and adjusted.
2836 (is_attribute_p): Moved from tree.h.
2837 * tree-inline.c: Add new includes.
2838 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
2839 (private_is_attribute_p): Remove.
2840 (private_lookup_attribute): Likewise.
2841 (private_lookup_attribute_by_prefix): Simplify.
2842 (remove_attribute): Use is_attribute_p.
2843 * tree.h: Remove removed declarations.
2845 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2848 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
2849 instead of computing it in the function. Formatting fix.
2850 (expand_case): Don't rely on default_edge being the first edge,
2851 clear it if removing it, pass default_edge to
2852 emit_case_dispatch_table.
2853 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
2856 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
2858 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
2859 insn in the function, emit NOP after the insn.
2861 2017-08-06 Tom de Vries <tom@codesourcery.com>
2863 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
2866 2017-08-06 Tom de Vries <tom@codesourcery.com>
2868 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
2871 2017-08-04 Yury Gribov <tetra2005@gmail.com>
2873 PR tree-optimization/57371
2874 * match.pd: New pattern.
2876 2017-08-04 Marek Polacek <polacek@redhat.com>
2879 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
2880 perform the computation in offset_int.
2882 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2884 PR tree-optimization/81136
2885 * tree-vectorizer.h: Include tree-hash-traits.h.
2886 (vec_base_alignments): New typedef.
2887 (vec_info): Add a base_alignments field.
2888 (vect_record_base_alignments): Declare.
2889 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
2891 (DR_IS_CONDITIONAL_IN_STMT): New macro.
2892 (create_data_ref): Add an is_conditional_in_stmt argument.
2893 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
2894 the is_conditional_in_stmt field.
2895 (data_ref_loc): Add an is_conditional_in_stmt field.
2896 (get_references_in_stmt): Set the is_conditional_in_stmt field.
2897 (find_data_references_in_stmt): Update call to create_data_ref.
2898 (graphite_find_data_references_in_stmt): Likewise.
2899 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
2900 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
2901 (vect_record_base_alignment): New function.
2902 (vect_record_base_alignments): Likewise.
2903 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
2904 for nested statements even if we fail to compute a misalignment.
2905 Use pooled base alignments for unconditional references.
2906 (vect_find_same_alignment_drs): Compare base addresses instead
2908 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
2909 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
2911 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2913 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
2914 Add an explicit name for the enum. Use auto_vec for slp_instances
2916 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
2918 (_bb_vec_info): Add a constructor and destructor.
2919 (vinfo_for_stmt): Return NULL for uids of -1 as well.
2920 (destroy_loop_vec_info): Delete.
2921 (vect_destroy_datarefs): Likewise.
2922 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
2923 (vec_info::vec_info): New function.
2924 (vec_info::~vec_info): Likewise.
2925 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
2926 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
2927 destroy_loop_vec_info.
2928 * tree-vect-loop.c (new_loop_vec_info): Replace with...
2929 (_loop_vec_info::_loop_vec_info): ...this.
2930 (destroy_loop_vec_info): Replace with...
2931 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
2932 the stmt_vec_infos. Leave handling of vec_info information to its
2933 destructor. Remove explicit vector releases.
2934 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
2935 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
2936 * tree-vect-slp.c (new_bb_vec_info): Replace with...
2937 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
2938 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
2939 (destroy_bb_vec_info): Replace with...
2940 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
2941 information to its destructor.
2942 (vect_slp_analyze_bb_1): Use new and delete instead of
2943 new_bb_vec_info and destroy_bb_vec_info.
2944 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
2947 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2949 * tree-data-ref.h (subscript): Add access_fn field.
2950 (data_dependence_relation): Add could_be_independent_p.
2951 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
2952 (same_access_functions): Move to tree-data-ref.c.
2953 * tree-data-ref.c (ref_contains_union_access_p): New function.
2954 (access_fn_component_p): Likewise.
2955 (access_fn_components_comparable_p): Likewise.
2956 (dr_analyze_indices): Add a reference to access_fn_component_p.
2957 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
2959 (constant_access_functions): Likewise.
2960 (add_other_self_distances): Likewise.
2961 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
2962 (initialize_data_dependence_relation): Use XCNEW and remove
2963 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
2964 of access functions that have the same type. Allow the
2965 subsequence to end with different bases in some circumstances.
2966 Record the chosen access functions in SUB_ACCESS_FN.
2967 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
2968 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
2969 (subscript_dependence_tester_1): Likewise dra and drb.
2970 (build_classic_dist_vector): Update calls accordingly.
2971 (subscript_dependence_tester): Likewise.
2972 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
2973 DDR_COULD_BE_INDEPENDENT_P.
2974 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
2975 comp_alias_ddrs instead of may_alias_ddrs.
2976 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
2978 (vect_analyze_data_ref_dependence): Use it if
2979 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
2980 distance vectors if that fails.
2981 (dependence_distance_ge_vf): New function.
2982 (vect_prune_runtime_alias_test_list): Use it. Don't clear
2983 LOOP_VINFO_MAY_ALIAS_DDRS.
2985 2017-08-04 Richard Biener <rguenther@suse.de>
2988 * fold-const.c (fold_binary_loc): Properly restrict
2989 minus_var0 && minus_var1 case when associating undefined overflow
2992 2017-08-04 Tom de Vries <tom@codesourcery.com>
2994 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
2996 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2998 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2999 Don't start diagnostic messages with a capital letter.
3000 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3002 (rs6000_invalid_builtin): Likewise.
3003 (rs6000_trampoline_init): Likewise.
3005 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3008 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3009 after setting changeable df flags.
3011 2017-08-03 Richard Biener <rguenther@suse.de>
3013 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3014 up if the use is in USE - X.
3016 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3018 * toplev.c (dumpfile.h): New include.
3019 (internal_error_reentered): New static function. Use it...
3020 (internal_error_function): ...here to handle reentered internal_error.
3022 2017-08-03 Richard Biener <rguenther@suse.de>
3025 * fold-const.c (split_tree): Add minus_var and minus_con
3026 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3027 here but always use minus_*.
3028 (associate_trees): Assert we never associate with MINUS_EXPR
3029 and NULL first operand. Do not recurse for PLUS_EXPR operands
3030 when associating as MINUS_EXPR either.
3031 (fold_binary_loc): Track minus_var and minus_con.
3033 2017-08-03 Tom de Vries <tom@codesourcery.com>
3036 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3037 ACCEL_COMPILER, apply finish_options on
3038 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3040 2017-08-03 Tom de Vries <tom@codesourcery.com>
3043 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3044 function_entry_patch_area_size > 0.
3046 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3049 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3050 instead of 10??LU, perform unit multiplication in wide_int,
3051 don't change alloc_object_size_limit if the limit is larger
3054 PR tree-optimization/81655
3055 PR tree-optimization/81588
3056 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3057 the case when ranges[i].low and high are 1 for unsigned type with
3061 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3062 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3064 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3066 * tree-vrp.h: Add include guard.
3068 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3071 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3072 (ud2): New insn pattern.
3073 * config/i386/i386.c (ix86_expand_epilogue):
3074 For naked functions, generate ud2 instead of trap insn.
3076 2017-08-02 Marek Polacek <polacek@redhat.com>
3079 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3081 2017-08-02 Tom de Vries <tom@codesourcery.com>
3082 Cesar Philippidis <cesar@codesourcery.com>
3084 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3085 Add missing edge probabilities.
3087 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3089 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3092 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3095 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3096 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3097 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3098 prologue_seq sequences - if any.
3100 2017-08-02 Richard Biener <rguenther@suse.de>
3102 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3103 via vectors if supported, integer extracts via punning if supported
3104 or otherwise vector extracts.
3106 2017-08-02 Richard Biener <rguenther@suse.de>
3108 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3110 (bitmap_insert_into_set): ... this.
3112 2017-08-02 Richard Biener <rguenther@suse.de>
3114 PR tree-optimization/81633
3116 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3118 PR tree-optimization/71752
3119 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3121 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3123 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3124 (machine_function::call_ms2sysv_pad_out): Remove field.
3125 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3126 (ix86_compute_frame_layout): Likewise.
3128 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3131 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3132 attribute with interrupt attribute.
3134 2017-08-01 Andrew Pinski <apinski@cavium.com>
3136 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3137 BIT_INSERT_EXPR's operand 1
3138 to see if the types precision matches.
3140 2017-08-01 Martin Liska <mliska@suse.cz>
3143 * builtins.c (expand_builtin_memcpy_args): Remove.
3144 (expand_builtin_memcpy): Call newly added function
3145 expand_builtin_memory_copy_args.
3146 (expand_builtin_memcpy_with_bounds): Likewise.
3147 (expand_builtin_mempcpy): Remove last argument.
3148 (expand_builtin_mempcpy_with_bounds): Likewise.
3149 (expand_builtin_memory_copy_args): New function created from
3150 expand_builtin_mempcpy_args with small modifications.
3151 (expand_builtin_mempcpy_args): Remove.
3152 (expand_builtin_stpcpy): Remove unused argument.
3153 (expand_builtin): Likewise.
3154 (expand_builtin_with_bounds): Likewise.
3156 2017-08-01 Martin Liska <mliska@suse.cz>
3159 Make mempcpy more optimal (PR middle-end/70140).
3161 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3164 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3165 __builtin_vec_cmpne verify both arguments are compatible vectors
3166 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3167 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3168 move computation of aligned to after checking the argument types.
3172 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3173 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3176 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3177 Doug Rupp <rupp@adacore.com>
3178 Olivier Hainque <hainque@adacore.com>
3180 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3181 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3183 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3184 for TARGET_OS_CPP_BUILTIN.
3185 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3186 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3188 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3189 passing required abi options to the assembler for EABI configurations.
3190 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3191 of .text.hot and .text.unlikely sections for kernel modules when
3192 using ARM style exceptions.
3193 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3194 options. Add EXTRA_CC1_SPEC.
3195 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3197 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3199 (ARM_TARGET2_DWARF_FORMAT): Define.
3200 * config/arm/t-vxworks: Adjust multilib control to removal of the
3201 Diab command line options.
3203 2017-08-01 Martin Liska <mliska@suse.cz>
3205 PR gcov-profile/81561
3206 * gcov.c (unblock): Make unblocking safe as we need to preserve
3207 index correspondence of blocks and block_lists.
3209 2017-08-01 Richard Biener <rguenther@suse.de>
3211 PR tree-optimization/81181
3212 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3213 (compute_antic): ... end of iteration here.
3215 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3217 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3218 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3219 (ftree-slp-vectorize): Likewise.
3220 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3221 can no longer be set independent of flag_tree_loop_vectorize.
3222 * omp-general.c (emp_max_vf): Likewise.
3223 * opts.c (enable_fdo_optimizations): Remove references to
3224 flag_tree_vectorize, these are now implicit.
3225 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3226 and leave it for the options machinery.
3228 2017-08-01 Martin Liska <mliska@suse.cz>
3231 * builtins.c (expand_builtin_memcpy_args): Remove.
3232 (expand_builtin_memcpy): Call newly added function
3233 expand_builtin_memory_copy_args.
3234 (expand_builtin_memcpy_with_bounds): Likewise.
3235 (expand_builtin_mempcpy): Remove last argument.
3236 (expand_builtin_mempcpy_with_bounds): Likewise.
3237 (expand_builtin_memory_copy_args): New function created from
3238 expand_builtin_mempcpy_args with small modifications.
3239 (expand_builtin_mempcpy_args): Remove.
3240 (expand_builtin_stpcpy): Remove unused argument.
3241 (expand_builtin): Likewise.
3242 (expand_builtin_with_bounds): Likewise.
3244 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3247 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3248 print "ds:" only for immediates in generic address space.
3250 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3253 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3254 (ix86_function_ok_for_sibcall): Return false for naked functions.
3256 2017-08-01 Richard Biener <rguenther@suse.de>
3258 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3259 (compute_antic): Seed worklist with exit block predecessors.
3260 * cfganal.c (dfs_find_deadend): For a cycle return the source
3261 of the edge closing it.
3263 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3265 * config/aarch64/aarch64.c
3266 (aarch64_can_const_movi_rtx_p): Move 0 check.
3268 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3270 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3271 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3273 * match.pd: Ditto in address comparison pattern.
3275 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3277 PR tree-optimization/81627
3278 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3279 closed ssa form for store-store chain.
3281 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3283 PR tree-optimization/81620
3284 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3285 for store-store chain.
3287 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3289 PR tree-optimization/81588
3290 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3291 ranges[i].in_p, invert comparison code ccode. For >/>=,
3292 swap rhs1 and rhs2 and comparison code unconditionally,
3293 for </<= don't do that. Don't swap rhs1/rhs2 again if
3294 ranges[i].in_p, instead invert comparison code ccode if
3295 opcode or oe->rank is BIT_IOR_EXPR.
3298 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3299 a direct optab to conversion optab.
3300 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3301 with GET_MODE_INNER as last argument instead of optab_handler.
3302 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3303 vector extraction if possible and optab is available.
3304 * expr.c (store_constructor): Use convert_optab_handler instead
3305 of optab_handler. Use vector initialization from smaller
3306 vectors if possible and optab is available.
3307 * tree-vect-stmts.c (vectorizable_load): Likewise.
3308 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3310 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3311 of vec_init from half-sized vectors with the same element mode.
3312 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3313 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3314 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3315 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3316 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3317 after mode in gen_vec_extract* calls.
3318 (vec_extract<mode>): Renamed to ...
3319 (vec_extract<mode><ssescalarmodelower>): ... this.
3320 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3321 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3322 element mode after mode in gen_vec_init* calls.
3323 (VEC_INIT_HALF_MODE): New mode iterator.
3324 (vec_init<mode>): Renamed to ...
3325 (vec_init<mode><ssescalarmodelower>): ... this.
3326 (vec_init<mode><ssehalfvecmodelower>): New expander.
3327 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3328 (vec_extractv2sfsf): ... this.
3329 (vec_initv2sf): Renamed to ...
3330 (vec_initv2sfsf): ... this.
3331 (vec_extractv2si): Renamed to ...
3332 (vec_extractv2sisi): ... this.
3333 (vec_initv2si): Renamed to ...
3334 (vec_initv2sisi): ... this.
3335 (vec_extractv4hi): Renamed to ...
3336 (vec_extractv4hihi): ... this.
3337 (vec_initv4hi): Renamed to ...
3338 (vec_initv4hihi): ... this.
3339 (vec_extractv8qi): Renamed to ...
3340 (vec_extractv8qiqi): ... this.
3341 (vec_initv8qi): Renamed to ...
3342 (vec_initv8qiqi): ... this.
3343 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3344 (vec_init<mode>): Renamed to ...
3345 (vec_init<mode><VEC_base_l>): ... this.
3346 (vec_extract<mode>): Renamed to ...
3347 (vec_extract<mode><VEC_base_l>): ... this.
3348 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3349 (vec_initv2sfsf): ... this.
3350 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3351 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3352 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3353 element mode after mode in gen_vec_init* calls.
3354 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3355 (vec_init<mode><Vel>): ... this.
3356 (vec_extract<mode>): Renamed to ...
3357 (vec_extract<mode><Vel>): ... this.
3358 * config/aarch64/iterators.md (Vel): New mode attribute.
3359 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3360 Add element mode after mode in gen_vec_extract* calls.
3361 * config/s390/vector.md (non_vec_l): New mode attribute.
3362 (vec_extract<mode>): Renamed to ...
3363 (vec_extract<mode><non_vec_l>): ... this.
3364 (vec_init<mode>): Renamed to ...
3365 (vec_init<mode><non_vec_l>): ... this.
3366 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3367 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3369 * config/arm/iterators.md (V_elem_l): New mode attribute.
3370 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3371 (vec_extract<mode><V_elem_l>): ... this.
3372 (vec_extractv2di): Renamed to ...
3373 (vec_extractv2didi): ... this.
3374 (vec_init<mode>): Renamed to ...
3375 (vec_init<mode><V_elem_l>): ... this.
3376 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3377 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3378 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3379 Add element mode after gen_vec_extract* calls.
3380 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3381 (vec_init<mode><unitmode>): ... this.
3382 (vec_extract<mode>): Renamed to ...
3383 (vec_extract<mode><unitmode>): ... this.
3384 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3385 (vec_init<mode><unitmode>): ... this.
3386 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3387 (vec_initv2sfsf): ... this.
3388 (vec_extractv2sf): Renamed to ...
3389 (vec_extractv2sfsf): ... this.
3390 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3391 Add element mode after gen_vec_extract* calls.
3392 * config/mips/mips.md (unitmode): New mode iterator.
3393 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3394 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3395 * config/spu/spu.md (inner_l): New mode attribute.
3396 (vec_init<mode>): Renamed to ...
3397 (vec_init<mode><inner_l>): ... this.
3398 (vec_extract<mode>): Renamed to ...
3399 (vec_extract<mode><inner_l>): ... this.
3400 * config/sparc/sparc.md (veltmode): New mode iterator.
3401 (vec_init<VMALL:mode>): Renamed to ...
3402 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3403 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3404 (vec_initv2sisi): ... this.
3405 (vec_initv2sf): Renamed to ...
3406 (vec_initv2sfsf): ... this.
3407 (vec_extractv2sf): Renamed to ...
3408 (vec_extractv2sfsf): ... this.
3409 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3410 (vec_init<mode>): Renamed to ...
3411 (vec_init<mode><VEC_base_l>): ... this.
3412 (vec_extract<mode>): Renamed to ...
3413 (vec_extract<mode><VEC_base_l>): ... this.
3414 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3415 (vec_initv2sfsf): ... this.
3416 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3417 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3418 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3419 gen_vec_init* calls.
3421 2017-08-01 Richard Biener <rguenther@suse.de>
3423 PR tree-optimization/81297
3424 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3425 TREE_OVERFLOW from INTEGER_CSTs.
3427 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3429 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3431 2017-07-31 Carl Love <cel@us.ibm.com>
3433 * config/rs6000/rs6000-c: Add support for built-in functions
3434 vector signed char vec_xl_be (signed long long, signed char *);
3435 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3436 vector signed int vec_xl_be (signed long long, signed int *);
3437 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3438 vector signed long long vec_xl_be (signed long long, signed long long *);
3439 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3440 vector signed short vec_xl_be (signed long long, signed short *);
3441 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3442 vector double vec_xl_be (signed long long, double *);
3443 vector float vec_xl_be (signed long long, float *);
3444 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3445 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3446 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3448 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3449 (altivec_expand_builtin): Add switch statement to call
3450 altivec_expand_xl_be for each builtin.
3451 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3452 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3453 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3454 __builtin_vsx_le_be_v16qi.
3455 * doc/extend.texi: Update the built-in documentation file for the
3456 new built-in functions.
3458 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3461 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3463 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3465 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3467 * config.gcc: Add z14.
3468 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3469 CPU model numbers for z13s and z14.
3470 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3472 * config/s390/s390-opts.h (enum processor_type): Rename
3473 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3474 * config/s390/s390.c (processor_table): Add field for CPU name to
3475 be passed to Binutils.
3476 (s390_asm_output_machine_for_arch): Use the new field in
3477 processor_table for Binutils.
3478 (s390_expand_builtin): Replace arch12 with z14.
3479 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3480 (s390_get_sched_attrmask): Likewise.
3481 (s390_get_unit_mask): Likewise.
3482 * config/s390/s390.opt: Add z14 to processor_type enum.
3484 2017-07-31 Martin Jambor <mjambor@suse.cz>
3487 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3488 regardless of optimization level.
3490 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3491 Martin Liska <mliska@suse.cz>
3493 * predict.def: Remove old comment and adjust probability.
3494 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3497 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3500 * config/i386/i386.c (ix86_function_naked): New function.
3501 (ix86_can_use_return_insn_p): Return false for naked functions.
3502 (ix86_expand_prologue): Skip prologue for naked functions.
3503 (ix86_expand_epilogue): Skip epilogue for naked functions
3504 and emit trap instruction.
3505 (ix86_warn_func_return): New function.
3506 (ix86_attribute_table): Add "naked" attribute specification.
3507 (TARGET_WARN_FUNC_RETURN): Define.
3508 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3510 2017-07-31 Martin Liska <mliska@suse.cz>
3512 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3513 (dump_gimple_bb_header): Always dump BB info.
3514 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3516 2017-07-31 Martin Liska <mliska@suse.cz>
3519 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3520 also with current_function_decl non-null equality.
3522 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3525 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3526 change type to the element type, instead add eltype variable and
3527 use it where we are interested in the element type.
3529 PR tree-optimization/81603
3530 * ipa-polymorphic-call.c
3531 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3532 offset arithmetic in offset_int, bail out if the resulting bit offset
3533 doesn't fit into shwi.
3535 2017-07-31 Martin Liska <mliska@suse.cz>
3537 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3538 (gimplify_save_expr): Fix comment.
3540 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3543 * config/i386/i386.c (ix86_function_arg): Update arguments for
3545 (ix86_compute_frame_layout): Set the initial stack offset to
3546 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3547 INCOMING_FRAME_SP_OFFSET.
3548 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3549 stack before exception handler returns.
3550 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3551 the 'ERROR_CODE' for exception handler.
3553 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3555 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3556 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3557 (ASM_OUTPUT_REG_POP): Ditto.
3558 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3559 instead of asm_fprintf to output pure string.
3561 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3563 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3564 to imported_module_or_decl hook.
3565 (debug_nothing_tree_tree_tree_bool): Remove.
3566 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3567 * debug.c (do_nothing_debug_hooks): Use
3568 debug_nothing_tree_tree_tree_bool_bool instead of
3569 debug_nothing_tree_tree_tree_bool.
3570 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3571 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3572 * sdbout.c (sdb_debug_hooks): Likewise.
3573 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3574 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3576 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3577 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3578 attribute, don't add anything.
3580 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3582 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3583 (fold_build2_stat_loc): Likewise.
3584 (fold_build3_stat_loc): Likewise.
3585 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3586 (fold_build1_loc): Remove macro.
3587 (fold_build2_loc): Likewise.
3588 (fold_build3_loc): Likewise.
3590 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3592 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3593 (gimple_build_debug_bind_source_stat): Likewise.
3594 * gimple.h (gimple_build_debug_bind): Remove macro.
3595 (gimple_build_debug_bind_source): Likewise.
3597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3599 * bitmap.c (bitmap_alloc): Adjust.
3600 (bitmap_gc_alloc): Likewise.
3601 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3603 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3605 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3606 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3607 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
3608 (bitmap_gc_alloc_stat): Likewise.
3609 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
3611 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3613 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
3614 * rtl.h (shallow_copy_rtx): Remove macro.
3616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3618 * emit-rtl.c (gen_raw_REG): Adjust.
3619 * gengenrtl.c (gendef): Likewise.
3620 * rtl.c (rtx_alloc_stat): Remove _stat from name.
3621 * rtl.h (rtx_alloc): Remove macro.
3623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3625 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
3626 (build_tree_list_stat): Likewise.
3627 * tree.h (build_tree_list): Remove macro.
3628 (build_tree_list_vec): Likewise.
3630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3632 * tree.c (make_vector_stat): Remove _stat from name.
3633 (build_vector_stat): Likewise.
3634 * tree.h (make_vector_stat): Remove macro.
3635 (build_vector_stat): Likewise.
3637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3639 * tree.h (build_var_debug_value): Remove prototype.
3641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3643 * tree.c (tree_cons_stat): Remove _stat from name.
3644 * tree.h (tree_cons): Remove macro.
3646 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3648 * tree.c (build_vl_exp_stat): Remove _stat from name.
3649 * tree.h (build_vl_exp): Remove macro.
3651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3653 * tree.c (build_decl_stat): Remove _stat from name.
3654 * tree.h (build_decl): Remove macro.
3656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3658 * gimple.c (gimple_build_with_ops_stat): Adjust.
3659 (gimple_alloc_stat): Remove _stat from name.
3660 * gimple.h (gimple_alloc): Remove macro.
3662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3664 * tree.c (make_tree_vec_stat): Remove _stat from name.
3665 (grow_tree_vec_stat): Likewise.
3666 * tree.h (make_tree_vec_stat): Adjust prototype.
3667 (grow_tree_vec_stat): Likewise.
3668 (make_tree_vec): Remove macro.
3669 (grow_tree_vec): Likewise.
3671 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3673 * fold-const.c (fold_build1_stat_loc): Adjust.
3674 (fold_build2_stat_loc): Likewise.
3675 (fold_build3_stat_loc): Likewise.
3676 * tree.c (build0_stat): Remove _stat from name.
3677 (build1_stat): Likewise.
3678 (build2_stat): Likewise.
3679 (build3_stat): Likewise.
3680 (build4_stat): Likewise.
3681 (build5_stat): Likewise.
3682 * tree.h (build1_loc): Remove macro, and rename _stat function
3684 (build2_loc): Likewise.
3685 (build3_loc): Likewise.
3686 (build4_loc): Likewise.
3687 (build5_loc): Likewise.
3689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3691 * tree.c (make_int_cst_stat): Remove _stat from name.
3692 * tree.h (make_int_cst_stat): Adjust prototype.
3693 (make_int_cst): Remove macro.
3695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3697 * tree.c (make_tre_binfo_stat): Remove _stat from name.
3698 * tree.h (make_tree_binfo_stat): Adjust prototype.
3699 (make_tree_binfo): Remove.
3701 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3703 * tree.c (copy_node_stat): Rename to copy_node.
3704 (build_distinct_type_copy): Adjust.
3705 * tree.h (copy_node_stat): Adjust prototype.
3706 (copy_node): Remove macro.
3708 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3710 * tree.c (make_node_stat): rename to make_node.
3711 (build_tree_list_stat): Adjust.
3712 (build0_stat): Likewise.
3713 (build2_stat): Likewise.
3714 (build3_stat): Likewise.
3715 (build4_stat): Likewise.
3716 (build5_stat): Likewise.
3717 (build_decl_stat): Likewise.
3718 * tree.h (make_node_stat): Adjust prototype.
3719 (make_node): remove macro.
3721 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
3723 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
3724 (PPC_FEATURE2_SCV): Likewise.
3725 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
3727 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3729 * config/aarch64/aarch64.c
3730 (aarch64_internal_mov_immediate): Add new special pattern.
3731 * config/aarch64/aarch64.md (*movdi_aarch64):
3732 Add reg/32bit const mov case.
3734 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3735 Richard Sandiford <richard.sandiford@linaro.org>
3737 * config/aarch64/aarch64.md (mov<mode>): Generalize.
3738 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3739 Add integer and movi cases.
3740 (movi-split-hf-df-sf split, fp16): New.
3741 (enabled): Added TARGET_FP_F16INST.
3742 * config/aarch64/iterators.md (GPF_HF): New.
3743 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
3745 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3747 * config/aarch64/aarch64.c
3748 (aarch64_simd_container_mode): Add prototype.
3749 (aarch64_expand_mov_immediate): Add HI support.
3750 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
3751 (aarch64_can_const_movi_rtx_p): New.
3752 (aarch64_preferred_reload_class):
3753 Remove restrictions of using FP registers for certain SIMD operations.
3754 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
3755 (aarch64_valid_floating_const): Add integer move validation.
3756 (aarch64_simd_imm_scalar_p): Remove.
3757 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
3758 (aarch64_legitimate_constant_p): Expand list of supported cases.
3759 * config/aarch64/aarch64-protos.h
3760 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
3761 (aarch64_reinterpret_float_as_int): New.
3762 (aarch64_simd_imm_scalar_p): Remove.
3763 * config/aarch64/constraints.md (Uvi): New.
3764 (Dd): Split into Ds and new Dd.
3765 * config/aarch64/aarch64.md (*movsi_aarch64):
3767 (*movdi_aarch64): Add SIMD mov case.
3769 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3771 * tree-predcom.c: (struct chain): Handle store-store chain in which
3772 stores for elimination only store loop invariant values.
3773 (execute_pred_commoning_chain): Ditto.
3774 (prepare_initializers_chain_store_elim): Ditto.
3775 (prepare_finalizers): Ditto.
3776 (is_inv_store_elimination_chain): New function.
3777 (initialize_root_vars_store_elim_1): New function.
3779 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3781 * tree-predcom.c: Revise general description of the pass.
3782 (enum chain_type): New enum type for store elimination.
3783 (struct chain): New field supporting store elimination.
3784 (struct component): Ditto.
3785 (dump_chain): Dump store-stores chain.
3786 (release_chain): Release resources.
3787 (split_data_refs_to_components): Compute and create component
3788 contains only stores for elimination.
3789 (get_chain_last_ref_at): New function.
3790 (make_invariant_chain): Initialization.
3791 (make_rooted_chain): Specify chain type in parameter and record it.
3792 (add_looparound_copies): Skip for store-stores chain.
3793 (determine_roots_comp): Compute type of chain and pass it to
3795 (initialize_root_vars_store_elim_2): New function.
3796 (finalize_eliminated_stores): New function.
3797 (remove_stmt): Handle store for elimination.
3798 (execute_pred_commoning_chain): Execute predictive commoning on
3800 (determine_unroll_factor): Skip unroll for store-stores chain.
3801 (prepare_initializers_chain_store_elim): New function.
3802 (prepare_initializers_chain): Hanlde store-store chain.
3803 (prepare_finalizers_chain, prepare_finalizers): New function.
3804 (tree_predictive_commoning_loop): Return integer value indicating
3805 if loop is unrolled or lcssa form is corrupted.
3806 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
3808 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3810 * tree-predcom.c (initialize_root): Delete.
3811 (execute_pred_commoning_chain): Initialize root vars and replace
3812 reference of non-combined chain directly, rather than call above
3815 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3817 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
3818 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
3820 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3822 * tree-predcom.c (struct chain): New field init_seq.
3823 (release_chain): Release init_seq.
3824 (prepare_initializers_chain): Record intialization stmts in above
3826 (insert_init_seqs): New function.
3827 (tree_predictive_commoning_loop): Call insert_init_seqs.
3829 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3831 * tree-predcom.c (determine_roots_comp): Skip trivial components.
3833 2017-07-28 Richard Biener <rguenther@suse.de>
3835 * match.pd: Remove superfluous :c.
3836 * genmatch.c (simplify::id): Add member.
3837 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
3839 (current_id): New global.
3840 (dt_node::parent): Move from ...
3841 (dt_operand::parent): ... here. Add for_id member.
3842 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
3843 (decision_tree::find_node): Relax order requirement when
3844 merging DT_TRUE nodes to ones inbetween the current simplify
3845 and the one we try to merge with. Add diagnostic whenever
3846 we need to enforce pattern order by not merging.
3847 (decision_tree::insert): Set current_id.
3848 (decision_tree::print_node): Dump parent node and for_id.
3849 (parser::last_id): Add member.
3850 (parser::push_simplify): Assign unique id.
3851 (parser::parser): Initialize last_id.
3853 2017-07-28 Martin Liska <mliska@suse.cz>
3856 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
3857 gimple_build_debug_bind.
3859 2017-07-28 Richard Biener <rguenther@suse.de>
3861 PR tree-optimization/81502
3862 * match.pd: Add pattern combining BIT_INSERT_EXPR with
3864 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
3866 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
3867 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
3868 for BIT_FIELD_REF args.
3869 * fold-const.c (make_bit_field_ref): Likewise.
3870 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3872 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3875 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
3876 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
3877 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
3878 Or it into SANITIZER_UNDEFINED.
3879 * ubsan.c: Include gimple-fold.h and varasm.h.
3880 (ubsan_expand_ptr_ifn): New function.
3881 (instrument_pointer_overflow): New function.
3882 (maybe_instrument_pointer_overflow): New function.
3883 (instrument_object_size): Formatting fix.
3884 (pass_ubsan::execute): Call instrument_pointer_overflow
3885 and maybe_instrument_pointer_overflow.
3886 * internal-fn.c (expand_UBSAN_PTR): New function.
3887 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
3888 * sanitizer.def (__ubsan_handle_pointer_overflow,
3889 __ubsan_handle_pointer_overflow_abort): New builtins.
3890 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
3891 * internal-fn.def (UBSAN_PTR): New internal function.
3892 * opts.c (sanitizer_opts): Add pointer-overflow.
3893 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
3894 * fold-const.c (build_range_check): Compute pointer range check in
3895 integral type if pointer arithmetics would be needed. Formatting
3898 2017-07-28 Martin Liska <mliska@suse.cz>
3901 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
3902 parameters that are of a variable-length.
3904 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
3906 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
3908 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
3909 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
3910 (CRT_CALL_STATIC_FUNCTION): Likewise.
3911 (ASM_DEFAULT_SPEC): New define.
3912 (ASM_SPEC32): Likewise.
3913 (ASM_SPEC64): Likewise.
3914 (ASM_SPEC_COMMON): Likewise.
3915 (ASM_SPEC): Likewise.
3916 (INVALID_64BIT): Likewise.
3917 (LINK_OS_DEFAULT_SPEC): Likewise.
3918 (LINK_OS_SPEC32): Likewise.
3919 (LINK_OS_SPEC64): Likewise.
3920 (POWERPC_LINUX): Likewise.
3921 (PTRDIFF_TYPE): Likewise.
3922 (RESTORE_FP_PREFIX): Likewise.
3923 (RESTORE_FP_SUFFIX): Likewise.
3924 (SAVE_FP_PREFIX): Likewise.
3925 (SAVE_FP_SUFFIX): Likewise.
3926 (SIZE_TYPE): Likewise.
3927 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
3928 (TARGET_64BIT): Likewise.
3929 (TARGET_64BIT): Likewise.
3930 (TARGET_AIX): Likewise.
3931 (WCHAR_TYPE_SIZE): Likewise.
3932 (WCHAR_TYPE): Undefine.
3933 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
3934 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
3935 (CPP_OS_RTEMS_SPEC): Delete.
3936 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
3937 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
3939 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
3941 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3943 PR tree-optimization/81578
3944 * tree-parloops.c (build_new_reduction): Bail out if
3945 reduction_code isn't one of the standard OpenMP reductions.
3946 Move the details printing after that decision.
3948 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3950 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
3951 related to reload_in_progress.
3952 (splat_input_operand): Likewise.
3953 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
3955 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
3957 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
3958 (TARGET_INSTANTIATE_DECLS): Likewise.
3959 (legitimate_indexed_address_p): Delete reload_in_progress code.
3960 (rs6000_debug_legitimate_address_p): Likewise.
3961 (rs6000_eliminate_indexed_memrefs): Likewise.
3962 (rs6000_emit_le_vsx_store): Likewise.
3963 (rs6000_emit_move_si_sf_subreg): Likewise.
3964 (rs6000_emit_move): Likewise.
3965 (register_to_reg_type): Likewise.
3966 (rs6000_pre_atomic_barrier): Likewise.
3967 (rs6000_machopic_legitimize_pic_address): Likewise.
3968 (rs6000_allocate_stack_temp): Likewise.
3969 (rs6000_address_for_fpconvert): Likewise.
3970 (rs6000_address_for_altivec): Likewise.
3971 (rs6000_secondary_memory_needed_rtx): Delete function.
3972 (rs6000_check_sdmode): Likewise.
3973 (rs6000_alloc_sdmode_stack_slot): Likewise.
3974 (rs6000_instantiate_decls): Likewise.
3975 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
3976 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
3977 Delete reload_in_progress.
3978 (*vec_reload_and_plus_<mptrsize>): Likewise.
3979 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
3980 (vsx_div_v2di): Likewise.
3981 (vsx_udiv_v2di): Likewise.
3983 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3985 * config/rs6000/rs6000.opt (mlra): Replace with stub.
3986 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
3987 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
3988 (rs6000_debug_reg_global): Delete print of LRA status.
3989 (rs6000_option_override_internal): Delete dead LRA related code.
3990 (rs6000_lra_p): Delete function.
3991 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
3993 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
3995 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
3996 * config/riscv/rtems.h: New file.
3998 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3999 Sudakshina Das <sudi.das@arm.com>
4001 * config/aarch64/aarch64.md
4002 (define_split for and<mode>3nr_compare): Move
4003 non aarch64_logical_operand to a register.
4004 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4005 register immediate operand to a register.
4006 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4008 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4011 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4013 2017-07-27 Richard Biener <rguenther@suse.de>
4015 PR tree-optimization/81573
4016 PR tree-optimization/81494
4017 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4018 multi defuse cycle case.
4020 2017-07-27 Richard Biener <rguenther@suse.de>
4022 PR tree-optimization/81571
4023 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4026 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4028 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4029 earlier and only if MASK_FPU is set. Adjust formatting.
4031 2017-07-27 Martin Liska <mliska@suse.cz>
4033 * opt-functions.awk: Add validation of value of Init.
4034 * optc-gen.awk: Pass new argument.
4036 2017-07-27 Martin Liska <mliska@suse.cz>
4038 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4039 Fix wrong condition.
4041 2017-07-27 Martin Liska <mliska@suse.cz>
4043 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4044 BBs and edges seen by autoFDO.
4046 2017-07-27 Richard Biener <rguenther@suse.de>
4048 PR tree-optimization/81502
4049 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4050 with incompatible but same sized type.
4051 (execute_update_addresses_taken): Likewise.
4053 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4055 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4056 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4058 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4061 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4062 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4063 Change s_operand to memory_operand.
4065 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4067 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4068 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4069 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4070 Emit instructions rather than returning an expression. Handle TFmode
4071 and KFmode by casting to TImode.
4072 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4073 (rs6000_emit_le_vsx_store): Likewise.
4074 * config/rs6000/vsx.md (VSX_TI): New iterator.
4075 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4076 (*vsx_le_undo_permute_<mode>): Likewise.
4077 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4078 emit the split sequence.
4079 (*vsx_le_perm_store_<mode>): Likewise.
4081 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4083 PR tree-optimization/81555
4084 PR tree-optimization/81556
4085 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4086 if true, force CHANGED for the recursive invocation.
4087 (reassociate_bb): Remember original length of ops array, pass
4088 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4090 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4091 attributes for noipa attribute. For naked attribute use
4092 lookup_attribute first before lookup_attribute_spec.
4093 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4095 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4096 for functions with noipa attribute.
4097 (cgraph_externally_visible_p): Return true for functions with noipa
4099 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4100 for functions with noipa attribute.
4101 * doc/extend.texi: Document noipa function attribute.
4102 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4103 also for functions with noipa attribute.
4104 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4106 2017-07-26 Andrew Pinski <apinski@cavium.com>
4108 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4109 vec_unalign_load_cost and vec_unalign_store_cost.
4111 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4113 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4114 -mvsx-small-integer option.
4115 (ISA_3_0_MASKS_IEEE): Likewise.
4116 (OTHER_VSX_VECTOR_MASKS): Likewise.
4117 (POWERPC_MASKS): Likewise.
4118 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4119 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4120 code, only testing for DImode being allowed in non-VSX floating
4122 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4123 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4125 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4126 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4127 TARGET_P8_VECTOR test.
4128 (rs6000_secondary_reload_simple_move): Likewise.
4129 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4130 since TARGET_P9_VECTOR was already tested.
4131 (rs6000_opt_masks): Remove -mvsx-small-integer.
4132 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4133 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4135 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4136 test for TARGET_VEXTRACTUB was used, and that uses
4138 (p9 extract splitter): Likewise.
4139 (vsx_extract_<mode>_di_p9): Likewise.
4140 (vsx_extract_<mode>_store_p9): Likewise.
4141 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4142 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4143 the elimination of TARGET_VSX_SMALL_INTEGER.
4144 (vsx_extract_<mode>_p8): Likewise.
4145 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4146 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4147 (vsx_set_<mode>_p9): Likewise.
4148 (vsx_set_v4sf_p9): Likewise.
4149 (vsx_set_v4sf_p9_zero): Likewise.
4150 (vsx_insert_extract_v4sf_p9): Likewise.
4151 (vsx_insert_extract_v4sf_p9_2): Likewise.
4152 * config/rs6000/rs6000.md (sign extend splitter): Change
4153 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4154 (floatsi<mode>2_lfiwax_mem): Likewise.
4155 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4156 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4157 since a test for TARGET_P9_VECTOR was used.
4158 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4159 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4160 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4161 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4162 TARGET_P8_VECTOR test.
4163 (fix_trunc<mode>si2_stfiwx): Likewise.
4164 (fix_trunc<mode>si2_internal): Likewise.
4165 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4166 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4168 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4169 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4170 TARGET_P8_VECTOR test.
4171 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4172 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4173 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4175 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4176 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4177 since a test for TARGET_P9_VECTOR was used.
4178 (splitter for loading small constants): Likewise.
4180 2017-07-26 Andrew Pinski <apinski@cavium.com>
4182 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4185 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4188 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4189 (fp_valid_at): Likewise.
4191 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4193 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4194 (qdf24xx_addrcost_table): Likewise.
4195 (cortexa57_tunings): Update to use generic_branch_cost.
4196 (cortexa72_tunings): Likewise.
4197 (cortexa73_tunings): Likewise.
4198 (qdf24xx_tunings): Likewise.
4200 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4202 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4203 (thunderx2t99_branch_cost): Likewise.
4204 (cortexa35_tunings): Update to use generic_branch_cost.
4205 (cortexa53_tunings): Likewise.
4206 (cortexa57_tunings): Likewise.
4207 (cortexa72_tunings): Likewise.
4208 (cortexa73_tunings): Likewise.
4209 (thunderx2t99_tunings): Likewise.
4211 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4213 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4214 (sparc_option_override): Honour MASK_FSMULD.
4215 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4216 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4217 * config/sparc/sparc.opt (mfsmuld): New option.
4218 * doc/invoke.texi (mfsmuld): Document option.
4220 2017-07-26 Marek Polacek <polacek@redhat.com>
4223 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4225 2017-07-26 Richard Biener <rguenther@suse.de>
4227 * gimple-match-head.c (do_valueize): Return OP if valueize
4229 (get_def): New helper to get at the def stmt of a SSA name
4231 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4232 do_valueize to get at the def stmt.
4233 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4235 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4238 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4240 2017-07-26 Martin Liska <mliska@suse.cz>
4243 * function.c (expand_function_start): Make expansion of
4244 nonlocal_goto_save_area after parm_birth_insn.
4246 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4248 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4249 from all CPU target flags enable members.
4251 2017-07-26 Richard Biener <rguenther@suse.de>
4253 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4254 (decision_tree::gen): Make 'type' const.
4255 (write_predicate): Likewise.
4257 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4259 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4260 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4261 (rs6000_option_override_internal): Likewise.
4262 (rs6000_expand_vector_set): Likewise.
4263 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4264 (TARGET_UPPER_REGS_SF): Likewise.
4265 (TARGET_UPPER_REGS_DI): Likewise.
4266 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4267 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4268 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4269 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4270 (Splitters for DI constants in Altivec registers): Likewise.
4271 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4272 (vsx_set_v4sf_p9): Likewise.
4273 (vsx_set_v4sf_p9_zero): Likewise.
4274 (vsx_insert_extract_v4sf_p9): Likewise.
4275 (vsx_insert_extract_v4sf_p9_2): Likewise.
4277 2017-07-25 Carl Love <cel@us.ibm.com>
4279 * doc/extend.texi: Update the built-in documentation file for the
4280 existing built-in functions
4281 vector signed char vec_cnttz (vector signed char);
4282 vector unsigned char vec_cnttz (vector unsigned char);
4283 vector signed short vec_cnttz (vector signed short);
4284 vector unsigned short vec_cnttz (vector unsigned short);
4285 vector signed int vec_cnttz (vector signed int);
4286 vector unsigned int vec_cnttz (vector unsigned int);
4287 vector signed long long vec_cnttz (vector signed long long);
4288 vector unsigned long long vec_cnttz (vector unsigned long long);
4290 2017-07-25 Andrew Pinski <apinski@cavium.com>
4292 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4293 accesses where the use is for the first operand of a BIT_INSERT.
4295 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4298 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4299 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4301 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4303 * config/i386/gstabs.h: Delete.
4304 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4306 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4308 * config/i386/i386.c (ix86_decompose_address): Do not check for
4309 register RTX when looking at index_reg or base_reg.
4310 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4312 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4314 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4315 to update EH info here.
4317 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4319 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4321 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4323 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4325 2017-07-25 Torsten Duwe <duwe@suse.de>
4327 * common.opt: Introduce -fpatchable-function-entry
4328 command line option, and its variables function_entry_patch_area_size
4329 and function_entry_patch_area_start.
4330 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4331 including a two-value parser.
4332 * target.def (print_patchable_function_entry): New target hook.
4333 * targhooks.h (default_print_patchable_function_entry): New function.
4334 * targhooks.c (default_print_patchable_function_entry): Likewise.
4335 * toplev.c (process_options): Switch off IPA-RA if
4336 patchable function entries are being generated.
4337 * varasm.c (assemble_start_function): Look at the
4338 patchable-function-entry command line switch and current
4339 function attributes and maybe generate NOP instructions by
4340 calling the print_patchable_function_entry hook.
4341 * doc/extend.texi: Document patchable_function_entry attribute.
4342 * doc/invoke.texi: Document -fpatchable_function_entry
4343 command line option.
4344 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4346 * doc/tm.texi: Re-generate.
4348 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4351 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4352 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4354 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4356 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4358 2017-07-25 Richard Biener <rguenther@suse.de>
4360 PR tree-optimization/81455
4361 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4362 not walk in cycles when looking for guards.
4364 2017-07-25 Richard Biener <rguenther@suse.de>
4366 PR tree-optimization/81529
4367 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4368 when optimizing backedge uses.
4370 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4372 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4374 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4375 to dl_section_ref. On AIX, append an expression to subtract
4376 the size of the section length to dl_section_ref.
4378 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4380 * configure.ac: If any of the config.* scripts fail, exit 1.
4381 * configure: Regenerate.
4383 2017-07-25 Richard Biener <rguenther@suse.de>
4386 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4387 of immediate uses, be more verbose on errors.
4389 2017-07-25 Richard Biener <rguenther@suse.de>
4391 PR tree-optimization/81510
4392 * tree-vect-loop.c (vect_is_simple_reduction): When the
4393 reduction stmt is not inside the loop bail out.
4395 2017-07-25 Richard Biener <rguenther@suse.de>
4397 PR tree-optimization/81303
4398 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4399 profitability check against LOOP_VINFO_NITERSM1.
4401 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4403 * domwalk.c (cmp_bb_postorder): Simplify.
4404 (sort_bbs_postorder): New function. Use it...
4405 (dom_walker::walk): ...here to optimize common cases.
4407 2017-07-25 Martin Liska <mliska@suse.cz>
4410 * ipa-visibility.c (function_and_variable_visibility): Make the
4411 redirection just on target that supports aliasing.
4412 Fix GNU coding style.
4414 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4417 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4419 * config/arm/rtems.h: Likewise.
4420 * config/bfin/rtems.h: Likewise.
4421 * config/i386/rtemself.h: Likewise.
4422 * config/lm32/rtems.h: Likewise.
4423 * config/m32c/rtems.h: Likewise.
4424 * config/m68k/rtemself.h: Likewise.
4425 * config/microblaze/rtems.h: Likewise.
4426 * config/mips/rtems.h: Likewise.
4427 * config/moxie/rtems.h: Likewise.
4428 * config/nios2/rtems.h: Likewise.
4429 * config/powerpcspe/rtems.h: Likewise.
4430 * config/rs6000/rtems.h: Likewise.
4431 * config/rtems.h: Likewise.
4432 * config/sh/rtems.h: Likewise.
4433 * config/sh/rtemself.h: Likewise.
4434 * config/sparc/rtemself.h: Likewise.
4436 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4439 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4440 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4441 * tree-ssa-structalias.c (alias_get_name): Same.
4443 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4446 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4447 instructions if no du chain is found.
4449 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4451 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4453 2017-07-25 Richard Biener <rguenther@suse.de>
4456 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4459 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4461 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4463 (ISA_2_7_MASKS_SERVER): Likewise.
4464 (ISA_3_0_MASKS_IEEE): Likewise.
4465 (OTHER_P8_VECTOR_MASKS): Likewise.
4466 (OTHER_VSX_VECTOR_MASKS): Likewise.
4467 (POWERPC_MASKS): Likewise.
4468 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4469 duplicate list of options.
4470 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4471 explicit -mupper-regs options.
4472 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4473 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4474 alias for -mupper-regs-df.
4475 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4476 (rs6000_init_hard_regno_mode_ok): Likewise.
4477 (rs6000_option_override_internal): Likewise.
4478 (rs6000_opt_masks): Likewise.
4479 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4480 options in terms of whether -mvsx or -mpower8-vector was used.
4481 (TARGET_UPPER_REGS_DI): Likewise.
4482 (TARGET_UPPER_REGS_SF): Likewise.
4483 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4484 -mupper-regs-* options.
4486 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4488 * passes.c (emergency_dump_function): Print some empty lines and a
4489 header before the RTL dump.
4491 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4493 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4495 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4498 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4499 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4501 2017-07-24 Carl Love <cel@us.ibm.com>
4503 * config/rs6000/rs6000-c.c: Add support for built-in functions
4504 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4505 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4506 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4507 vec_extract_fp_from_shortl): Add defines for the two builtins.
4508 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4509 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4511 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4512 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4513 * doc/extend.texi: Update the built-in documentation file for the
4514 new built-in function.
4516 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4519 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4521 * doc/generic.texi: Likewise.
4522 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4523 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4525 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4527 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4528 (aarch64_mls_elt_merge<mode>): Likewise.
4530 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4532 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4533 having __cxa_atexit.
4535 2017-07-23 Michael Collison <michael.collison@arm.com>
4537 * config/arm/arm.c (arm_option_override): Deprecate
4538 use of -mstructure-size-boundary.
4539 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4540 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4542 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4545 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4546 Reduce cost estimate for direct moves.
4548 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4551 * config/i386/i386.c (ix86_option_override_internal): Disable
4552 BMI, BMI2 and TBM instructions for -m16.
4554 2017-07-21 Carl Love <cel@us.ibm.com>
4556 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4557 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4558 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4559 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4560 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4561 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4562 VMULOSW): New enum "unspec" values.
4563 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4564 altivec_vmulosw): New patterns.
4565 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4566 VMULOSW): Add definitions.
4568 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4570 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4571 (qdf24xx): Likewise.
4572 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4573 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4574 (AARCH64_FL_V8_1): Renumber.
4575 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4576 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4577 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4578 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4579 rdma to feature modifiers list.
4581 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4584 * ipa-visibility (function_and_variable_visibility): Convert
4585 recursive PLT call to direct call if appropriate.
4587 2017-07-21 Andrew Pinski <apinski@cavium.com>
4589 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4590 operand 1 to see if the types precision matches.
4591 * fold-const.c (operand_equal_p): Likewise.
4593 2017-07-21 Richard Biener <rguenther@suse.de>
4595 PR tree-optimization/81303
4596 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4597 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4598 (vect_peeling_hash_get_lowest_cost): Adjust.
4599 (vect_enhance_data_refs_alignment): Likewise. Use
4600 vect_get_peeling_costs_all_drs to compute the penalty for no
4601 peeling to match up costs.
4603 2017-07-21 Richard Biener <rguenther@suse.de>
4605 PR tree-optimization/81500
4606 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4607 we didn't identify a reduction path.
4609 2017-07-21 Tom de Vries <tom@codesourcery.com>
4610 Cesar Philippidis <cesar@codesourcery.com>
4612 PR gcov-profile/81442
4613 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
4616 2017-07-21 Tom de Vries <tom@codesourcery.com>
4619 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
4621 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
4622 nvptx_override_options_after_change.
4624 2017-07-21 Ulrich Drepper <drepper@redhat.com>
4626 * dwarf2out.c (output_file_names): Avoid double testing for
4629 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
4631 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
4633 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4635 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
4637 (try_crossjump_to_edge): Do not punt on partitioned functions.
4639 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4641 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4642 Put all BBs reachable only via paths crossing cold region to cold
4644 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
4646 2016-07-21 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/81303
4649 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
4650 into account prologue and epilogue iterations when raising
4651 min_profitable_iters to sth at least covering one vector iteration.
4653 2017-07-21 Tamar Christina <tamar.christina@arm.com>
4655 * config/arm/arm.c (arm_test_cpu_arch_dat):
4658 2017-07-20 Nathan Sidwell <nathan@acm.org>
4660 Remove TYPE_METHODS.
4661 * tree.h (TYPE_METHODS): Delete.
4662 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
4663 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
4664 (dbxout_type_methods): Scan TYPE_FIELDS.
4665 (dbxout_type): Don't check TYPE_METHODS here.
4666 * function.c (use_register_for_decl): Always ignore register for
4667 class types when not optimizing.
4668 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
4669 * tree.c (free_lang_data_in_type): Stitch out member functions and
4670 templates from TYPE_FIELDS.
4671 (build_distinct_type_copy, verify_type_variant,
4672 verify_type): Member fns are on TYPE_FIELDS.
4673 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
4674 * tree-pretty-print.c (dump_generic_node): Likewise.
4676 2017-07-20 Jakub Jelinek <jakub@redhat.com>
4679 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
4680 V2TImode and V4TImode.
4681 (ix86_expand_vector_extract): Likewise.
4682 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
4683 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
4684 (ssescalarmode): Handle V4TImode and V2TImode.
4685 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
4686 (*vec_extractv2ti, *vec_extractv4ti): New insns.
4687 (VEXTRACTI128_MODE): New mode iterator.
4688 (splitter for *vec_extractv?ti first element): New.
4689 (VEC_INIT_MODE): New mode iterator.
4690 (vec_init<mode>): Consolidate 3 expanders into one using
4691 VEC_INIT_MODE mode iterator.
4693 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4695 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
4696 non_spilled_static_chain_regno_p.
4698 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4700 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
4702 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
4704 * bb-reorder.c (connect_traces): Allow copying of blocks within
4707 2017-07-20 Richard Biener <rguenther@suse.de>
4709 * gimple.h (gimple_phi_result): Add gphi * overload.
4710 (gimple_phi_result_ptr): Likewise.
4711 (gimple_phi_arg): Likewise. Adjust index assert to only
4712 allow actual argument accesses rather than all slots available
4714 (gimple_phi_arg_def): Add gphi * overload.
4715 * tree-phinodes.c (make_phi_node): Initialize only actual
4717 (resize_phi_node): Clear memory not covered by old node,
4718 do not initialize excess argument slots.
4719 (reserve_phi_args_for_new_edge): Initialize new argument slot
4722 2017-07-20 Bin Cheng <bin.cheng@arm.com>
4724 PR tree-optimization/81388
4726 2016-07-21 Bin Cheng <bin.cheng@arm.com>
4728 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
4729 by removing computation of may_be_zero.
4731 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4732 Tom de Vries <tom@codesourcery.com>
4735 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
4736 when gimple level profile disagrees with what RTL expander did.
4738 2017-07-20 Richard Biener <rguenther@suse.de>
4740 PR tree-optimization/61171
4741 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
4742 (vect_analyze_stmt): Add slp instance parameter.
4743 (vectorizable_reduction): Likewise.
4744 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
4745 (vect_is_simple_reduction): Deal with chains not detected
4746 as SLP reduction chain, specifically not properly associated
4747 chains containing a mix of plus/minus.
4748 (get_reduction_op): Remove.
4749 (get_initial_defs_for_reduction): Simplify, pass in whether
4750 this is a reduction chain, pass in the SLP node for the PHIs.
4751 (vect_create_epilog_for_reduction): Get the SLP instance as
4753 (vectorizable_reduction): Get the SLP instance as arg.
4754 During analysis remember the SLP node with the PHIs in the
4755 instance. Simplify getting at the vectorized reduction PHIs.
4756 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
4757 through SLP instance.
4758 (vect_slp_analyze_operations): Likewise.
4759 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
4760 (vect_transform_stmt): Likewise.
4762 2017-07-20 Tom de Vries <tom@codesourcery.com>
4764 PR tree-optimization/81489
4765 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
4766 read of phi arg location to before loop that modifies phi.
4768 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4770 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
4773 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4776 * except.c (execute): Fix ordering issue.
4778 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4780 PR rtl-optimization/81423
4781 * combine.c (make_compound_operation_int): Don't try to optimize
4782 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
4784 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4786 PR rtl-optimization/81423
4787 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
4788 with a constant that is -1 in the truncated to mode.
4790 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4792 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
4793 (determine_unlikely_bbs): ... here.
4794 * predict.h (propagate_unlikely_bbs_forward): Declare.
4795 * cfgexpand.c (pass_expand::execute): Use it.
4796 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
4798 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
4799 propagate_unlikely_bbs_forward.
4801 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4804 * except.c (maybe_add_nop_after_section_switch): New function.
4807 2017-07-19 Tom de Vries <tom@codesourcery.com>
4809 * gimple.h (gimple_phi_set_arg): Make assert more strict.
4811 2017-07-19 Tom de Vries <tom@codesourcery.com>
4813 * gimple.h (gimple_phi_arg): Make assert more strict.
4815 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
4817 * config.gcc (powerpc*-*-*): Add mmintrin.h.
4818 * config/rs6000/mmintrin.h: New file.
4819 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
4821 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4823 PR tree-optimization/81346
4824 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
4826 2017-07-19 Tom de Vries <tom@codesourcery.com>
4828 * config/nvptx/nvptx.md (VECIM): Add V2DI.
4830 2017-07-19 Tom de Vries <tom@codesourcery.com>
4832 * config/nvptx/nvptx-modes.def: Add V2DImode.
4833 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
4834 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
4835 (nvptx_output_mov_insn): Handle lack of mov.b128.
4836 (nvptx_print_operand): Handle 'H' and 'L' codes.
4837 (nvptx_vector_mode_supported): Allow V2DImode.
4838 (nvptx_preferred_simd_mode): New function.
4839 (nvptx_data_alignment): New function.
4840 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
4841 nvptx_preferred_simd_mode.
4842 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
4844 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
4846 2017-07-19 Tom de Vries <tom@codesourcery.com>
4848 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
4849 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
4850 (nvptx_vector_mode_supported): New function. Allow V2SImode.
4851 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
4852 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
4853 (mov<VECIM>_insn): New define_insn.
4854 (define_expand "mov<VECIM>): New define_expand.
4856 2017-07-19 Tom de Vries <tom@codesourcery.com>
4858 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
4860 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4862 PR tree-optimization/81346
4863 * fold-const.h (fold_div_compare, range_check_type): Declare.
4864 * fold-const.c (range_check_type): New function.
4865 (build_range_check): Use range_check_type.
4866 (fold_div_compare): No longer static, rewritten into
4867 a match.pd helper function.
4868 (fold_comparison): Don't call fold_div_compare here.
4869 * match.pd (X / C1 op C2): New optimization using fold_div_compare
4872 2017-07-19 Nathan Sidwell <nathan@acm.org>
4874 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
4875 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
4876 * tree.c (find_decls_types_r, verify_type): Use
4877 TYPE_{MIN,MAX}_VALUE_RAW.
4878 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
4879 (hash_tree): Likewise.
4880 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4882 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4885 2017-07-18 Tom de Vries <tom@codesourcery.com>
4888 * omp-expand.c (expand_omp_for_static_chunk): Handle
4889 equal-argument loop exit phi.
4891 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
4894 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
4895 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
4896 operand 2 predicate.
4897 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
4898 operand 2 predicate.
4899 (ror,rol -> rorx splitters): Use const_int_operand as
4900 operand 2 predicate.
4902 2017-06-18 Richard Biener <rguenther@suse.de>
4904 PR tree-optimization/81410
4905 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
4906 the gap in the ! slp_perm SLP case after each group.
4908 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4911 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
4914 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4917 * predict.c (set_even_probabilities): Cleanup; do not affect
4918 probabilities that are already known.
4919 (combine_predictions_for_bb): Call even when count is set.
4921 2017-07-18 Nathan Sidwell <nathan@acm.org>
4923 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
4926 2017-07-18 Bin Cheng <bin.cheng@arm.com>
4929 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
4930 optimization for loop niter analysis.
4932 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
4935 * config/avr/avr.c (avr_optimize_casesi): Don't use
4936 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
4938 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4940 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
4941 body_cost_vec from _vect_peel_extended_info.
4942 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
4943 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
4946 2017-07-18 Bin Cheng <bin.cheng@arm.com>
4948 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
4950 2017-07-18 Richard Biener <rguenther@suse.de>
4952 PR tree-optimization/80620
4953 PR tree-optimization/81403
4954 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
4955 info when re-using a VN table entry.
4957 2017-07-18 Richard Biener <rguenther@suse.de>
4959 PR tree-optimization/81418
4960 * tree-vect-loop.c (vectorizable_reduction): Properly compute
4961 vectype_in. Verify that with lane-reducing reduction operations
4962 we have a single def-use cycle.
4964 2017-07-17 Carl Love <cel@us.ibm.com>
4966 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
4968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4969 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4970 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4971 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4972 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4973 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4974 VMULOSW): New enum "unspec" values.
4975 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4976 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4977 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4978 altivec_vmulosw): New patterns.
4979 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4980 VMULOSW): Add definitions.
4981 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
4983 * config/alpha/alpha.c: Include predict.h.
4985 2017-07-17 Yury Gribov <tetra2005@gmail.com>
4987 * tree-vrp.c (compare_assert_loc): Fix comparison function
4988 to return predictable results.
4990 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4992 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
4995 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
4996 * doc/invoke.texi (mexpand-adddi): Update text.
4998 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5000 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5001 that also clobbers the CC register. The old expand code is moved
5003 (*arc_clzsi2): ... here.
5004 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5005 the CC register. The old expand code is moved to ...
5006 (arc_ctzsi2): ... here.
5008 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5010 * config/arc/arc.opt (mindexed-loads): Use initial value
5011 TARGET_INDEXED_LOADS_DEFAULT.
5012 (mauto-modify-reg): Use initial value
5013 TARGET_AUTO_MODIFY_REG_DEFAULT.
5014 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5015 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5016 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5017 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5019 2017-07-17 Martin Liska <mliska@suse.cz>
5022 * opts.c (finish_options): Do not allow -fgnu-tm
5023 w/ -fsanitize={kernel-,}address. Say sorry.
5025 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5028 * tree-loop-distribution.c (classify_partition): Only assert on
5029 numer of iterations.
5030 (merge_dep_scc_partitions): Delete prameter. Update function call.
5031 (distribute_loop): Remove code handling loop with unknown niters.
5032 (pass_loop_distribution::execute): Skip loop with unknown niters.
5034 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5037 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5038 function sort_partitions_by_post_order.
5040 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5042 PR tree-optimization/81374
5043 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5044 the max index of basic blocks, rather than number of basic blocks.
5046 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5048 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5050 (arc_legitimate_pic_operand_p): Likewise.
5051 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5053 (arc_needs_pcl_p): Likewise.
5054 (arc_legitimate_pc_offset_p): Likewise.
5055 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5056 function is also used in constrains.md.
5057 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5058 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5059 PLUS. Only return true/false in known cases, otherwise assert.
5060 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5061 is already called in arc_legitimate_constant_p.
5062 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5064 (LEGITIMATE_PIC_OPERAND_P): Use
5065 arc_raw_symbolic_reference_mentioned_p function.
5066 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5071 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5072 Andrew Burgess <andrew.burgess@embecosm.com>
5074 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5075 (arc_return_address_register): New function.
5076 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5077 (arc_handle_fndecl_attribute): Add naked attribute.
5078 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5079 (TARGET_WARN_FUNC_RETURN): Likewise.
5080 (arc_allocate_stack_slots_for_args): New function.
5081 (arc_warn_func_return): Likewise.
5082 (machine_function): Change type fn_type.
5083 (arc_compute_function_type): Consider new naked function type,
5084 change function return type.
5085 (arc_must_save_register): Adapt to handle new
5086 arc_compute_function_type's return type.
5087 (arc_expand_prologue): Likewise.
5088 (arc_expand_epilogue): Likewise.
5089 (arc_return_address_regs): Delete.
5090 (arc_return_address_register): New function.
5091 (arc_epilogue_uses): Use above function.
5092 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5093 (arc_function_type): Change encoding, add naked type.
5094 (ARC_INTERRUPT_P): Change to handle the new encoding.
5095 (ARC_FAST_INTERRUPT_P): Likewise.
5096 (ARC_NORMAL_P): Define.
5097 (ARC_NAKED_P): Likewise.
5098 (arc_compute_function_type): Delete prototype.
5099 * config/arc/arc.md (in_ret_delay_slot): Use
5100 arc_return_address_register function.
5101 (simple_return): Likewise.
5102 (p_return_i): Likewise.
5104 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5106 PR tree-optimization/81428
5107 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5108 can't be built for those types.
5110 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5112 Remove stuff dead since r239246.
5114 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5115 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5116 (avr_inform_devices): Remove dead stuff.
5118 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5120 * config/arm/arm_neon.h: Fix softp typo.
5122 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5124 PR tree-optimization/81365
5125 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5126 aggregate moves onto bb predecessor edges, make sure there are no
5127 loads that could alias the lhs in between the start of bb and the
5130 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5133 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5134 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5135 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5137 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5139 PR tree-optimization/81396
5140 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5141 (init_symbolic_number): Initialize it to 1.
5142 (perform_symbolic_merge): Add n_ops from both operands into the new
5144 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5145 without base_addr as useless if they need more than one operation.
5146 (bswap_replace): Handle !bswap case for NULL base_addr.
5148 2017-07-17 Tom de Vries <tom@codesourcery.com>
5151 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5154 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5156 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5157 conditional builtin define __FIX_LEON3FT_B2BST.
5159 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5161 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5162 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5165 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5167 PR rtl-optimization/81424
5168 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5169 to remove potential trapping from operands if -fnon-call-exceptions.
5171 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5173 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5174 profile_proability for scalling.
5175 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5177 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5179 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5181 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5183 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5184 fixpoint arithmetics.
5186 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5188 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5189 fixpoint arithmetics.
5191 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5193 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5194 fixpoint arithmetics.
5196 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5198 * profile-count.h (profile_probability::from_reg_br_prob_note,
5199 profile_probability::to_reg_br_prob_note): New functions.
5200 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5201 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5202 * predict.c (probability_reliable_p): Update.
5203 (edge_probability_reliable_p): Update.
5204 (br_prob_note_reliable_p): Update.
5205 (invert_br_probabilities): Update.
5206 (add_reg_br_prob_note): New function.
5207 (combine_predictions_for_insn): Update.
5208 * asan.c (asan_clear_shadow): Update.
5209 * cfgbuild.c (compute_outgoing_frequencies): Update.
5210 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5211 (update_br_prob_note): Update.
5212 (rtl_verify_edges): Update.
5213 (purge_dead_edges): Update.
5214 (fixup_reorder_chain): Update.
5215 * emit-rtl.c (try_split): Update.
5216 * ifcvt.c (cond_exec_process_insns): Update.
5217 (cond_exec_process_if_block): Update.
5218 (dead_or_predicable): Update.
5219 * internal-fn.c (expand_addsub_overflow): Update.
5220 (expand_neg_overflow): Update.
5221 (expand_mul_overflow): Update.
5222 * loop-doloop.c (doloop_modify): Update.
5223 * loop-unroll.c (compare_and_jump_seq): Update.
5224 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5225 * predict.h: Update.
5226 * reorg.c (mostly_true_jump): Update.
5228 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5229 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5230 * config/arc/arc.c: (emit_unlikely_jump): Update.
5231 * config/arm/arm.c: (emit_unlikely_jump): Update.
5232 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5233 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5234 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5235 (ix86_print_operand): Update.
5236 (ix86_split_fp_branch): Update.
5237 (predict_jump): Update.
5238 * config/ia64/ia64.c (ia64_print_operand): Update.
5239 * config/mmix/mmix.c (mmix_print_operand): Update.
5240 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5241 (rs6000_expand_split_stack_prologue): Update.
5242 * config/rs6000/rs6000.c: Update.
5243 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5244 (s390_expand_vec_movstr): Update.
5245 (s390_expand_cs_tdsi): Update.
5246 (s390_expand_split_stack_prologue): Update.
5247 * config/sh/sh.c (sh_print_operand): Update.
5248 (expand_cbranchsi4): Update.
5249 (expand_cbranchdi4): Update.
5250 * config/sparc/sparc.c (output_v9branch): Update.
5251 * config/spu/spu.c (get_branch_target): Update.
5252 (ea_load_store_inline): Update.
5253 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5254 * config/tilepro/tilepro.c: Update.
5256 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5258 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5259 (gimplify_save_expr): Likewise.
5261 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5263 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5265 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5267 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5269 * timevar.def (TV_IPA_FNSUMMARY): Define.
5271 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5273 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5274 to back store errata sensitive sequence from being generated.
5275 (sqrtdf2_fix): Likewise.
5277 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5279 * tree-ssa-threadupdate.c (compute_path_counts,
5280 update_joiner_offpath_counts): Use profile_probability.
5282 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5285 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5287 * config/arm/arm-c.c (arm_cpu_builtins): Define
5288 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5290 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5292 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5293 array entries to represent __ieee128 versions of the
5294 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5295 scalar_extract_sig, and scalar_insert_exp built-in functions.
5296 (altivec_resolve_overloaded_builtin): Add special case handling
5297 for the __builtin_scalar_insert_exp function, as represented by
5298 the P9V_BUILTIN_VEC_VSIEDP constant.
5299 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5300 exponent support for __ieee128 argument.
5301 (VSESQP): Add scalar extract signature support for __ieee128
5303 (VSTDCNQP): Add scalar test negative support for __ieee128
5305 (VSIEQP): Add scalar insert exponent support for __int128 argument
5306 with __ieee128 result.
5307 (VSIEQPF): Add scalar insert exponent support for __ieee128
5308 argument with __ieee128 result.
5309 (VSTDCQP): Add scalar test data class support for __ieee128
5311 (VSTDCNQP): Add overload support for scalar test negative with
5313 (VSTDCQP): Add overload support for scalar test data class
5315 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5317 (UNSPEC_VSX_SIEXPQP): New constant.
5318 (xsxexpqp): New insn for VSX scalar extract exponent quad
5320 (xsxsigqp): New insn for VSX scalar extract significand quad
5322 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5323 precision with floating point argument.
5324 (xststdcqp): New expand for VSX scalar test data class quad
5326 (xststdcnegqp): New expand for VSX scalar test negative quad
5328 (xststdcqp): New insn to match expansions for VSX scalar test data
5329 class quad precision and VSX scalar test negative quad precision.
5330 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5331 special case operand checking to enforce that second operand of
5332 VSX scalar test data class with quad precision argument is a 7-bit
5334 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5335 prototypes and descriptions of __ieee128 versions of
5336 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5337 scalar_test_data_class, and scalar_test_neg built-in functions.
5339 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5341 PR tree-optimization/81162
5342 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5343 replace a negate with an add.
5345 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5347 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5349 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5351 * config/arm/arm-c.c (arm_cpu_builtins): Define
5352 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5354 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5356 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5357 (armv8-r): Set ARM Cortex-R52 as default CPU.
5358 * config/arm/arm-tables.opt: Regenerate.
5359 * config/arm/arm-tune.md: Regenerate.
5360 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5362 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5363 extension for -mcpu=cortex-r52.
5365 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5367 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5368 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5369 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5370 (fp-armv8): Define it as FP_ARMv8 only.
5371 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5372 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5374 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5375 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5376 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5377 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5378 than TARGET_FPU_ARMV8.
5379 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5380 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5381 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5383 * config/arm/neon.md (neon_vrint): Likewise.
5384 (neon_vcvt): Likewise.
5385 (neon_<fmaxmin_op><mode>): Likewise.
5386 (<fmaxmin><mode>3): Likewise.
5387 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5388 * config/arm/predicates.md (arm_cond_move_operator): Check against
5389 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5391 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5393 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5396 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5398 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5399 loop in comment with memset.
5401 2017-07-14 Martin Liska <mliska@suse.cz>
5403 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5404 * dwarf2out.c (is_java): Remove the function.
5405 (output_pubname): Remove usage of the function.
5406 (lower_bound_default): Remove usage of DW_LANG_Java.
5407 (gen_compile_unit_die): Likewise.
5408 * gcc.c: Remove compiler defaults for .java and .zip files.
5409 * gimple-expr.c (remove_suffix): Change as there's no longer
5410 extension than 4-letter one.
5411 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5412 (gimplify_save_expr): Likewise.
5413 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5414 as it's possible even for other languages than Java.
5415 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5416 * lto-opts.c (lto_write_options): Remove reference to Java.
5417 * opts.c (strip_off_ending): Update file extension handling.
5418 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5419 * tree-eh.c (lower_resx): Likewise.
5420 * tree.c (free_lang_data_in_type): Remove dead code.
5421 (find_decls_types_r): Likewise.
5422 (build_common_builtin_nodes): Remove Java from a comment.
5423 (verify_type): Remove dead code.
5424 * varasm.c (assemble_external): Remove Java from a comment.
5426 2017-07-14 Martin Liska <mliska@suse.cz>
5428 * opts.c (finish_options): Add quotes.
5429 (common_handle_option): Likewise.
5431 2017-07-14 Martin Liska <mliska@suse.cz>
5433 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5434 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5436 * dwarf2out.c (lower_bound_default): Do not handle
5438 (gen_compile_unit_die): Likewise.
5439 * gcc.c: Remove default extension binding for GNU Pascal.
5440 * stmt.c: Remove Pascal language from a comment.
5441 * xcoffout.c: Likewise.
5443 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5446 * diagnostic-show-locus.c (fixit_cmp): New function.
5447 (layout::layout): Sort m_fixit_hints.
5448 (column_range::column_range): Assert that the values are valid.
5449 (struct char_span): New struct.
5450 (correction::overwrite): New method.
5451 (struct source_line): New struct.
5452 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5453 calls in terms of classes source_line and char_span, and
5454 correction::overwrite.
5455 (selftest::test_overlapped_fixit_printing_2): New function.
5456 (selftest::diagnostic_show_locus_c_tests): Call it.
5458 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5460 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5461 early if there is no lhs.
5463 2017-07-13 Martin Liska <mliska@suse.cz>
5465 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5466 (gen_reference_type_die): Likewise.
5467 * stor-layout.c: Remove Pascal-related comment.
5469 2017-07-13 Martin Liska <mliska@suse.cz>
5471 * opts.c (finish_options): Add quotes to error messages.
5472 (parse_sanitizer_options): Likewise.
5474 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5476 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5478 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5480 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5482 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5484 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5486 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5488 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5491 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5492 provides the hardware capability bits, define the macro
5493 __BUILTIN_CPU_SUPPORTS__.
5494 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5495 if GLIBC does not provide the hardware capability bits. Add a
5496 gcc_unreachable call if the built-in cpu function is neither
5497 __builtin_cpu_is nor __builtin_cpu_supports.
5498 (rs6000_get_function_versions_dispatcher): Change the warning
5499 that an old GLIBC is used which does not export the capability
5500 bits to be an error.
5501 * doc/extend.texi (target_clones attribute): Document the
5502 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5503 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5504 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5505 the macros defined by GCC if the newer GLIBC is available.
5507 2017-07-12 Jeff Law <law@redhat.com>
5509 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5510 remaining includes slightly.
5511 * config/riscv/riscv-builtins.c: Include profile-count.h.
5513 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5516 * config/avr/avr.c (avr_set_current_function): In diagnostic
5517 messages: Quote keywords and (parts of) identifiers.
5518 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5521 2017-07-12 Carl Love <cel@us.ibm.com>
5523 * config/rs6000/rs6000-c.c: Add support for built-in functions
5524 vector bool char vec_revb (vector bool char);
5525 vector bool short vec_revb (vector short char);
5526 vector bool int vec_revb (vector bool int);
5527 vector bool long long vec_revb (vector bool long long);
5528 * doc/extend.texi: Update the built-in documentation file for the
5529 new built-in functions.
5531 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5533 * config/s390/s390.md: Remove movcc splitter.
5535 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5537 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5538 load/store on condition.
5540 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5543 * config/avr/avr.c (avr_encode_section_info)
5544 [progmem && !TREE_READONLY]: Error if progmem object needs
5547 2017-07-11 Michael Collison <michael.collison@arm.com>
5549 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5552 2017-07-11 Carl Love <cel@us.ibm.com>
5554 * config/rs6000/rs6000-c.c: Add support for builtins
5555 vector unsigned int vec_parity_lsbb (vector signed int);
5556 vector unsigned int vec_parity_lsbb (vector unsigned int);
5557 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5558 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5559 vector unsigned long long vec_parity_lsbb (vector signed long long);
5560 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5561 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5562 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5563 * doc/extend.texi: Update the built-in documentation file for the
5564 new built-in functions.
5566 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5568 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5569 (layout::m_primary_loc): New field.
5570 (layout::layout): Initialize new field. Move location filtering
5571 logic from here to...
5572 (layout::maybe_add_location_range): ...this new method. Add
5573 support for filtering to just the lines already specified by other
5575 (layout::will_show_line_p): New method.
5576 (gcc_rich_location::add_location_if_nearby): New method.
5577 (selftest::test_add_location_if_nearby): New test function.
5578 (selftest::diagnostic_show_locus_c_tests): Call it.
5579 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5582 2017-07-11 Tom de Vries <tom@codesourcery.com>
5584 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5585 (bb_first_real_insn): New function.
5586 (nvptx_single): Add extra initialization of broadcasted condition
5589 2017-07-11 Nathan Sidwell <nathan@acm.org>
5591 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5593 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5595 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5596 Binutils doc as TEXI will mess them up.
5597 * doc/invoke.texi (AVR Options): Same here.
5599 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5601 * config/sparc/sparc.opt (mfix-ut700): New option.
5602 (mfix-gr712rc): Likewise.
5603 (sparc_fix_b2bst): New variable.
5604 * doc/invoke.texi (SPARC options): Document them.
5605 (ARM options): Fix warnings.
5606 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5607 instructions to prevent sequences that can trigger the store-store
5608 errata for certain LEON3FT processors.
5609 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
5610 (sparc_option_override): Set sparc_fix_b2bst appropriately.
5611 * config/sparc/sparc.md (fix_b2bst): New attribute.
5612 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
5614 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
5617 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
5619 (*rsqrtsf2_sse): Ditto.
5621 (div<mode>3): Macroize insn from divdf3 and divsf3
5622 using MODEF mode iterator.
5624 2017-07-10 Martin Sebor <msebor@redhat.com>
5626 PR tree-optimization/80397
5627 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
5628 instead of testing for equality to INTEGER_TYPE.
5630 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
5632 * config.gcc: Remove uclibc from arc target spec.
5634 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
5636 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
5638 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5641 * lto-wrapper.c (remove_option): New function.
5642 (merge_and_complain): Merge PIC/PIE options more realistically.
5644 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5646 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
5650 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
5651 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
5652 * config.in: Regenerate.
5653 * configure: Regenerate.
5654 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
5655 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
5656 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
5657 (TARGET_GASISR_PROLOGUES): ...target mask.
5658 * common/config/avr/avr-common.c
5659 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
5660 Set -mgas-isr-prologues.
5661 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
5662 INSERT_PASS_BEFORE for it.
5663 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
5664 * config/avr/avr.c (avr_option_override)
5665 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
5666 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
5667 (avr_attribute_table) <no_gccisr>: Add new function attribute.
5668 (avr_set_current_function) <is_no_gccisr>: Init machine field.
5669 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
5671 (make_avr_pass_pre_proep): New function.
5672 (emit_push_sfr) <treg>: Add argument to function and use it
5674 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
5675 and set machine->gasisr.yes.
5676 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
5677 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
5678 __gcc_isr.n_pushed to .L__stack_usage.
5679 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
5680 (avr_asm_final_postscan_insn): ...this new static function.
5681 * config/avr/avr.h (machine_function)
5682 <is_no_gccisr, use_L__stack_usage>: New fields.
5683 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
5684 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
5685 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
5686 (gasisr, *gasisr): New expander and insn.
5687 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5688 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
5689 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
5691 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
5693 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
5696 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5698 Move jump-tables out of .text again.
5701 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
5702 (ASM_OUTPUT_ADDR_VEC): New function.
5703 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
5704 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
5705 INSN_ADDRESSes as asm comment.
5706 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
5707 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
5708 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
5709 * config/avr/avr.md (*tablejump): Adjust comment.
5710 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
5711 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
5713 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
5714 (avr_output_addr_vec): New proto.
5715 (avr_log_t) <insn_addresses>: New field.
5717 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
5720 * config/i386/i386.c (ix86_function_arg_advance): Set
5721 outgoing_args_on_stack to true if there are outgoing arguments
5723 (ix86_function_arg): Likewise.
5724 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
5725 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
5726 * config/i386/i386.h (machine_function): Add
5727 outgoing_args_on_stack.
5729 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
5731 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5733 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
5735 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
5737 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
5738 (REAL_H): Remove $(MACHMODE_H).
5739 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
5741 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
5742 $(MACHMODE_H) and double-int.h.
5743 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
5745 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
5748 2017-07-07 Andrew Pinski <apinski@cavium.com>
5750 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
5751 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
5753 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5755 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
5756 Add warning if GCC was not configured to link against a GLIBC that
5757 exports the hardware capability bits.
5758 (make_resolver_func): Make resolver function private and not a
5759 COMDAT function. Create the name with clone_function_name instead
5760 of make_unique_name.
5763 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
5764 correct operand in doing the split.
5766 2017-07-07 Carl Love <cel@us.ibm.com>
5768 * config/rs6000/rs6000-c: Add support for built-in function
5769 vector unsigned short vec_pack_to_short_fp32 (vector float,
5771 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
5772 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
5773 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
5774 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
5775 (convert_4f32_8i16): Add define_expand.
5776 * doc/extend.texi: Update the built-in documentation file for the
5777 new built-in function.
5779 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5781 * config/sparc/m8.md: New file.
5782 * config/sparc/sparc.md: Include m8.md.
5784 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5786 * config/sparc/sparc.opt: New option -mvis4b.
5787 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
5788 (sparc_option_override): Handle VIS4B.
5789 (enum sparc_builtins): Define
5790 SPARC_BUILTIN_DICTUNPACK{8,16,32},
5791 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
5792 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
5793 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
5794 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
5795 (check_constant_argument): New function.
5796 (sparc_vis_init_builtins): Define builtins
5797 __builtin_vis_dictunpack{8,16,32},
5798 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
5799 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
5800 __builtin_vis_fpcmpde{8,16,32}shl and
5801 __builtin_vis_fpcmpur{8,16,32}shl.
5802 (sparc_expand_builtin): Check that the constant operands to
5803 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
5804 constant and in range.
5805 * config/sparc/sparc-c.c (sparc_target_macros): Handle
5807 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
5808 (SPARC_IMM5_P): Likewise.
5809 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
5810 (enabled): Handle vis4b.
5811 (UNSPEC_DICTUNPACK): New unspec.
5812 (UNSPEC_FPCMPSHL): Likewise.
5813 (UNSPEC_FPUCMPSHL): Likewise.
5814 (UNSPEC_FPCMPDESHL): Likewise.
5815 (UNSPEC_FPCMPURSHL): Likewise.
5816 (cpu_feature): New CPU feature `vis4b'.
5817 (dictunpack{8,16,32}): New insns.
5818 (FPCSMODE): New mode iterator.
5819 (fpcscond): New code iterator.
5820 (fpcsucond): Likewise.
5821 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
5822 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
5823 (fpcmpde{8,16,32}{si,di}shl): Likewise.
5824 (fpcmpur{8,16,32}{si,di}shl): Likewise.
5825 * config/sparc/constraints.md: Define constraints `q' for unsigned
5826 2-bit integer constants and `t' for unsigned 5-bit integer
5828 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
5830 (imm5_operand_dictunpack16): Likewise.
5831 (imm5_operand_dictunpack32): Likewise.
5832 (imm2_operand): Likewise.
5833 * doc/invoke.texi (SPARC Options): Document -mvis4b.
5834 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
5835 ditunpack* and fpcmp*shl builtins.
5837 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5839 * config.gcc: Handle m8 in --with-{cpu,tune} options.
5840 * config.in: Add HAVE_AS_SPARC6 define.
5841 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
5843 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
5845 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
5846 (CPP_CPU_SPEC): Handle m8.
5847 (ASM_CPU_SPEC): Likewise.
5848 * config/sparc/sparc-opts.h (enum processor_type): Add
5850 * config/sparc/sparc.c (m8_costs): New struct.
5851 (sparc_option_override): Handle TARGET_CPU_m8.
5852 (sparc32_initialize_trampoline): Likewise.
5853 (sparc64_initialize_trampoline): Likewise.
5854 (sparc_issue_rate): Likewise.
5855 (sparc_register_move_cost): Likewise.
5856 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
5857 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
5858 (ASM_CPU64_DEFAULT_SPEC): Likewise.
5859 (CPP_CPU_SPEC): Handle M8.
5860 (ASM_CPU_SPEC): Likewise.
5861 (AS_M8_FLAG): Define.
5862 * config/sparc/sparc.md: Add m8 to the cpu attribute.
5863 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
5864 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
5866 * configure: Regenerate.
5867 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
5870 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5872 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
5874 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
5875 ("*movdi_insn_sp32"): Do not set v3pipe.
5876 ("*movsi_insn"): Likewise.
5877 ("*movdi_insn_sp64"): Likewise.
5878 ("*movsf_insn"): Likewise.
5879 ("*movdf_insn_sp32"): Likewise.
5880 ("*movdf_insn_sp64"): Likewise.
5881 ("*zero_extendsidi2_insn_sp64"): Likewise.
5882 ("*sign_extendsidi2_insn"): Likewise.
5883 ("*mov<VM32:mode>_insn"): Likewise.
5884 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5885 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5886 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5887 ("<vlop:code><VL:mode>3"): Likewise.
5888 ("*not_<vlop:code><VL:mode>3"): Likewise.
5889 ("*nand<VL:mode>_vis"): Likewise.
5890 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
5891 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
5892 ("one_cmpl<VL:mode>2"): Likewise.
5893 ("faligndata<VM64:mode>_vis"): Likewise.
5894 ("alignaddrsi_vis"): Likewise.
5895 ("alignaddrdi_vis"): Likweise.
5896 ("alignaddrlsi_vis"): Likewise.
5897 ("alignaddrldi_vis"): Likewise.
5898 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5899 ("bmaskdi_vis"): Likewise.
5900 ("bmasksi_vis"): Likewise.
5901 ("bshuffle<VM64:mode>_vis"): Likewise.
5902 ("cmask8<P:mode>_vis"): Likewise.
5903 ("cmask16<P:mode>_vis"): Likewise.
5904 ("cmask32<P:mode>_vis"): Likewise.
5905 ("pdistn<P:mode>_vis"): Likewise.
5906 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5908 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5910 * config/sparc/sparc.md ("subtype"): New insn attribute.
5911 ("*wrgsr_sp64"): Set insn subtype.
5912 ("*rdgsr_sp64"): Likewise.
5913 ("alignaddrsi_vis"): Likewise.
5914 ("alignaddrdi_vis"): Likewise.
5915 ("alignaddrlsi_vis"): Likewise.
5916 ("alignaddrldi_vis"): Likewise.
5917 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5918 ("fexpand_vis"): Likewise.
5919 ("fpmerge_vis"): Likewise.
5920 ("faligndata<VM64:mode>_vis"): Likewise.
5921 ("bshuffle<VM64:mode>_vis"): Likewise.
5922 ("cmask8<P:mode>_vis"): Likewise.
5923 ("cmask16<P:mode>_vis"): Likewise.
5924 ("cmask32<P:mode>_vis"): Likewise.
5925 ("fchksm16_vis"): Likewise.
5926 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
5927 ("fmean16_vis"): Likewise.
5928 ("fp<plusminus_insn>64_vis"): Likewise.
5929 ("<plusminus_insn>v8qi3"): Likewise.
5930 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5931 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
5932 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
5933 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
5934 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
5935 ("*movqi_insn"): Likewise.
5936 ("*movhi_insn"): Likewise.
5937 ("*movsi_insn"): Likewise.
5938 ("movsi_pic_gotdata_op"): Likewise.
5939 ("*movdi_insn_sp32"): Likewise.
5940 ("*movdi_insn_sp64"): Likewise.
5941 ("movdi_pic_gotdata_op"): Likewise.
5942 ("*movsf_insn"): Likewise.
5943 ("*movdf_insn_sp32"): Likewise.
5944 ("*movdf_insn_sp64"): Likewise.
5945 ("*zero_extendhisi2_insn"): Likewise.
5946 ("*zero_extendqihi2_insn"): Likewise.
5947 ("*zero_extendqisi2_insn"): Likewise.
5948 ("*zero_extendqidi2_insn"): Likewise.
5949 ("*zero_extendhidi2_insn"): Likewise.
5950 ("*zero_extendsidi2_insn_sp64"): Likewise.
5951 ("ldfsr"): Likewise.
5952 ("prefetch_64"): Likewise.
5953 ("prefetch_32"): Likewise.
5954 ("tie_ld32"): Likewise.
5955 ("tie_ld64"): Likewise.
5956 ("*tldo_ldub_sp32"): Likewise.
5957 ("*tldo_ldub1_sp32"): Likewise.
5958 ("*tldo_ldub2_sp32"): Likewise.
5959 ("*tldo_ldub_sp64"): Likewise.
5960 ("*tldo_ldub1_sp64"): Likewise.
5961 ("*tldo_ldub2_sp64"): Likewise.
5962 ("*tldo_ldub3_sp64"): Likewise.
5963 ("*tldo_lduh_sp32"): Likewise.
5964 ("*tldo_lduh1_sp32"): Likewise.
5965 ("*tldo_lduh_sp64"): Likewise.
5966 ("*tldo_lduh1_sp64"): Likewise.
5967 ("*tldo_lduh2_sp64"): Likewise.
5968 ("*tldo_lduw_sp32"): Likewise.
5969 ("*tldo_lduw_sp64"): Likewise.
5970 ("*tldo_lduw1_sp64"): Likewise.
5971 ("*tldo_ldx_sp64"): Likewise.
5972 ("*mov<VM32:mode>_insn"): Likewise.
5973 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5974 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5976 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5978 * config/sparc/sparc.md ("type"): New insn type viscmp.
5979 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
5981 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
5982 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
5983 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5984 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
5986 ("n7_vis_logical_11cycle"): Likewise.
5987 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
5988 * config/sparc/niagara2.md ("niag3_vis": Likewise.
5989 * config/sparc/niagara.md ("niag_vis"): Likewise.
5990 * config/sparc/ultra3.md ("us3_fga"): Likewise.
5991 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
5993 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5995 * config/sparc/sparc.md: New instruction type `bmask'.
5996 (bmaskdi_vis): Use the `bmask' type.
5997 (bmasksi_vis): Likewise.
5998 * config/sparc/ultra3.md (us3_array): Likewise.
5999 * config/sparc/niagara7.md (n7_array): Likewise.
6000 * config/sparc/niagara4.md (n4_array): Likewise.
6001 * config/sparc/niagara2.md (niag2_vis): Likewise.
6002 (niag3_vis): Likewise.
6003 * config/sparc/niagara.md (niag_vis): Likewise.
6005 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6007 * ipa-comdats.c: Remove optimize check from gate.
6008 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6009 for functions not optimized.
6010 (ipa_fn_summary_read): Skip optimize check.
6011 (ipa_fn_summary_write): Likewise.
6012 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6014 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6016 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6017 (check_callers): Give up on caller which is not optimized.
6018 (inline_small_functions): Likewise.
6019 (ipa_inline): Do not give up when not optimizing.
6020 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6021 away unoptimizes cdtors.
6022 (whole_program_function_and_variable_visibility): Do
6023 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6024 * ipa.c (process_references): Do not check optimize.
6025 (symbol_table::remove_unreachable_nodes): Update optimize check.
6026 (set_writeonly_bit): Update optimize check.
6027 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6028 (pass_ipa_single_use::gate): Remove.
6030 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6032 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6033 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6034 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6035 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6036 permute_load, permute_store, adjust_extract, adjust_splat,
6037 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6038 replace_swap_with_copy, dump_swap_insn_table,
6039 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6040 recombine_lvx_pattern, recombine_stvx_pattern,
6041 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6042 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6043 to file rs6000-p8swap.c.
6044 * config/rs6000/rs6000-p8swap.c: New file.
6045 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6046 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6047 and rs6000*-*-* targets.
6049 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6051 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6053 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6055 * lto-wrapper.c (merge_and_complain): Do not merge
6056 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6057 fsigned_zeros, ftrapping_math, fwrapv.
6058 (append_compiler_options): Do not track these options.
6059 (append_linker_options): Likewie
6061 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6063 * cgraphunit.c (cgraph_node::finalize_function): When
6064 !flag_toplevel_reorde set no_reorder flag.
6065 (varpool_node::finalize_decl): Likewise.
6066 (symbol_table::compile): Drop no toplevel reorder path.
6068 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6070 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6071 edges; zero probability is not better than uninitialized.
6073 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6075 * asan.h (asan_sanitize_allocas_p): Declare.
6076 * asan.c (asan_sanitize_allocas_p): New function.
6077 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6078 (handle_builtin_alloca): Likewise.
6079 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6080 if !asan_sanitize_allocas_p.
6081 * params.def (asan-instrument-allocas): Add new option.
6082 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6083 * opts.c (common_handle_option): Disable allocas sanitization for
6086 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6088 * asan.c: Include gimple-fold.h.
6089 (get_last_alloca_addr): New function.
6090 (handle_builtin_stackrestore): Likewise.
6091 (handle_builtin_alloca): Likewise.
6092 (asan_emit_allocas_unpoison): Likewise.
6093 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6094 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6095 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6096 (instrument_builtin_call): Pass gimple iterator to
6097 get_mem_refs_of_builtin_call.
6098 (last_alloca_addr): New global.
6099 * asan.h (asan_emit_allocas_unpoison): Declare.
6100 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6101 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6102 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6103 if function calls alloca.
6104 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6105 * gimple-fold.h (replace_call_with_value): Declare.
6106 * internal-fn.c: Include asan.h.
6107 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6108 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6110 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6112 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6113 (C_SELFTEST_FLAGS): New.
6114 (CPP_SELFTEST_FLAGS): New.
6115 (SELFTEST_DEPS): New, from deps of s-selftest.
6116 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6117 (CPP_SELFTEST_DEPS): New.
6118 (selftest): Add dependency on s-selftest-c++.
6119 (s-selftest): Rename to...
6120 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6121 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6122 than SELFTEST_FLAGS.
6123 (selftest-gdb): Rename to...
6124 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6126 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6127 (selftest-valgrind): Rename to...
6128 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6130 (selftest-valgrind): Reintroduce as an alias for
6131 selftest-c-valgrind.
6132 (s-selftest-c++): New.
6133 (selftest-c++-gdb): New.
6134 (selftest-c++-valgrind): New.
6136 2017-07-06 Olivier Hainque <hainque@adacore.com>
6138 * gcc.c (process_command): When deciding if undefined variables
6139 should be ignored when processing specs, accept "gcc -v" as well.
6141 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6143 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6144 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6146 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6148 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6149 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6150 * config/arm/arm-tables.opt: Regenerate.
6151 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6153 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6155 2017-07-06 Carl Love <cel@us.ibm.com>
6157 * ChangeLog: Clean up from mid air collision
6159 2017-07-06 Carl Love <cel@us.ibm.com>
6161 * config/rs6000/rs6000-c.c: Add support for built-in functions
6162 vector signed int vec_subc (vector signed int, vector signed int);
6163 vector signed __int128 vec_subc (vector signed __int128,
6164 vector signed __int128);
6165 vector unsigned __int128 vec_subc (vector unsigned __int128,
6166 vector unsigned __int128);
6167 vector signed int vec_sube (vector signed int, vector signed int,
6169 vector unsigned int vec_sube (vector unsigned int,
6170 vector unsigned int,
6171 vector unsigned int);
6172 vector signed __int128 vec_sube (vector signed __int128,
6173 vector signed __int128,
6174 vector signed__int128);
6175 vector unsigned __int128 vec_sube (vector unsigned __int128,
6176 vector unsigned __int128,
6177 vector unsigned __int128);
6178 vector signed int vec_subec (vector signed int, vector signed int,
6180 vector unsigned int vec_subec (vector unsigned int,
6181 vector unsigned int,
6182 vector unsigned int);
6183 vector signed __int128 vec_subec (vector signed __int128,
6184 vector signed __int128,
6185 vector signed__int128);
6186 vector unsigned __int128 vec_subec (vector unsigned __int128,
6187 vector unsigned __int128,
6188 vector unsigned __int128);
6189 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6190 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6191 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6192 BU_ALTIVEC_OVERLOAD_X definitions.
6193 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6194 * doc/extend.texi: Update the built-in documentation file for the new
6197 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6200 * diagnostic-show-locus.c (layout::layout): Use start and finish
6201 spelling location for the start and finish of each range.
6202 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6203 Add unused aspect param.
6204 * input.c (expand_location_1): Add "aspect" param, and use it
6205 to access the correct part of the location.
6206 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6208 (expand_location_to_spelling_point): Likewise.
6209 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6210 param, and pass it to expand_location_1.
6212 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6214 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6215 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6216 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6217 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6218 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6219 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6220 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6221 _mm_maskz_getmant_ss): New intrinsics.
6222 (__builtin_ia32_getexpss128_mask): Changed to ...
6223 __builtin_ia32_getexpss128_round ... this.
6224 (__builtin_ia32_getexpsd128_mask): Changed to ...
6225 __builtin_ia32_getexpsd128_round ... this.
6226 * config/i386/i386-builtin-types.def
6227 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6228 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6229 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6230 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6231 __builtin_ia32_getmantss_mask_round): New builtins.
6232 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6233 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6234 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6235 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6236 * config/i386/sse.md
6237 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6238 avx512f_sgetexp<mode><mask_scalar_name>
6239 <round_saeonly_scalar_name> ... this.
6240 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6241 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6242 vgetexp<ssescalarmodesuffix>
6243 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6244 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6245 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6246 avx512f_vgetmant<mode><mask_scalar_name>
6247 <round_saeonly_scalar_name> ... this.
6248 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6249 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6250 vgetmant<ssescalarmodesuffix>
6251 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6252 %0<mask_scalar_operand4>, %1, %2
6253 <round_saeonly_scalar_mask_op4>, %3} ... this.
6254 * config/i386/subst.md (mask_scalar_operand4,
6255 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6256 round_saeonly_scalar_nimm_predicate): New subst attributes.
6258 2017-07-06 Julia Koval <julia.koval@intel.com>
6260 * config/i386/i386.c (ix86_erase_embedded_rounding):
6261 Remove code for old rounding pattern.
6263 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6265 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6267 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6269 * doc/sourcebuild.texi (Test Directives, Variants of
6270 dg-require-support): Add documentation for dg-require-stack-check.
6272 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6274 * config/i386/subst.md (mask_scalar, round_scalar,
6275 round_saeonly_scalar): New meta-templates.
6276 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6277 round_scalar_mask_operand3, round_scalar_mask_op3,
6278 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6279 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6280 round_saeonly_scalar_constraint,
6281 round_saeonly_scalar_prefix): New subst attribute.
6282 * config/i386/sse.md
6283 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6284 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6285 <round_scalar_name> ... this.
6286 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6287 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6288 <round_scalar_name> ... this.
6289 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6290 <sse>_vm<code><mode>3<mask_scalar_name>
6291 <round_saeonly_scalar_name> ... this.
6292 (v<plusminus_mnemonic><ssescalarmodesuffix>
6293 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6294 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6295 v<plusminus_mnemonic><ssescalarmodesuffix>
6296 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6297 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6298 (v<multdiv_mnemonic><ssescalarmodesuffix>
6299 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6300 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6301 v<multdiv_mnemonic><ssescalarmodesuffix>
6302 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6303 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6304 (v<maxmin_float><ssescalarmodesuffix>
6305 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6306 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6307 v<maxmin_float><ssescalarmodesuffix>
6308 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6309 %0<mask_scalar_operand3>, %1, %<iptr>2
6310 <round_saeonly_scalar_mask_op3>} ... this.
6312 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6314 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6315 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6317 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6318 Alan Hayward <alan.hayward@arm.com>
6319 David Sherwood <david.sherwood@arm.com>
6321 * combine.c (simplify_if_then_else): Remove "enum" before
6323 * compare-elim.c (can_eliminate_compare): Likewise.
6324 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6326 (aarch64_lookup_simd_builtin_type): Likewise.
6327 (aarch64_simd_builtin_type): Likewise.
6328 (aarch64_init_simd_builtin_types): Likewise.
6329 (aarch64_simd_expand_args): Likewise.
6330 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6332 (aarch64_reverse_mask): Likewise.
6333 (aarch64_simd_emit_reg_reg_move): Likewise.
6334 (aarch64_gen_adjusted_ldpstp): Likewise.
6335 (aarch64_ccmp_mode_to_code): Likewise.
6336 (aarch64_operands_ok_for_ldpstp): Likewise.
6337 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6338 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6340 (aarch64_min_divisions_for_recip_mul): Likewise.
6341 (aarch64_reassociation_width): Likewise.
6342 (aarch64_get_condition_code_1): Likewise.
6343 (aarch64_simd_emit_reg_reg_move): Likewise.
6344 (aarch64_simd_attr_length_rglist): Likewise.
6345 (aarch64_reverse_mask): Likewise.
6346 (aarch64_operands_ok_for_ldpstp): Likewise.
6347 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6348 (aarch64_gen_adjusted_ldpstp): Likewise.
6349 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6351 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6352 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6353 (arm_lookup_simd_builtin_type): Likewise.
6354 (arm_simd_builtin_type): Likewise.
6355 (arm_init_simd_builtin_types): Likewise.
6356 (arm_expand_builtin_args): Likewise.
6357 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6358 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6359 (ft32_setup_incoming_varargs): Likewise.
6360 (ft32_function_arg): Likewise.
6361 (ft32_function_arg_advance): Likewise.
6362 (ft32_pass_by_reference): Likewise.
6363 (ft32_arg_partial_bytes): Likewise.
6364 (ft32_valid_pointer_mode): Likewise.
6365 (ft32_addr_space_pointer_mode): Likewise.
6366 (ft32_addr_space_legitimate_address_p): Likewise.
6367 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6369 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6370 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6371 (iamcu_alignment): Likewise.
6372 (canonicalize_vector_int_perm): Likewise.
6373 (ix86_noce_conversion_profitable_p): Likewise.
6374 (ix86_mpx_bound_mode): Likewise.
6375 (ix86_operands_ok_for_move_multiple): Likewise.
6376 * config/microblaze/microblaze-protos.h
6377 (microblaze_expand_conditional_branch_reg): Likewise.
6378 * config/microblaze/microblaze.c
6379 (microblaze_expand_conditional_branch_reg): Likewise.
6380 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6382 (rs6000_reassociation_width): Likewise.
6383 (rs6000_invalid_binary_op): Likewise.
6384 (fusion_p9_p): Likewise.
6385 (emit_fusion_p9_load): Likewise.
6386 (emit_fusion_p9_store): Likewise.
6387 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6389 (riscv_hard_regno_mode_ok_p): Likewise.
6390 (riscv_address_insns): Likewise.
6391 (riscv_split_symbol): Likewise.
6392 (riscv_legitimize_move): Likewise.
6393 (riscv_function_value): Likewise.
6394 (riscv_hard_regno_nregs): Likewise.
6395 (riscv_expand_builtin): Likewise.
6396 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6397 (riscv_build_integer): Likewise.
6398 (riscv_split_integer): Likewise.
6399 (riscv_legitimate_constant_p): Likewise.
6400 (riscv_cannot_force_const_mem): Likewise.
6401 (riscv_regno_mode_ok_for_base_p): Likewise.
6402 (riscv_valid_base_register_p): Likewise.
6403 (riscv_valid_offset_p): Likewise.
6404 (riscv_valid_lo_sum_p): Likewise.
6405 (riscv_classify_address): Likewise.
6406 (riscv_legitimate_address_p): Likewise.
6407 (riscv_address_insns): Likewise.
6408 (riscv_load_store_insns): Likewise.
6409 (riscv_force_binary): Likewise.
6410 (riscv_split_symbol): Likewise.
6411 (riscv_force_address): Likewise.
6412 (riscv_legitimize_address): Likewise.
6413 (riscv_move_integer): Likewise.
6414 (riscv_legitimize_const_move): Likewise.
6415 (riscv_legitimize_move): Likewise.
6416 (riscv_address_cost): Likewise.
6417 (riscv_subword): Likewise.
6418 (riscv_output_move): Likewise.
6419 (riscv_canonicalize_int_order_test): Likewise.
6420 (riscv_emit_int_order_test): Likewise.
6421 (riscv_function_arg_boundary): Likewise.
6422 (riscv_pass_mode_in_fpr_p): Likewise.
6423 (riscv_pass_fpr_single): Likewise.
6424 (riscv_pass_fpr_pair): Likewise.
6425 (riscv_get_arg_info): Likewise.
6426 (riscv_function_arg): Likewise.
6427 (riscv_function_arg_advance): Likewise.
6428 (riscv_arg_partial_bytes): Likewise.
6429 (riscv_function_value): Likewise.
6430 (riscv_pass_by_reference): Likewise.
6431 (riscv_setup_incoming_varargs): Likewise.
6432 (riscv_print_operand): Likewise.
6433 (riscv_elf_select_rtx_section): Likewise.
6434 (riscv_save_restore_reg): Likewise.
6435 (riscv_for_each_saved_reg): Likewise.
6436 (riscv_register_move_cost): Likewise.
6437 (riscv_hard_regno_mode_ok_p): Likewise.
6438 (riscv_hard_regno_nregs): Likewise.
6439 (riscv_class_max_nregs): Likewise.
6440 (riscv_memory_move_cost): Likewise.
6441 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6442 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6443 (rl78_addr_space_address_mode): Likewise.
6444 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6446 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6447 (rs6000_reassociation_width): Likewise.
6448 (rs6000_invalid_binary_op): Likewise.
6449 (fusion_p9_p): Likewise.
6450 (emit_fusion_p9_load): Likewise.
6451 (emit_fusion_p9_store): Likewise.
6452 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6453 (ok_for_simple_move_operands): Likewise.
6454 (ok_for_simple_move_strict_operands): Likewise.
6455 (ok_for_simple_arith_logic_operands): Likewise.
6456 (visium_legitimize_reload_address): Likewise.
6457 (visium_select_cc_mode): Likewise.
6458 (output_cbranch): Likewise.
6459 (visium_split_double_move): Likewise.
6460 (visium_expand_copysign): Likewise.
6461 (visium_expand_int_cstore): Likewise.
6462 (visium_expand_fp_cstore): Likewise.
6463 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6464 (visium_function_arg): Likewise.
6465 (visium_function_arg_advance): Likewise.
6466 (visium_libcall_value): Likewise.
6467 (visium_setup_incoming_varargs): Likewise.
6468 (visium_legitimate_constant_p): Likewise.
6469 (visium_legitimate_address_p): Likewise.
6470 (visium_legitimize_address): Likewise.
6471 (visium_secondary_reload): Likewise.
6472 (visium_register_move_cost): Likewise.
6473 (visium_memory_move_cost): Likewise.
6474 (prepare_move_operands): Likewise.
6475 (ok_for_simple_move_operands): Likewise.
6476 (ok_for_simple_move_strict_operands): Likewise.
6477 (ok_for_simple_arith_logic_operands): Likewise.
6478 (visium_function_value_1): Likewise.
6479 (rtx_ok_for_offset_p): Likewise.
6480 (visium_legitimize_reload_address): Likewise.
6481 (visium_split_double_move): Likewise.
6482 (visium_expand_copysign): Likewise.
6483 (visium_expand_int_cstore): Likewise.
6484 (visium_expand_fp_cstore): Likewise.
6485 (visium_split_cstore): Likewise.
6486 (visium_select_cc_mode): Likewise.
6487 (visium_split_cbranch): Likewise.
6488 (output_cbranch): Likewise.
6489 (visium_print_operand_address): Likewise.
6490 * expmed.c (flip_storage_order): Likewise.
6491 * expmed.h (emit_cstore): Likewise.
6492 (flip_storage_order): Likewise.
6493 * genrecog.c (validate_pattern): Likewise.
6494 * hsa-gen.c (gen_hsa_addr): Likewise.
6495 * internal-fn.c (expand_arith_overflow): Likewise.
6496 * ira-color.c (allocno_copy_cost_saving): Likewise.
6497 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6498 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6499 (process_invariant_for_inheritance): Likewise.
6500 * lra-eliminations.c (move_plus_up): Likewise.
6501 * omp-low.c (lower_oacc_reductions): Likewise.
6502 * simplify-rtx.c (simplify_subreg): Likewise.
6503 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6504 (TARGET_CHKP_BOUND_MODE): Likewise..
6505 * targhooks.c (default_chkp_bound_mode): Likewise.
6506 (default_setup_incoming_vararg_bounds): Likewise.
6507 * targhooks.h (default_chkp_bound_mode): Likewise.
6508 (default_setup_incoming_vararg_bounds): Likewise.
6509 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6510 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6511 (have_whole_vector_shift): Likewise.
6512 * tree-vect-stmts.c (vectorizable_load): Likewise.
6513 * doc/tm.texi: Regenerate.
6515 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6517 Graceful degrade if Binutils PR21472 is not available.
6520 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6521 .rodata in flash test fails.
6522 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6523 * confgure: Regenerate.
6524 * config.in: Regenerate.
6525 * config/avr/avr.c (avr_asm_named_section)
6526 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6527 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6528 (avr_asm_init_sections): Same.
6530 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6532 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6533 (fma<VH:mode>4_intrinsic): Likewise.
6534 (*fmsub<VCVTF:mode>4): Likewise.
6535 (*fmsub<VH:mode>4_intrinsic): Likewise.
6537 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6540 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6541 Don't depend on "optimize > 0".
6542 (out_movhi_r_mr, out_movqi_mr_r): Same.
6543 (out_movhi_mr_r, out_movqi_r_mr): Same.
6544 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6545 io_address_operand on "optimize > 0".
6547 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6549 * tree-loop-distribution.c: Add general explanantion on the pass.
6550 (generate_loops_for_partition): Mark distributed loop.
6551 (pg_add_dependence_edges): New parameter. Handle alias data
6552 dependence specially and record it in the parameter if asked.
6553 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6554 (init_partition_graph_vertices, add_partition_graph_edge): New.
6555 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6556 (free_partition_graph_vdata, build_partition_graph): New.
6557 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6558 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6559 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6560 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6561 (version_for_distribution_p, finalize_partitions): New.
6562 (distribute_loop): Handle alias data dependence specially. Factor
6563 out loop fusion code as functions and call these functions.
6565 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6567 * tree-loop-distribution.c (classify_partition): New parameter and
6568 better handle reduction statement.
6569 (rdg_build_partitions): Revise comment.
6570 (distribute_loop): Compute statements in all partitions and pass it
6571 to classify_partition.
6573 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6575 * tree-loop-distribution.c (enum partition_type): New.
6576 (struct partition): New field type.
6577 (partition_merge_into): Add parameter. Update partition type.
6578 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6579 (build_rdg_partition_for_vertex): Compute partition type.
6580 (rdg_build_partitions): Dump partition type.
6581 (distribute_loop): Update calls to partition_merge_into.
6583 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6585 * tree-loop-distribution.c (struct ddr_hasher): New.
6586 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6588 (classify_partition): Call get_data_dependence.
6589 (pg_add_dependence_edges): Ditto.
6590 (distribute_loop): Release data dependence hash table.
6592 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6594 * tree-loop-distribution.c (ref_base_address): Delete.
6595 (similar_memory_accesses): Rename ...
6596 (share_memory_accesses): ... to this. Check if partitions access
6597 the same memory reference.
6598 (distribute_loop): Call share_memory_accesses.
6600 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6602 * tree-loop-distribution.c (struct partition): New field recording
6604 (partition_alloc, partition_free): Init and release data refs.
6605 (partition_merge_into): Merge data refs.
6606 (build_rdg_partition_for_vertex): Collect data refs for partition.
6607 (pg_add_dependence_edges): Change parameters from vector to bitmap.
6609 (distribute_loop): Remve data refs from vertice data of partition
6612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6614 * tree-loop-distribution.c (params.h): Include header file.
6615 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
6616 (datarefs_vec): New global var.
6617 (create_rdg_vertices): Use datarefs_vec directly.
6618 (free_rdg): Don't free data references.
6619 (build_rdg): Update use. Don't free data references.
6620 (distribute_loop): Compute global variable for data references.
6621 Bail out if there are too many data references.
6623 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6625 * tree-loop-distribution.c (loop_nest): New global var.
6626 (build_rdg): Use loop directly, rather than loop nest.
6627 (pg_add_dependence_edges): Remove loop nest parameter. Use global
6629 (distribute_loop): Compute global variable loop nest. Update use.
6631 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6633 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
6634 (partition_merge_into): New parameter. Dump reason for fusion.
6635 (distribute_loop): Update use of partition_merge_into.
6637 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6639 * tree-loop-distribution.c (bb_top_order_index): New.
6640 (bb_top_order_index_size, bb_top_order_cmp): New.
6641 (stmts_from_loop): Use topological order.
6642 (pass_loop_distribution::execute): Compute and release topological
6643 order for basic blocks.
6645 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6647 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
6650 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6652 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
6653 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
6654 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
6655 * internal-fn.def (LOOP_DIST_ALIAS): New.
6656 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
6657 (fold_loop_internal_call): ... this.
6658 (vect_loop_dist_alias_call): New function.
6659 (set_uid_loop_bbs): Call fold_loop_internal_call.
6660 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
6663 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6666 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
6667 Require dead FLAGS_REG at the beginning of a peephole.
6669 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6672 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
6673 arguments in the call to __builtin_ia32_sbb_u32.
6674 (_subborrow_u64): Swap _X and _Y arguments in the call to
6675 __builtin_ia32_sbb_u64.
6677 2017-07-04 Jakub Jelinek <jakub@redhat.com>
6680 * tree-vrp.c (compare_assert_loc): Turn into a function template
6681 with stable template parameter. Only test if a->e is NULL,
6682 !a->e == !b->e has been verified already. Use e == NULL or
6683 e != NULL instead of e or ! e tests. If stable is true, don't use
6684 iterative_hash_expr, on the other side allow a or b or both NULL
6685 and sort the NULLs last.
6686 (process_assert_insertions): Sort using compare_assert_loc<false>
6687 instead of compare_assert_loc, later sort using
6688 compare_assert_loc<true> before calling process_assert_insertions_for
6689 in a loop. Use break instead of continue once seen NULL pointer.
6691 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6693 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6694 Cortex-R7 and Cortex-R8 processors.
6696 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6698 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
6699 uninitialized while src is not.
6701 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
6703 * common/config/arm/arm-common.c: Adjust include path for
6705 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
6706 (arm-cpu.h): Create in build directory. Adjust dependency rules.
6707 (arm-cpu-data.h): Likewise.
6708 (arm-cpu-cdata.h): Likewise.
6709 * config/arm/arm-cpu.h: Delete.
6710 * config/arm/arm-cpu-cdata.h: Delete.
6711 * config/arm/arm-cpu-data.h: Delete.
6713 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
6715 * config/arm/arm-cpus.in (cortex-a55): New.
6716 (cortex-a75): Likewise.
6717 (cortex-a75.cortex-a55): Likewise.
6718 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
6720 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
6721 * config/arm/arm-cpu-cdata.h: Regenerate.
6722 * config/arm/arm-cpu-data.h: Regenerate.
6723 * config/arm/arm-cpu.h: Regenerate.
6724 * config/arm/arm-tables.opt: Regenerate.
6725 * config/arm/arm-tune.md: Regenerate.
6727 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6729 * haifa-sched.c (sched_create_recovery_edges): Update profile.
6731 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6733 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
6736 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
6738 PR tree-optimization/81292
6739 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
6740 full_string_p, also call adjust_related_strinfos if the adjustment
6741 is simple, otherwise invalidate related strinfos.
6743 2017-07-04 Martin Liska <mliska@suse.cz>
6746 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
6747 newly created variable as DECL_IGNORED_P.
6749 2017-07-04 Martin Liska <mliska@suse.cz>
6752 * ipa-inline.c (inline_small_functions):
6753 Use xstrdup_for_dump.
6755 2017-07-04 Tom de Vries <tom@codesourcery.com>
6757 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
6759 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6762 * config/darwin.c (darwin_function_switched_text_sections):
6765 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
6767 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
6769 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6771 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
6773 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6775 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
6776 min_profitable_iters, and th as inclusive lower bounds.
6777 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
6778 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
6779 for min_profitable_iters and min_profitable_estimate.
6780 (vect_transform_loop): Treat th as an inclusive lower bound.
6781 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6783 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6786 * config/darwin.c (darwin_function_switched_text_sections):
6787 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
6788 in two pieces, and suppress the use of buf.
6790 2017-07-03 Nathan Sidwell <nathan@acm.org>
6792 * hash-table.h (hash_table_mod1): Fix indentation.
6794 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6797 * predict.c (force_edge_cold): Be more careful about propagation
6799 * profile-count.h (profile_probability::guessed,
6800 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
6802 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
6804 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
6806 * doc/invoke.texi (rcpc architecture extension): Document it.
6808 2017-07-03 Richard Biener <rguenther@suse.de>
6810 PR tree-optimization/60510
6811 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
6812 the scalar reduction PHI and use it.
6813 (vectorizable_reduction): Properly guard the single_defuse_cycle
6814 path for non-SLP reduction chains where we cannot use it.
6815 Rework reduc_def/index and vector type deduction. Rework
6816 vector operand gathering during reduction op code-gen.
6817 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
6818 chains dissolve the chain and leave it to non-SLP reduction
6821 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6823 * tree-data-ref.h (dr_alignment): Declare.
6824 * tree-data-ref.c (dr_alignment): New function.
6825 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
6826 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
6828 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
6830 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6832 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
6833 and base_misalignment fields.
6834 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
6835 * tree-data-ref.c: Include builtins.h.
6836 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
6837 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
6838 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
6839 * tree-vect-data-refs.c: Include tree-cfg.h.
6840 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
6841 fields instead of calculating an alignment here.
6842 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
6843 innermost_loop_behavior fields.
6845 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6847 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
6849 (DR_STEP_ALIGNMENT): New macro.
6850 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
6851 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
6852 (create_data_ref): Print it.
6853 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
6854 to tell whether the step preserves vector (mis)alignment.
6855 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6856 Move the check for an integer step and generalise to all INTEGER_CST.
6857 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
6858 Print the outer step alignment.
6860 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6862 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
6863 with offset_alignment.
6864 (DR_ALIGNED_TO): Delete.
6865 (DR_OFFSET_ALIGNMENT): New macro.
6866 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
6867 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
6868 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
6869 (create_data_ref): Likewise.
6870 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6871 (vect_analyze_data_refs): Likewise.
6872 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
6873 creating dummy innermost behavior.
6875 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6877 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
6878 with a "innermost_loop_behavior *" and refeence tree.
6879 * tree-data-ref.c (dr_analyze_innermost): Likewise.
6880 (create_data_ref): Update call accordingly.
6881 * tree-predcom.c (find_looparound_phi): Likewise.
6883 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6885 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
6886 fields with dr_wrt_vec_loop.
6887 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
6888 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
6889 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
6890 (vect_dr_behavior): New function.
6891 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6892 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
6893 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
6894 track whether the step preserves the misalignment.
6895 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6896 Use vect_dr_behavior.
6897 (vect_setup_realignment): Update call accordingly.
6898 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
6899 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
6900 call to vect_create_addr_base_for_vector_ref.
6901 (vect_create_cond_for_align_checks): Likewise.
6902 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
6903 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
6904 (vect_recog_mask_conversion_pattern): Likewise.
6905 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
6906 (new_stmt_vec_info): Remove redundant zeroing.
6908 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6910 * common/config/arm/arm-common.c (arm_be8_option): New function.
6911 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
6912 (ISA_ARMv6): Add isa_bit_be8.
6913 * config/arm/arm.h (arm_be8_option): Add prototype.
6914 (BE8_SPEC_FUNCTION): New define.
6915 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
6916 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
6917 (mlittle-endian): Similarly.
6918 (mbe8, mbe32): New options.
6919 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
6920 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
6922 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6924 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
6926 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6928 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
6929 (cleanup_tree_cfg_bb): Use it.
6930 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
6932 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
6934 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6937 * loop-doloop.c (add_test): Update profile.
6939 2017-07-03 Martin Liska <mliska@suse.cz>
6942 * sanopt.c (rewrite_usage_of_param): New function.
6943 (sanitize_rewrite_addressable_params): Likewise.
6944 (pass_sanopt::execute): Call rewrite_usage_of_param.
6946 2017-07-03 Richard Biener <rguenther@suse.de>
6948 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
6949 back to using VIEW_CONVERT_EXPR.
6951 2017-07-03 Martin Liska <mliska@suse.cz>
6954 * doc/extend.texi: Document when a resolver function is
6955 generated for target_clones.
6957 2017-07-03 Martin Liska <mliska@suse.cz>
6959 * asan.c (asan_emit_stack_protection): Unpoison just red zones
6960 and shadow memory of auto variables which are subject of
6961 use-after-scope sanitization.
6962 (asan_expand_mark_ifn): Add do set only when is_poison.
6964 2016-07-03 Richard Biener <rguenther@suse.de>
6966 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
6968 (vect_force_simple_reduction): Record reduction def -> phi mapping.
6969 (vectorizable_reduction): Perform reduction PHI creation when
6970 visiting a reduction PHI and adjust and simplify code generation
6971 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
6972 (vect_transform_loop): Visit reduction PHIs.
6973 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
6974 defs into the SLP tree.
6975 (vect_build_slp_tree): Reduction defs terminate the recursion.
6976 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
6978 (vect_get_vec_defs_for_stmt_copy): Export.
6979 (vect_get_vec_defs): Likewise.
6980 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
6982 (vect_get_vec_defs_for_stmt_copy): Declare.
6983 (vect_get_vec_defs): Likewise.
6985 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6987 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
6988 parameter with a "loop" parameter and use it instead of the
6989 loop containing DR_STMT. Don't check simple_iv when doing
6990 BB analysis. Describe the two analysis modes in the comment.
6992 2017-07-03 Tom de Vries <tom@codesourcery.com>
6994 PR tree-optimization/69468
6995 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
6996 (find_same_succ_bb): Handle ignore_edge_flags.
6998 2017-07-03 Tom de Vries <tom@codesourcery.com>
7000 PR tree-optimization/81192
7001 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7003 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7005 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7007 2017-07-03 Tom de Vries <tom@codesourcery.com>
7009 PR tree-optimization/81192
7010 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7011 BB_SAME_SUCC (bb) == NULL.
7013 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7015 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7018 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7020 * dumpfile.c: Include profile-count.h
7021 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7023 (insert_cond_bb): Update profile.
7024 * tree-cfg.h (insert_cond_bb): Update prototype.
7025 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7026 * tree-dump.c: Do not include tree-cfg.
7028 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7030 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7032 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7034 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7037 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7039 * tree-complex.c (expand_complex_div_wide): update profile.
7041 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7042 Alan Hayward <alan.hayward@arm.com>
7043 David Sherwood <david.sherwood@arm.com>
7045 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7046 (CORETYPES_H): New define.
7047 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7048 (insn-modes-inline.h, s-modes-inline-h): New rules.
7049 (generated_files): Add insn-modes-inline.h.
7050 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7051 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7052 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7053 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7054 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7055 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7056 (build/gencodes.o, build/genconditions.o): Likewise.
7057 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7058 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7059 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7060 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7061 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7062 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7063 * coretypes.h: Include everything up to real.h for generators.
7064 Include insn-modes.h first. Include wide-int-print.h after
7065 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7066 * machmode.h: Don't include insn-modes.h here.
7067 * function-tests.c: Remove includes of signop.h, machmode.h,
7068 double-int.h and wide-int.h.
7070 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7072 * optc-save-gen.awk: Likewise.
7073 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7074 * godump.c: Remove include of wide-int-print.h.
7075 * pretty-print.h: Likewise.
7076 * wide-int-print.cc: Likewise.
7077 * wide-int.cc: Likewise.
7078 * hash-map-tests.c: Remove include of signop.h.
7079 * hash-set-tests.c: Likewise.
7080 * rtl-tests.c: Likewise.
7081 * mkconfig.sh: Remove include of machmode.h.
7082 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7084 (emit_insn_modes_inline_h): ...this new function. Emit the code
7085 into an insn-modes-inline.h header file, adding appropriate
7086 include guards and end comments.
7087 (emit_insn_modes_c_header): Remove include of machmode.h.
7088 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7090 (main): Handle -i flag and call emit_insn_modes_inline_h when
7093 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7095 * tree-ssa-strlen.c (strinfo): Rename the length field to
7096 nonzero_chars. Add a full_string_p field.
7097 (compare_nonzero_chars, zero_length_string_p): New functions.
7098 (get_addr_stridx): Add an offset_out parameter.
7099 Use compare_nonzero_chars.
7100 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7101 (new_strinfo): Update after above changes to strinfo.
7102 (set_endptr_and_length): Set full_string_p.
7103 (get_string_length): Update after above changes to strinfo.
7104 (unshare_strinfo): Update call to new_strinfo.
7105 (maybe_invalidate): Likewise.
7106 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7107 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7108 as a uhwi instead of an shwi. Update after above changes to
7109 strinfo and new_strinfo.
7110 (zero_length_string): Assert that chainsi contains full strings.
7111 Use zero_length_string_p. Update call to new_strinfo.
7112 (adjust_related_strinfos): Update after above changes to strinfo.
7113 Copy full_string_p from origsi.
7114 (adjust_last_stmt): Use zero_length_string_p.
7115 (handle_builtin_strlen): Update after above changes to strinfo and
7116 new_strinfo. Install the lhs as the string length if the previous
7117 entry didn't describe a full string.
7118 (handle_builtin_strchr): Update after above changes to strinfo
7120 (handle_builtin_strcpy): Likewise.
7121 (handle_builtin_strcat): Likewise.
7122 (handle_builtin_malloc): Likewise.
7123 (handle_pointer_plus): Likewise.
7124 (handle_builtin_memcpy): Likewise. Track nonzero characters
7125 that aren't necessarily followed by a nul terminator.
7126 (handle_char_store): Likewise.
7128 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7130 PR tree-optimization/80769
7131 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7132 for malloc and calloc. Document the new invariant that all related
7133 strinfos have delayed lengths or none do.
7134 (verify_related_strinfos): Move earlier in file.
7135 (set_endptr_and_length): New function, split out from...
7136 (get_string_length): ...here. Also set the lengths of related
7138 (zero_length_string): Assert that chainsi has known (rather than
7140 (adjust_related_strinfos): Likewise.
7142 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7144 PR tree-optimization/81136
7145 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7146 assert that two references with the same misalignment have the same
7147 compile-time misalignment if those compile-time misalignments
7150 2017-07-01 Andi Kleen <ak@linux.intel.com>
7152 * print-tree.c (print_node): Print all attributes.
7154 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7156 * cfg.c (scale_bbs_frequencies): New function.
7157 * cfg.h (scale_bbs_frequencies): Declare it.
7158 * cfgloopanal.c (single_likely_exit): Cleanup.
7159 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7161 (scale_loop_profile): Likewise.
7162 (loop_version): Likewise.
7163 (create_empty_loop_on_edge): Update.
7164 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7165 scale_loop_frequencies, scale_loop_profile, loopify,
7166 loop_version): Update prototypes.
7167 * modulo-sched.c (sms_schedule): Update.
7168 * predict.c (unlikely_executed_edge_p): Also check probability.
7169 (probably_never_executed_edge_p): Fix typo.
7170 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7171 * tree-parloops.c (gen_parallel_loop): Update.
7172 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7173 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7174 * tree-ssa-loop-split.c (split_loop): Update.
7175 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7176 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7177 (vect_loop_versioning): Update.
7178 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7180 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7182 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7184 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7186 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7187 to keep profile consistent.
7189 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7191 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7192 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7193 * profile-count.h (max_safe_multiplier): Make unsigned.
7194 (profile_count::guessed_zero): New.
7196 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7198 * bb-reorder.c (fix_up_crossing_landing_pad,
7199 fix_crossing_conditional_branches): Use make_single_succ_edge
7200 to keep profile consistent.
7202 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7204 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7207 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7210 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7211 the right scopes, make sure cond_jump isn't preserved between multiple
7212 iterations. Search for fallthru edge whenever there are 3+ edges and
7213 use find_fallthru_edge for it.
7215 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7217 Patch by Alexander Monakov <amonakov@ispras.ru>
7218 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7219 probabilities consistently.
7221 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7223 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7224 emit_cmp_and_jump_insns.
7226 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7229 * tree-inline.c (expand_call_inline): Combine profile statuses.
7231 2017-06-30 Andrew Pinski <apinski@cavium.com>
7233 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7234 fold_stmt returned true.
7236 2017-06-30 Nathan Sidwell <nathan@acm.org>
7238 * ggc.h (empty_string): Delete.
7239 * cfgexpand.c (expand_asm_stmt): Use plain "".
7240 * optabs.c (expand_asm_memory_barrier): Likewise.
7241 * stringpool.c (empty_string): Delete.
7242 (digit_vector, digit_string): Delete.
7243 (ggc_alloc_string): Use plain "", don't optimize single digit
7244 strings. Use ggc_alloc_atomic.
7246 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7248 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7249 comparison set and one other set, use the cost of the non-comparison
7252 2017-06-30 Nathan Sidwell <nathan@acm.org>
7254 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7257 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7259 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7260 loops. Remove now unneeded calls to gimple_switch_set_label() that
7261 just set removed labels to NULL_TREE.
7263 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7265 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7266 (set_range_info): ...here. Only call set_range_info_raw if domain
7268 (set_nonzero_bits): Call set_range_info_raw.
7269 * tree-ssanames.h (set_range_info_raw): New.
7271 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7274 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7275 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7276 of nonimmediate_operand and <store_mask_constraint> instead of m
7277 for the input operand. For V8FI iterator, always split if input
7278 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7279 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7280 <store_mask_predicate> instead of register_operand and
7281 <store_mask_constraint> instead of v for the input operand. Make
7282 sure both operands aren't MEMs for if not <mask_applied>.
7284 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7286 * lto-wrapper.c (copy_file) Close both file descriptors before
7289 2017-06-30 Martin Liska <mliska@suse.cz>
7292 * multiple_target.c (create_dispatcher_calls): Make ifunc
7293 also for function that don't have calls or are not referenced.
7295 2017-06-30 Richard Biener <rguenther@suse.de>
7297 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7298 analyze the first scalar stmt. Move vector type computation
7299 for the BB case here from ...
7300 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7301 live operation processing in the SLP case properly.
7303 2017-06-30 Richard Biener <rguenther@suse.de>
7305 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7307 2017-06-30 Martin Liska <mliska@suse.cz>
7310 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7311 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7313 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7315 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7318 2017-06-30 Richard Biener <rguenther@suse.de>
7320 PR tree-optimization/81249
7321 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7322 condition reduction result to original scalar type.
7324 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7326 * profile-count.h (enum profile_quality): Fix typos and whitespace
7329 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7331 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7332 type for branch probabilities.
7334 2017-06-29 Julian Brown <julian@codesourcery.com>
7335 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7337 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7338 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7339 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7340 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7342 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7344 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7345 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7346 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7347 CC usage from generic code to here.
7348 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7349 CC usage into the target macros.
7351 2017-06-29 Maya Rashish <coypu@sdf.org>
7353 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7356 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7358 * arm/arm-builtins.c: Include profile-count.h
7359 * except.c (sjlj_emit_function_enter): Use
7360 profile_probability::unlikely.
7362 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7364 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7365 and tocrel_offset be pointer args rather than implicitly using
7367 (legitimate_constant_pool_address_p, rs6000_emit_move,
7368 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7369 tocrel_offset and use in toc_relative_expr_p call.
7370 (print_operand, print_operand_address): Use static tocrel_base_oac
7371 and tocrel_offset_oac.
7372 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7375 2017-06-29 Maya Rashish <coypu@sdf.org>
7377 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7379 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7381 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7382 objects, take into account only the alignment of 'op0' and 'mode1' if
7385 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7387 * ccmp.c (ccmp_tree_comparison_p): New function.
7388 (ccmp_candidate_p): Update to use above function.
7389 (get_compare_parts): New function.
7390 (expand_ccmp_next): Update to use new functions.
7391 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7393 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7394 take mode as argument.
7395 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7396 * expr.c (expand_expr_real_1): Pass mode as argument.
7398 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7400 * combine.c (combine_instructions): Print insns to dump_file, together
7403 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7405 * asan.c (asan_emit_stack_protection): Update.
7406 (create_cond_insert_point): Update.
7407 * auto-profile.c (afdo_propagate_circuit): Update.
7408 * basic-block.h (struct edge_def): Turn probability to
7409 profile_probability.
7410 (EDGE_FREQUENCY): Update.
7411 * bb-reorder.c (find_traces_1_round): Update.
7412 (better_edge_p): Update.
7413 (sanitize_hot_paths): Update.
7414 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7415 (make_single_succ_edge): Update.
7416 (check_bb_profile): Update.
7417 (dump_edge_info): Update.
7418 (update_bb_profile_for_threading): Update.
7419 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7420 probabilitycount to 0.
7421 * cfgbuild.c (compute_outgoing_frequencies): Update.
7422 * cfgcleanup.c (try_forward_edges): Update.
7423 (outgoing_edges_match): Update.
7424 (try_crossjump_to_edge): Update.
7425 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7426 (expand_gimple_tailcall): Update.
7427 (construct_init_block): Use make_single_succ_edge.
7428 (construct_exit_block): Use make_single_succ_edge.
7429 * cfghooks.c (verify_flow_info): Update.
7430 (redirect_edge_succ_nodup): Update.
7431 (split_edge): Update.
7432 (account_profile_record): Update.
7433 * cfgloopanal.c (single_likely_exit): Update.
7434 * cfgloopmanip.c (scale_loop_profile): Update.
7435 (set_zero_probability): Remove.
7436 (duplicate_loop_to_header_edge): Update.
7437 * cfgloopmanip.h (loop_version): Update prototype.
7438 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7439 (force_nonfallthru_and_redirect): Update.
7440 (update_br_prob_note): Update.
7441 (rtl_verify_edges): Update.
7442 (purge_dead_edges): Update.
7443 (rtl_lv_add_condition_to_bb): Update.
7444 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7445 * cgraphunit.c (init_lowered_empty_function): Update.
7446 (cgraph_node::expand_thunk): Update.
7447 * cilk-common.c: Include profile-count.h
7448 * dojump.c (inv): Remove.
7449 (jumpifnot): Update.
7450 (jumpifnot_1): Update.
7451 (do_jump_1): Update.
7453 (do_jump_by_parts_greater_rtx): Update.
7454 (do_compare_rtx_and_jump): Update.
7455 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7456 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7457 * dwarf2cfi.c: Include profile-count.h
7458 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7459 (sjlj_emit_dispatch_table): Likewise.
7460 * explow.c: Include profile-count.h
7461 * expmed.c (emit_store_flag_force): Update.
7462 (do_cmp_and_jump): Update.
7463 * expr.c (compare_by_pieces_d::generate): Update.
7464 (compare_by_pieces_d::finish_mode): Update.
7465 (emit_block_move_via_loop): Update.
7466 (store_expr_with_bounds): Update.
7467 (store_constructor): Update.
7468 (expand_expr_real_2): Update.
7469 (expand_expr_real_1): Update.
7470 * expr.h (try_casesi, try_tablejump): Update prototypes.
7471 * gimple-pretty-print.c (dump_probability): Update.
7472 (dump_profile): New.
7473 (dump_gimple_label): Update.
7474 (dump_gimple_bb_header): Update.
7475 * graph.c (draw_cfg_node_succ_edges): Update.
7476 * hsa-gen.c (convert_switch_statements): Update.
7477 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7478 (find_if_case_1): Update.
7479 (find_if_case_2): Update.
7480 * internal-fn.c (expand_arith_overflow_result_store): Update.
7481 (expand_addsub_overflow): Update.
7482 (expand_neg_overflow): Update.
7483 (expand_mul_overflow): Update.
7484 (expand_vector_ubsan_overflow): Update.
7485 * ipa-cp.c (good_cloning_opportunity_p): Update.
7486 * ipa-split.c (split_function): Use make_single_succ_edge.
7487 * ipa-utils.c (ipa_merge_profiles): Update.
7488 * loop-doloop.c (add_test): Update.
7489 (doloop_modify): Update.
7490 * loop-unroll.c (compare_and_jump_seq): Update.
7491 (unroll_loop_runtime_iterations): Update.
7492 * lra-constraints.c (lra_inheritance): Update.
7493 * lto-streamer-in.c (input_cfg): Update.
7494 * lto-streamer-out.c (output_cfg): Update.
7495 * mcf.c (adjust_cfg_counts): Update.
7496 * modulo-sched.c (sms_schedule): Update.
7497 * omp-expand.c (expand_omp_for_init_counts): Update.
7498 (extract_omp_for_update_vars): Update.
7499 (expand_omp_ordered_sink): Update.
7500 (expand_omp_for_ordered_loops): Update.
7501 (expand_omp_for_generic): Update.
7502 (expand_omp_for_static_nochunk): Update.
7503 (expand_omp_for_static_chunk): Update.
7504 (expand_cilk_for): Update.
7505 (expand_omp_simd): Update.
7506 (expand_omp_taskloop_for_outer): Update.
7507 (expand_omp_taskloop_for_inner): Update.
7508 * omp-simd-clone.c (simd_clone_adjust): Update.
7509 * optabs.c (expand_doubleword_shift): Update.
7510 (expand_abs): Update.
7511 (emit_cmp_and_jump_insn_1): Update.
7512 (expand_compare_and_swap_loop): Update.
7513 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7514 * predict.c (predictable_edge_p): Update.
7515 (edge_probability_reliable_p): Update.
7516 (set_even_probabilities): Update.
7517 (combine_predictions_for_insn): Update.
7518 (combine_predictions_for_bb): Update.
7519 (propagate_freq): Update.
7520 (estimate_bb_frequencies): Update.
7521 (force_edge_cold): Update.
7522 * profile-count.c (profile_count::dump): Add missing space into dump.
7523 (profile_count::debug): Add newline.
7524 (profile_count::differs_from_p): Explicitly convert to unsigned.
7525 (profile_count::stream_in): Update.
7526 (profile_probability::dump): New member function.
7527 (profile_probability::debug): New member function.
7528 (profile_probability::differs_from_p): New member function.
7529 (profile_probability::differs_lot_from_p): New member function.
7530 (profile_probability::stream_in): New member function.
7531 (profile_probability::stream_out): New member function.
7532 * profile-count.h (profile_count_quality): Rename to ...
7533 (profile_quality): ... this one.
7534 (profile_probability): New.
7535 (profile_count): Update.
7536 * profile.c (compute_branch_probabilities): Update.
7537 * recog.c (peep2_attempt): Update.
7538 * sched-ebb.c (schedule_ebbs): Update.
7539 * sched-rgn.c (find_single_block_region): Update.
7540 (compute_dom_prob_ps): Update.
7541 (schedule_region): Update.
7542 * sel-sched-ir.c (compute_succs_info): Update.
7543 * stmt.c (struct case_node): Update.
7544 (do_jump_if_equal): Update.
7545 (get_outgoing_edge_probs): Update.
7546 (conditional_probability): Update.
7547 (emit_case_dispatch_table): Update.
7548 (expand_case): Update.
7549 (expand_sjlj_dispatch_table): Update.
7550 (emit_case_nodes): Update.
7551 * targhooks.c: Update.
7552 * tracer.c (better_p): Update.
7553 (find_best_successor): Update.
7554 * trans-mem.c (expand_transaction): Update.
7555 * tree-call-cdce.c: Update.
7556 * tree-cfg.c (gimple_split_edge): Upate.
7557 (move_sese_region_to_fn): Upate.
7558 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7559 * tree-eh.c (lower_resx): Upate.
7560 (cleanup_empty_eh_move_lp): Upate.
7561 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7562 * tree-inline.c (copy_edges_for_bb): Update.
7563 (copy_cfg_body): Update.
7564 * tree-parloops.c (gen_parallel_loop): Update.
7565 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7566 (gimple_gen_time_profiler): Update.
7567 * tree-ssa-dce.c (remove_dead_stmt): Update.
7568 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7569 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7570 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7571 (unloop_loops): Update.
7572 (try_peel_loop): Update.
7573 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7574 * tree-ssa-loop-split.c (connect_loops): Update.
7575 (split_loop): Update.
7576 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7577 (hoist_guard): Update.
7578 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7579 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7580 (value_replacement): Update.
7581 * tree-ssa-reassoc.c (branch_fixup): Update.
7582 * tree-ssa-tail-merge.c (replace_block_by): Update.
7583 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7584 (create_edge_and_update_destination_phis): Update.
7585 (compute_path_counts): Update.
7586 (recompute_probabilities): Update.
7587 (update_joiner_offpath_counts): Update.
7588 (freqs_to_counts_path): Update.
7589 (duplicate_thread_path): Update.
7590 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7591 (struct switch_conv_info): Update.
7592 (gen_inbound_check): Update.
7593 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7594 (vect_do_peeling): Update.
7595 (vect_loop_versioning): Update.
7596 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7597 (optimize_mask_stores): Update.
7598 * ubsan.c (ubsan_expand_null_ifn): Update.
7599 * value-prof.c (gimple_divmod_fixed_value): Update.
7600 (gimple_divmod_fixed_value_transform): Update.
7601 (gimple_mod_pow2): Update.
7602 (gimple_mod_pow2_value_transform): Update.
7603 (gimple_mod_subtract): Update.
7604 (gimple_mod_subtract_transform): Update.
7605 (gimple_ic): Update.
7606 (gimple_stringop_fixed_value): Update.
7607 (gimple_stringops_transform): Update.
7608 * value-prof.h: Update.
7610 2017-06-29 Carl Love <cel@us.ibm.com>
7612 * config/rs6000/rs6000-c.c: Add support for built-in functions
7613 vector signed int vec_signed (vector float);
7614 vector signed long long vec_signed (vector double);
7615 vector signed int vec_signed2 (vector double, vector double);
7616 vector signed int vec_signede (vector double);
7617 vector signed int vec_signedo (vector double);
7618 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
7619 instruction generator.
7620 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7621 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
7622 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
7624 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
7625 vunsignede_v2df): Add define_expands.
7626 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
7627 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
7628 VEC_UNSIGNEDO): Add definitions.
7629 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7630 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
7631 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
7632 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
7633 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
7634 * config/rs6000/altivec.h (vec_signed, vec_signed2,
7635 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
7636 vec_unsignede, vec_unsignedo): Add builtin defines.
7637 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
7639 * doc/extend.texi: Update the built-in documentation file for the
7640 new built-in functions.
7642 2017-06-29 Richard Biener <rguenther@suse.de>
7644 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
7645 reduction chains to LOOP_VINFO_REDUCTIONS.
7646 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
7647 SLP reductions after processing reduction chains.
7649 2017-06-29 Nathan Sidwell <nathan@acm.org>
7651 * builtins.c (fold_builtin_FUNCTION): Use
7652 lang_hooks.decl_printable_name.
7654 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
7657 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
7658 with only one label.
7659 * stmt.c (expand_case): Assert NCASES is greater than one.
7661 2017-06-29 Richard Biener <rguenther@suse.de>
7663 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
7665 (group_case_labels): Likewise.
7666 (find_taken_edge): Push sanity checking on val to workers...
7667 (find_taken_edge_cond_expr): ... here
7668 (find_taken_edge_switch_expr): ... and here, handle cases
7669 with just a default label.
7670 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
7671 (group_case_labels): Likewise.
7672 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
7673 group_case_labels does anything cleanup the CFG again.
7675 2017-06-29 Bin Cheng <bin.cheng@arm.com>
7677 PR tree-optimization/81196
7678 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
7679 exit condition comparing two IVs.
7681 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
7683 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
7684 profile to the dummy entry at the end of the list of architectures.
7685 * config/arm/arm-cpu-cdata.h: Regenerated.
7687 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7688 Michael Collison <michael.collison@arm.com>
7691 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
7693 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
7694 (*aarch64_reg_<mode>3_minus_mask): New pattern.
7695 (*aarch64_<optab>_reg_di3_mask2): New pattern.
7696 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
7697 of shift when the shift amount is masked with constant equal to
7698 the size of the mode.
7699 * config/aarch64/predicates.md (subreg_lowpart_operator): New
7702 2017-06-29 Martin Liska <mliska@suse.cz>
7704 * config/i386/i386.opt: Change range from [1,5] to [0,5].
7706 2017-06-29 Yury Gribov <tetra2005@gmail.com>
7709 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
7712 (edge_growth_cache_entry::edge_growth_cache_entry): New
7714 (reset_edge_growth_cache): Update to use constructor.
7716 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7718 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
7719 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7720 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7722 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
7724 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
7725 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
7727 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
7729 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
7730 (*-linux-uclibc*): Add t-uclibc tmake_file.
7731 * config/t-musl: New.
7732 * config/t-uclibc: New.
7734 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
7736 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
7738 (gen_comm_data): Emit architectural setting of arch_prof.
7739 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
7741 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
7742 (armv8-m.base, armv8-m.main): Likewise.
7743 * arm-protos.h (arm_build_target): Add profile field.
7744 (arch_option): Likewise.
7745 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
7747 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
7748 arm_active_target.profile.
7750 2017-06-28 Richard Biener <rguenther@suse.de>
7753 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
7754 TYPE_OVERFLOW_WRAPS.
7755 * match.pd (negate_expr_p): Likewise.
7756 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
7757 fold_build2, not fold_binary.
7759 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7761 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
7762 Convert memory address to Pmode.
7763 (aarch64_print_operand): Assert MEM operands are always Pmode.
7765 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7768 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
7769 Remove redundant if.
7770 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
7771 * config/arm/aarch-common-protos.h
7772 (aarch_forward_to_shift_is_not_shifted_re): Remove.
7773 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
7775 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
7778 * multiple_target.c (create_dispatcher_calls): Set the default
7779 clone to be static, not public.
7781 2017-06-28 Richard Biener <rguenther@suse.de>
7783 * tree-vect-loop.c (vectorizable_reduction): Move special
7784 cond reduction IV var creation ...
7785 (vect_create_epilog_for_reduction): ... here. Remove induction_index
7786 parameter. Use STMT_VINFO_VECTYPE.
7787 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
7790 2017-06-28 Martin Liska <mliska@suse.cz>
7793 * ipa-visibility.c (non_local_p): Handle visibility.
7795 2017-06-28 Martin Liska <mliska@suse.cz>
7798 * common.opt: Add IntegerRange to various options.
7799 * opt-functions.awk (integer_range_info): New function.
7800 * optc-gen.awk: Add integer_range_info to cl_options struct.
7801 * opts-common.c (decode_cmdline_option): Handle
7802 CL_ERR_INT_RANGE_ARG.
7803 (cmdline_handle_error): Likewise.
7804 * opts.c (print_filtered_help): Show valid interval in
7805 when --help is provided.
7806 * opts.h (struct cl_option): Add range_min and range_max fields.
7807 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
7809 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
7811 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
7812 (x * C EQ/NE y * C): New transformation.
7814 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
7816 * genmultilib (combination_space): Accept '+' in option names.
7818 2017-06-28 Martin Liska <mliska@suse.cz>
7821 * asan.c (instrument_derefs): Bail out inner references
7822 that are hard register variables.
7824 2017-06-28 Jakub Jelinek <jakub@redhat.com>
7827 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
7828 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
7830 2017-06-28 Richard Biener <rguenther@suse.de>
7832 * tree-vectorizer.h (vect_get_vec_defs): Remove.
7833 (vect_get_slp_defs): Adjust.
7834 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
7836 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
7838 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7839 get_initial_defs_for_reduction instead of vect_get_vec_defs.
7840 (vectorizable_reduction): Adjust.
7841 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
7843 (vect_get_slp_defs): Likewise.
7844 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
7845 (vectorizable_bswap): Adjust.
7846 (vectorizable_call): Likewise.
7847 (vectorizable_conversion): Likewise.
7848 (vectorizable_assignment): Likewise.
7849 (vectorizable_shift): Likewise.
7850 (vectorizable_operation): Likewise.
7851 (vectorizable_store): Likewise.
7852 (vectorizable_condition): Likewise.
7853 (vectorizable_comparison): Likewise.
7855 2017-06-28 Michael Collison <michael.collison@arm.com>
7858 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
7860 (arm_gen_movmemqi): Removed unused variable 'i'.
7861 Convert 'for' loop into 'while' loop.
7862 (arm_expand_prologue): Remove last unnecessary set of insn.
7863 (thumb_pop): Remove unused variable 'pushed_words'.
7864 (thumb_exit): Remove last unnecessary set of regs_to_pop.
7866 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7868 * config/s390/predicates.md: Use s390_rel_address_ok_p.
7869 * config/s390/s390-protos.h: Add prototype of
7870 s390_rel_address_ok_p.
7871 * config/s390/s390.c (s390_got_symbol): New function.
7872 (s390_rel_address_ok_p): New function.
7873 (legitimize_pic_address): Use s390_rel_address_ok_p.
7874 (s390_load_got): Use s390_got_symbol.
7875 (s390_option_override): Issue error if
7876 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
7877 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
7879 * config/s390/s390.opt: New option mpic-data-is-text-relative.
7881 2017-06-27 Andrew Pinski <apinski@cavium.com>
7883 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
7884 (X * copysign (1.0, X)): New pattern.
7885 (X * copysign (1.0, -X)): New pattern.
7886 (copysign (-1.0, CST)): New pattern.
7888 2017-06-27 Joseph Myers <joseph@codesourcery.com>
7890 * genmultilib (combination_space): Remove variable.
7891 Validate reuse rules against regular expression for any sequence
7892 of multilib options in any order.
7894 2017-06-27 Michael Collison <michael.collison@arm.com>
7896 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
7897 call aarch64_split_simd_combine.
7898 * (aarch64_combine_internal<mode>): Delete pattern.
7899 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
7900 Allow register and subreg operands.
7902 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7904 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
7905 specific need, just fallback on defaults.
7906 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
7908 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7909 Olivier Hainque <hainque@adacore.com>
7911 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
7913 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
7914 targets. Pick a default if no particular attempt applied.
7915 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
7918 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7920 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
7921 (x86_64-wrs-vxworks7): Likewise.
7923 2017-06-27 Marek Polacek <polacek@redhat.com>
7926 * ubsan.c (instrument_null): Check get_base_address's result for null.
7928 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
7930 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
7932 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
7934 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
7935 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
7936 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
7937 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
7939 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
7940 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
7941 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
7942 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
7943 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
7944 BUILT_IN_FEUPDATEENV): New builtins.
7945 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
7946 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
7947 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
7948 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
7950 (builtin_structptr_types): Adjust size.
7951 * tree.c (builtin_structptr_types): Add four entries.
7953 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7954 Olivier Hainque <hainque@adacore.com>
7956 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
7957 (TLS_SYM): New local macro, forcing reference to __tls__ on
7958 link command lines for VxWorks 7 RTPs, triggering initialization
7960 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
7961 OS features TLS support, true for RTPs on VxWorks 7.
7962 * config/vxworks.c (vxworks_override_options): Setup emutls
7965 2017-06-27 Jakub Jelinek <jakub@redhat.com>
7967 * predict.c (test_prediction_value_range): Use -1U instead of -1
7968 to avoid narrowing conversion warning.
7969 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
7970 to avoid narrowing conversion warning.
7971 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
7973 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
7975 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7977 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
7978 64bit configurations.
7979 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
7980 (SIZE_TYPE): Likewise.
7981 * config/vxworks.c (vxworks_emutls_var_fields): Use
7982 long_unsigned_type_node instead of unsigned_type_node as the offset
7983 field type, which is "pointer" mode in emutls.c.
7985 2017-06-27 Jakub Jelinek <jakub@redhat.com>
7988 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
7991 * gimple-fold.c (replace_call_with_call_and_fold): Handle
7992 gimple_vuse copying separately from gimple_vdef copying.
7994 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7996 * value-prof.c (free_hist): Remove call to memset and the enclosing if
7999 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8000 Olivier Hainque <hainque@adacore.com>
8002 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8003 for all vxworks7 targets.
8004 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8005 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8006 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8007 variations for VX6/VX7 and 32/64bits later on in ...
8008 (VXWORKS_LIB_SPEC): Leverage new macros.
8009 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8010 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8012 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8014 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8015 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8017 2017-06-26 Carl Love <cel@us.ibm.com>
8019 * config/rs6000/rs6000-c.c: Add support for built-in functions
8020 vector bool char vec_reve (vector bool char);
8021 vector signed char vec_reve (vector signed char);
8022 vector unsigned char vec_reve (vector unsigned char);
8023 vector bool int vec_reve (vector bool int);
8024 vector signed int vec_reve (vector signed int);
8025 vector unsigned int vec_reve (vector unsigned int);
8026 vector bool long long vec_reve (vector bool long long);
8027 vector signed long long vec_reve (vector signed long long);
8028 vector unsigned long long vec_reve (vector unsigned long long);
8029 vector bool short vec_reve (vector bool short);
8030 vector signed short vec_reve (vector signed short);
8031 vector double vec_reve (vector double);
8032 vector float vec_reve (vector float);
8033 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8034 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8035 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8036 (altivec_vreve): New pattern.
8037 * config/rs6000/altivec.h (vec_reve): New define.
8038 * doc/extend.texi (vec_rev): Update the built-in documentation file
8039 for the new built-in functions.
8041 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8043 PR tree-optimization/71815
8044 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8046 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8048 (slsr_process_phi): Likewise.
8049 (replace_uncond_cands_and_profitable_phis): Don't replace a
8050 multiply candidate with a stride of 1 (copy or cast).
8051 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8053 (lowest_cost_path): Likewise.
8054 (total_savings): Likewise.
8056 2017-06-26 Richard Biener <rguenther@suse.de>
8059 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8060 Use def_builtin_pure for all gather builtins.
8062 2017-06-26 Richard Biener <rguenther@suse.de>
8064 PR tree-optimization/81203
8065 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8068 2017-06-26 Marek Polacek <polacek@redhat.com>
8071 * doc/invoke.texi: Document -Wmultistatement-macros.
8073 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8075 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8076 arm_neon_ok_no_float_abi effective target.
8078 2017-06-26 Richard Biener <rguenther@suse.de>
8080 PR tree-optimization/80928
8081 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8082 (copy_bbs): Set BB_DUPLICATED flag early.
8083 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8085 (execute_on_shrinking_pred): Likewise.
8086 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8087 BB_DUPLICATED blocks.
8088 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8089 iterate over all PHIs considering removal of *gsi.
8091 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8093 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8096 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8098 * config/rs6000/rs6000-string.c: (expand_block_clear,
8099 do_load_for_compare, select_block_compare_mode,
8100 compute_current_alignment, expand_block_compare,
8101 expand_strncmp_align_check, expand_strn_compare,
8102 expand_block_move, rs6000_output_load_multiple)
8103 Move functions related to string/block move/compare
8105 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8106 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8107 for this function which is now used in two files.
8108 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8109 * config.gcc: Add rs6000-string.o to extra_objs for
8110 targets powerpc*-*-* and rs6000*-*-*.
8112 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8115 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8116 32-bit, since indexed is not valid for DImode.
8117 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8118 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8119 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8120 (define_peephole2 for Altivec d-form store): Likewise.
8123 * multiple_target.c (create_dispatcher_calls): Only create the
8124 dispatcher call if the function is the default clone of a
8127 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8130 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8131 a call, force the call to not be a tail call.
8133 2017-06-23 Jeff Law <law@redhat.com>
8135 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8138 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8140 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8141 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8142 handling for early expansion of vector shifts (sl,sr,sra,rl).
8143 (builtin_function_type): Add vector shift right instructions
8144 to the unsigned argument list.
8146 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8148 rtl-optimizatoin/79286
8149 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8150 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8151 trap. PIC register plus a const unspec without offset can never trap.
8153 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8155 * tree.h (builtin_structptr_type): New type.
8156 (builtin_structptr_types): Declare new array.
8157 * tree.c (builtin_structptr_types): New array.
8158 (free_lang_data, build_common_tree_nodes): Use it.
8160 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8163 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8166 2017-06-22 Matt Turner <mattst88@gmail.com>
8168 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8169 Lake models to skylake case. Assume skylake for unknown
8170 models with clflushopt.
8172 2017-06-22 Jeff Law <law@redhat.com>
8174 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8175 frame sizes that do not satisfy aarch64_uimm12_shift.
8177 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8179 * profile-count.h (apply_probability,
8180 apply_scale, probability_in): Fix checks for zero.
8182 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8184 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8185 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8187 2016-06-22 Richard Biener <rguenther@suse.de>
8189 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8190 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8191 REDUC_MAX_EXPR support.
8192 (vectorizable_reduction): Likewise.
8193 (vect_create_epilog_for_reduction): Likewise.
8195 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8197 * match.pd (A / (1 << B) -> A >> B): New.
8198 * generic-match-head.c: Include optabs-tree.h.
8199 * gimple-match-head.c: Likewise.
8200 * optabs-tree.h (target_supports_op_p): New.
8201 * optabs-tree.c (target_supports_op_p): New.
8203 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8205 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8206 $gcc_cv_ld --help output.
8207 (gcc_cv_ld_demangle): Likewise.
8208 (gcc_cv_ld_eh_frame_hdr): Likewise.
8209 (gcc_cv_ld_pie): Likewise.
8210 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8211 (gcc_cv_ld_buildid): Likewise.
8212 (gcc_cv_ld_sysroot): Likewise.
8213 (ld_bndplt_support): Likewise.
8214 (ld_pushpopstate_support): Likewise.
8215 * configure: Regenerate.
8216 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8218 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8221 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8222 operands indexes to avoid gap between operands and match_dups.
8224 2017-06-21 Andrew Pinski <apinski@cavium.com>
8226 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8227 Increment Arith_shift and Arith_shift_reg by 1.
8228 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8230 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8231 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8232 (aarch64_strip_extend): Add new argument and test for it.
8233 (aarch64_cheap_mult_shift_p): New function.
8234 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8235 add a cost if it is true.
8236 Update calls to aarch64_strip_extend.
8237 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8239 2017-06-21 Andrew Pinski <apinski@cavium.com>
8241 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8243 (thunderxt88): Likewise.
8244 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8245 (thunderx_prefetch_tune): New variable.
8246 (thunderx2t99_prefetch_tune): Update for the correct values.
8247 (thunderxt88_tunings): New variable.
8248 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8249 generic_prefetch_tune.
8250 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8252 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8254 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8255 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8256 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8257 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8258 (aarch64_atomic_cas<mode>, GPI): Likewise.
8260 2017-06-21 Martin Liska <mliska@suse.cz>
8262 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8263 statements on cold and hot labels.
8264 * predict.c (tree_estimate_probability_bb): Remove the
8265 prediction from this place.
8267 2017-06-21 Martin Liska <mliska@suse.cz>
8269 PR tree-optimization/79489
8270 * gimplify.c (maybe_add_early_return_predict_stmt): New
8272 (gimplify_return_expr): Call the function.
8273 * predict.c (tree_estimate_probability_bb): Remove handling
8275 * predict.def: Update comment about early return predictor.
8276 * gimple-predict.h (is_gimple_predict): New function.
8277 * predict.def: Change default value of early return to 66.
8278 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8280 * passes.def: Put pass_strip_predict_hints to the beginning of
8283 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8285 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8286 FUNCTION_DECL declarations.
8287 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8289 (dwaf2out_decl): Likewise.
8290 * godump.c (go_early_global_decl): Skip call to the real debug hook
8291 for FUNCTION_DECL declarations.
8292 * passes.c (rest_of_decl_compilation): Skip call to the
8293 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8294 -fdump-go-spec is passed.
8296 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8298 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8299 Reorder for compactness.
8300 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8301 (def_builtin_pure, def_builtin_pure2): New functions.
8302 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8304 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8306 * match.pd (nop_convert): New predicate.
8307 ((A +- CST1) +- CST2): Allow some NOP conversions.
8309 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8312 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8313 with ctors/dtors if GOVD_SHARED is set.
8315 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8317 * config/aarch64/aarch64.md (movti_aarch64):
8318 Emit mov rather than orr.
8319 (movtf_aarch64): Likewise.
8320 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8321 Emit mov rather than orr.
8323 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8325 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8326 Swap alternatives, make integer dup more expensive.
8328 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8330 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8331 Return true for non-tls symbols.
8333 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8335 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8336 (cortex-a75): Likewise.
8337 (cortex-a75.cortex-a55): Likewise.
8338 * config/aarch64/aarch64-tune.md: Regenerate.
8339 * doc/invoke.texi (-mtune): Document new values for -mtune.
8341 2017-06-21 Tom de Vries <tom@codesourcery.com>
8343 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8345 (Effective-Target Keywords, Other attributes): Suggest using
8346 dg-add-options stack_size feature to get stack limit in stack_size
8347 effective target documentation.
8349 2017-06-21 Julian Brown <julian@codesourcery.com>
8350 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8352 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8353 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8354 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8356 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8357 attribute type list for neon_multiply.
8358 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8359 attribute type list for neon_multiply.
8360 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8361 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8362 attribute type list for neon_multiply.
8363 * config/arm/types.md (crypto_pmull): Add.
8364 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8365 attribute type list.
8367 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8369 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8372 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8374 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8375 to make sure not to dereference a NULL cost_classes_ptr pointer.
8377 2017-06-20 Carl Love <cel@us.ibm.com>
8379 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8380 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8381 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8382 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8383 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8384 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8385 VMULOSW): New enum "unspec" values.
8386 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8387 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8388 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8389 altivec_vmulosw): New patterns.
8390 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8391 VMULOSW): Add definitions.
8393 2017-06-20 Julia Koval <julia.koval@intel.com>
8395 * config/i386/i386.c: Fix rounding expand for new pattern.
8396 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8398 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8400 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8401 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8403 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8405 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8408 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8410 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8411 architecture, then by implementer ID.
8412 * config/aarch64/aarch64-tune.md: Regenerate.
8414 2017-06-20 Richard Biener <rguenther@suse.de>
8417 * fold-const.c (split_tree): Fold to type before negating.
8419 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8421 * diagnostic-show-locus.c
8422 (selftest::test_fixit_deletion_affecting_newline): New function.
8423 (selftest::diagnostic_show_locus_c_tests): Call it.
8425 2017-06-20 Andreas Schwab <schwab@suse.de>
8428 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8431 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8433 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8434 __ARM_FEATURE_COPROC according to support.
8436 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8438 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8439 Rewritten to avoid overflow for > 32-bit pointers.
8442 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8443 by removing enum keyword.
8444 (ubsan_type_descriptor): Likewise. Formatting fix.
8447 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8448 splitter): Require TARGET_SSE2 in the condition.
8450 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8453 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8454 for doing vector set of SFmode on ISA 3.0.
8455 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8456 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8458 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8459 SFmode value into a V4SF variable that was extracted from another
8460 V4SF variable without converting the element to double precision
8461 and back to single precision vector format.
8462 (vsx_insert_extract_v4sf_p9_2): Likewise.
8464 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8466 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8467 in UWHI to avoid undefined overflow.
8470 * ubsan.h (enum ubsan_encode_value_phase): New.
8471 (ubsan_encode_value): Change second argument to
8472 enum ubsan_encode_value_phase with default value of
8473 UBSAN_ENCODE_VALUE_GENERIC.
8474 * ubsan.c (ubsan_encode_value): Change second argument to
8475 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8476 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8477 create_tmp_var_raw instead of create_tmp_var and use a
8479 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8480 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8481 ubsan_encode_value callers.
8484 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8485 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8486 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8488 2017-06-19 Richard Biener <rguenther@suse.de>
8491 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8492 estimates if we changed anything.
8494 2017-06-19 Richard Biener <rguenther@suse.de>
8496 PR tree-optimization/80887
8497 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8498 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8499 simplified lookups, then reset mprts_hook.
8500 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8502 (try_to_simplify): Likewise.
8504 2017-06-19 Martin Liska <mliska@suse.cz>
8507 * gimplify.c (gimplify_switch_expr):
8508 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8510 2017-06-19 Martin Liska <mliska@suse.cz>
8512 * doc/install.texi: Document that PGO runs in 4 stages.
8514 2017-06-19 Martin Liska <mliska@suse.cz>
8517 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8518 to dispatcher function name.
8519 * multiple_target.c (replace_function_decl): New function.
8520 (create_dispatcher_calls): Redirect both edges and references.
8522 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8524 * profile-count.c (profile_count::dump): Dump quality.
8525 (profile_count::differs_from_p): Update for unsigned val.
8526 * profile-count.h (profile_count_quality): New enum.
8527 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8529 2017-06-19 Richard Biener <rguenther@suse.de>
8531 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8532 struct function as arg.
8533 (estimate_numbers_of_iterations): Export overload with loop arg.
8534 (free_numbers_of_iterations_estimates_loop): Use an overload of
8535 free_numbers_of_iterations_estimates instead.
8536 * tree-cfg.c (remove_bb): Adjust.
8537 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8538 * tree-parloops.c (gen_parallel_loop): Likewise.
8539 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8541 (tree_unroll_loops_completely): Likewise.
8542 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8543 Use an overload instead and export.
8544 (estimated_loop_iterations): Adjust.
8545 (max_loop_iterations): Likewise.
8546 (likely_max_loop_iterations): Likewise.
8547 (estimate_numbers_of_iterations): Take struct function as arg
8549 (loop_exits_before_overflow): Adjust.
8550 (free_numbers_of_iterations_estimates_loop): Use an overload.
8551 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8552 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8554 2017-06-19 Richard Biener <rguenther@suse.de>
8557 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8558 RANGE_EXPR conservatively.
8560 2017-06-16 Carl Love <cel@us.ibm.com>
8562 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8563 definitions for vec_float, vec_float2, vec_floato,
8564 vec_floate built-ins.
8565 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8566 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8568 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8569 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8570 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8571 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8572 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8573 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8574 vec_floato): Add builtin defines.
8575 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8576 Update the built-in documentation file for the new built-in
8579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8581 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8582 (mthumb): Mark as the negative of -marm.
8584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8586 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8588 (ARM Options, -mtune): Document that this accepts the same
8589 extension options as -mcpu.
8590 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8594 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8595 permitted extensions.
8597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8599 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8600 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8601 (armv8-m.main): Add option +nodsp.
8602 * config/arm/arm-cpu-cdata.h: Regenerated.
8604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8606 * config/arm/t-fuchsia: New file.
8607 * config.gcc (arm*-*-fuchsia*): Use it.
8609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8611 * config/arm/t-symbian: Rewrite for new option infrastructure.
8613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8615 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
8616 (MULTILIB_REQUIRED): Likewise.
8618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8620 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
8621 (MULTILIB_RESUE): Likewise.
8622 (MULTILIB_MATCHES): Likewise.
8623 (MULTLIB_REQUIRED): Likewise.
8625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8627 * config/arm/t-rtems: Rewrite for new option framework.
8629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8631 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
8632 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
8633 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
8634 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
8635 * config/arm/t-multilib: ... here.
8636 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
8637 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
8638 armv7-a and armv8*-a when A-profile libraries have not been built.
8639 * config/arm/t-rmprofile: Rewrite.
8641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8643 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
8644 with a backslash. Remove the backslash after substituting unescaped
8646 * doc/fragments.texi (MULTILIB_REUSE): Document it.
8648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8650 * config.gcc: (arm*-*-*): When building a-profile libraries, force
8651 the driver to pass through the default setting of -mfloat-abi.
8652 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
8654 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
8655 (all_feat_combs): New rule.
8656 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
8658 * config/arm/t-aprofile: Rewrite.
8660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8662 * config/arm/arm.h (FPUTYPE_AUTO): Define.
8663 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
8664 fpu is not specified by the user/command-line.
8665 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
8666 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
8667 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
8668 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
8669 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
8670 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
8672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8674 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
8675 * config/arm/t-arm-elf: Rewritten.
8677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8679 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
8680 have some floating-point instructions.
8681 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
8682 (TARGET_MAYBE_HARD_FLOAT): New macro.
8683 * config/arm/arm-builtins.c (arm_init_builtins): Use
8684 TARGET_MAYBE_HARD_FLOAT.
8685 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
8687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8689 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
8690 (configargs.h): Include it.
8691 (arm_print_hint_for_fpu_option): New function.
8692 (arm_parse_fpu_option): New function.
8693 (candidate_extension): New class.
8694 (arm_canon_for_multilib): New function.
8695 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
8696 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
8697 (ARCH_CANONICAL_SPECS): New macro.
8698 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
8700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8702 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
8703 are set after handling multilib fragments. Set target_cpu_default2
8706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8708 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
8710 (arm*-*-*): Set target_cpu_default2 to a quoted string.
8711 * config/arm/parsecpu.awk (check_cpu): Validate any extension
8713 (check_arch): Likewise.
8714 * config/arm/arm.c (arm_configure_build_target): Handle
8715 TARGET_CPU_DEFAULT being a string constant. Scan any feature
8716 options in the default.
8718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8720 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
8721 when an option is an alias of another.
8722 * config/arm/parsecpu.awk (optalias): New parser token.
8723 (gen_comm_data): Mark non-alias options as such. Emit entries
8724 for extension aliases.
8725 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
8726 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
8727 (armv6kz, armv6zk, armv6t2): Likewise.
8728 (armv7): Make vfpv3-d16 an alias.
8729 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
8731 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
8732 Sort in canonical order.
8733 (armv8-a): Sort in canonical order.
8734 (armv8.1-a, armv8.2-a): Likewise.
8735 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
8737 (cortex-a9): Sort in canonical order.
8738 * config/arm/arm.c (selftests.h): Include it.
8739 (arm_test_cpu_arch_data): New function.
8740 (arm_run_self_tests): New function.
8741 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
8742 (targetm): Move declaration to the end of the file.
8743 * arm-cpu-cdata.h: Regenerated.
8745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8747 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
8748 call to target_mode_check describing the type of option passed.
8749 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
8750 (arm_target_thumb_only): Use arm_parse_arch_option_name or
8751 arm_parse_cpu_option_name to match parameters against list of
8753 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
8754 arm_arch_core_flags data structure.
8755 * config/arm/arm-cpu_cdata.h: Regenerated.
8757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8759 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
8761 (arm_print_hint_for_cpu_option): Likewise.
8762 (arm_print_hint_for_arch_option): Likewise.
8763 (arm_parse_cpu_option_name): Likewise.
8764 (arm_parse_arch_option_name): Likewise.
8765 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
8766 of entries in the all_fpus list.
8767 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
8768 (arm_parse_cpu_option_name): Declare.
8769 (arm_parse_arch_option_name): Declare.
8770 (arm_parse_option_features): Declare.
8771 (arm_intialize_isa): Declare.
8772 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
8774 (gen_comm_data): ... here. Make definitions non-static.
8775 * config/arm/arm-cpu-data.h: Regenerated.
8776 * config/arm/arm-cpu-cdata.h: Regenerated.
8778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8780 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
8781 (cpu_arch_extension): New structure.
8782 (cpu_arch_option, arch_option, cpu_option): New structures.
8783 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
8785 (gen_data): Generate new format data tables.
8786 * config/arm/arm.c (cpu_tune): New structure.
8787 (cpu_option, processors): Delete.
8788 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
8789 (arm_print_hint_for_cpu_option): ... this and ...
8790 (arm_print_hint_for_arch_option): ... this.
8791 (arm_parse_arch_cpu_name): Delete. Replace with ...
8792 (arm_parse_cpu_option_name): ... this and ...
8793 (arm_parse_arch_option_name): ... this.
8794 (arm_unrecognized_feature): Change type of target parameter to
8796 (arm_parse_arch_cpu_features): Delete. Replace with ...
8797 (arm_parse_option_features): ... this.
8798 (arm_configure_build_target): Rework to use new configuration data
8800 (arm_print_tune_info): Rework for new configuration data tables.
8801 * config/arm/arm-cpu-data.h: Regenerated.
8802 * config/arm/arm-cpu.h: Regenerated.
8804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8806 * Makefile.in (OBJS): Move sbitmap.o from here ...
8807 (OBJS-libcommon): ... to here.
8809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8811 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
8812 (ISA_ALL_CRYPTO): New macro.
8813 (ISA_ALL_SIMD): New macro
8814 (ISA_ALL_FP): New macro.
8815 * config/arm/arm.c (fpu_bitlist): Update initializer.
8816 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
8818 (arm9e): Add fpu. Add option for nofp
8819 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
8820 (arm926ej-s, arm1026ej-s): Likewise.
8821 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
8822 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
8823 neon-fp16, neon-vfpv4, nofp and nosimd.
8824 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
8825 (cortex-a8): Add fpu. Add option for nofp.
8826 (cortex-a9): Add fpu. Add options for nosimd and nofp.
8827 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
8828 (cortex-r4f): Add fpu.
8829 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
8830 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
8832 (cortex-r8): Likewise.
8833 (cortex-m4): Add fpu. Add option for nofp.
8834 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
8835 (cortex-a17.cortex-a7): Likewise.
8836 (cortex-a32): Add fpu. Add options for crypto and nofp.
8837 (cortex-a35, cortex-a53): Likewise.
8838 (cortex-a57): Add fpu. Add option for crypto.
8839 (cortex-a72, cortex-a73): Likewise.
8840 (exynos-m1): Likewise.
8841 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
8842 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
8843 (cortex-m33): Add fpu. Add option for nofp.
8844 * config/arm/arm-cpu-cdata.h: Regenerated
8845 * config/arm/arm-cpu-data.h: Regenerated.
8847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8849 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
8850 (armv5te, armv5tej): Likewise.
8851 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
8852 (armv7): Add options fp and vfpv3-d16.
8853 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
8854 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
8856 (armv7ve): Likewise.
8857 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
8858 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
8859 (armv8-a): Add nocrypto option.
8860 (armv8.1-a, armv8.2-a): Likewise.
8861 (armv8-m.main): add options fp, fp.dp and nofp.
8863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8865 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
8867 (armv8-a+crc): Delete.
8868 (armv8.1-a): Add options simd, crypto and nofp.
8869 (armv8.2-a): Add options fp16, simd, crypto and nofp.
8870 (armv8.2-a+fp16): Delete.
8871 (armv8-m.main): Add option dsp.
8872 (armv8-m.main+dsp): Delete.
8873 (cortex-a8): Add fpu. Add nofp option.
8874 (cortex-a9): Add fpu. Add nofp and nosimd options.
8875 * config/arm/parsecpu.awk (gen_data): Generate option tables and
8876 link to main cpu and architecture data structures.
8877 (gen_comm_data): Only put isa attributes from the main architecture
8879 (option): New statement for architecture and CPU entries.
8880 * arm.c (struct cpu_option): New structure.
8881 (struct processors): Add entry for options.
8882 (arm_unrecognized_feature): New function.
8883 (arm_parse_arch_cpu_name): Ignore any characters after the first
8885 (arm_parse_arch_cpu_feature): New function.
8886 (arm_configure_build_target): Separate out any CPU and architecture
8887 features and parse separately. Don't error out if -mfpu=auto is
8888 used with only an architecture string.
8889 (arm_print_asm_arch_directives): New function.
8890 (arm_file_start): Call it.
8891 * config/arm/arm-cpu-cdata.h: Regenerated.
8892 * config/arm/arm-cpu-data.h: Likewise.
8893 * config/arm/arm-tables.opt: Likewise.
8895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8897 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
8898 assembler when it is not -mfpu=auto.
8900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8902 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
8903 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
8904 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
8905 (ASM_CPU_SPEC): Rewrite.
8906 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
8907 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
8908 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
8909 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
8910 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
8911 copied string is NUL-terminated. Also strip any characters prefixed
8913 (arm_rewrite_selected_arch): New function.
8914 (arm_rewrite_march): New function.
8916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8918 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
8919 (x_arm_cpu_string, x_arm_tune_string): Likewise.
8920 (march, mcpu, mtune): Convert to string-based options.
8921 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
8922 (arm_parse_arch_cpu_name): New function.
8923 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
8924 identify selected architecture or CPU.
8925 (arm_option_save): New function.
8926 (TARGET_OPTION_SAVE): Redefine.
8927 (arm_option_restore): Restore string options.
8928 (arm_option_print): Print string options.
8930 2017-06-16 Martin Sebor <msebor@redhat.com>
8932 PR tree-optimization/80933
8933 PR tree-optimization/80934
8934 * builtins.c (fold_builtin_3): Do not handle bcmp here.
8935 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
8936 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
8937 (gimple_fold_builtin): Call them.
8939 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8941 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
8942 as unlikely; update profile.
8944 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8946 * predict.c (force_edge_cold): Handle declaring edges impossible
8949 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8951 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
8953 (try_unroll_loop_completely): Fix reporting.
8955 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8957 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
8959 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
8962 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
8963 if given a non-constant argument for an intrinsic which requires a
8966 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8968 * profile.c (compare_freqs): New function.
8969 (branch_prob): Sort edge list.
8970 (find_spanning_tree): Assume that the list is priority sorted.
8972 2017-06-16 Richard Biener <rguenther@suse.de>
8974 PR tree-optimization/81090
8975 * passes.def (pass_record_bounds): Remove.
8976 * tree-pass.h (make_pass_record_bounds): Likewise.
8977 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
8978 make_pass_record_bounds): Likewise.
8979 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
8980 not free niter estimates at the beginning but at the end.
8981 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
8983 2017-06-16 Richard Biener <rguenther@suse.de>
8985 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
8986 initializer to workaround ICE in host GCC 4.8.
8988 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8990 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
8992 (clone_inlined_nodes): Update.
8994 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8996 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
8997 prefetch settings, and enable prefetching by default at -O3.
8999 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9001 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9002 Set flag_prefetch_loop_arrays according to tuning data.
9004 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9006 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9008 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9009 [Unrelated to main purpose of the patch] Place the pointer field last
9010 to enable type checking errors when tune structure are wrongly merged.
9011 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9012 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9013 (thunderx2t99_prefetch_tune): New tune constants.
9014 (tune_params *_tunings): Update all tunings (no functional change).
9015 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9016 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9017 from tunings structures.
9019 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9022 * ubsan.c (instrument_null): Add T argument, use it instead
9023 of computing it based on IS_LHS.
9024 (instrument_object_size): Likewise.
9025 (pass_ubsan::execute): Adjust instrument_null and
9026 instrument_object_size callers to pass gimple_get_lhs or
9027 gimple_assign_rhs1 result to it. Use instrument_null instead of
9028 calling get_base_address and instrument_mem_ref. Handle
9029 aggregate call arguments for object-size sanitization.
9031 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9033 PR tree-optimization/81089
9034 * tree-vrp.c (is_masked_range_test): Validate operands of
9037 2017-06-15 Martin Sebor <msebor@redhat.com>
9040 * dumpfile.c (dump_register): Avoid calling memset to initialize
9041 a class with a default ctor.
9042 * gcc.c (struct compiler): Remove const qualification.
9043 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9044 * hash-table.h: Ditto.
9045 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9047 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9048 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9050 * params.h (struct param_info): Make struct members non-const.
9051 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9052 with default initialization.
9053 * vec.h (vec_copy_construct, vec_default_construct): New helper
9055 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9056 with vec_copy_construct.
9057 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9058 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9059 * doc/invoke.texi (-Wclass-memaccess): Document.
9061 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9063 * emit-rtl.h (is_leaf): Update comment about local
9066 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9069 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9070 for a variable to have a section before checking if the section has a
9072 Set section to.persistent if persistent attribute is set.
9073 Warn if .persistent attribute is used on an automatic variable.
9075 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9077 PR rtl-optimization/80474
9078 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9080 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9082 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9085 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9087 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9088 (casesi): Emit bounds checking as RTL.
9089 (casesi_internal_mips16_<mode>): Remove bounds checking.
9091 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9093 * config/xtensa/xtensa.c (xtensa_option_override): Append
9094 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9095 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9096 xtensa_doloop_hooks): Define unconditionally.
9097 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9099 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9100 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9101 for it in xtensa_option_override.
9102 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9103 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9105 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9107 * doc/cppopts.texi: Document '-' special value to -MF.
9109 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9111 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9112 (cortex_a53_fconst): Likewise.
9113 (cortex_a53_fpmul): Likewise.
9114 (cortex_a53_f_load_64): Likewise.
9115 (cortex_a53_f_load_many): Likewise.
9116 (cortex_a53_advsimd_alu): Likewise.
9117 (cortex_a53_advsimd_alu_q): Likewise.
9118 (cortex_a53_advsimd_mul): Likewise.
9119 (cortex_a53_advsimd_mul_q): Likewise.
9120 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9121 Add missing fmul, r2f_cvt and fconst cases.
9123 2017-06-14 Richard Biener <rguenther@suse.de>
9126 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9128 (fold_binary_loc): When associating do not treat pre-existing
9129 TREE_OVERFLOW on literal constants as a reason to allow
9130 TREE_OVERFLOW on associated literal constants.
9132 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9134 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9135 (MASK_FEATURES): New macro.
9136 * config/sparc/sparc.c (sparc_option_override): Remove the special
9137 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9139 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9141 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9142 a division of 0 if non-call exceptions are enabled.
9144 2017-06-14 Andrew Pinski <apinski@cavium.com>
9145 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9148 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9149 Improve vector initialization code gen for only variable case.
9151 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9153 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9155 2017-06-14 Richard Biener <rguenther@suse.de>
9157 PR tree-optimization/81083
9158 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9161 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9163 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9164 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9165 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9166 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9167 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9168 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9170 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9172 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9173 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9175 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9177 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9179 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9181 * config/rs6000/t-rtems: Don't handle SPE.
9183 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9185 * config/rs6000/t-linux: Don't handle SPE.
9187 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9189 * config/rs6000/eabispe.h: Delete file.
9191 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9193 * config/rs6000/t-spe: Delete file.
9195 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9197 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9198 (rs6000_legitimate_offset_address_p): Return false for anything in
9199 V2SImode or V2SFmode.
9201 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9203 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9204 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9205 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9207 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9208 (rs6000_legitimate_offset_address_p): Ditto.
9209 (rs6000_emit_move): Ditto.
9210 (rs6000_init_builtins): Remove V4HI_type_node.
9212 2017-06-13 Martin Liska <mliska@suse.cz>
9215 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9216 (gate_asan): Likewise.
9217 * asan.h (asan_no_sanitize_address_p): Remove the function.
9218 (sanitize_flags_p): New function.
9219 * builtins.def: Fix coding style.
9220 * common.opt: Use renamed enum value.
9221 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9222 * doc/extend.texi: Document no_sanitize attribute.
9223 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9224 to SANITIZE_UNDEFINED_NONDEFAULT.
9225 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9226 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9227 Use sanitize_flags_p.
9228 * gimplify.c (gimplify_function_tree): Likewise.
9229 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9230 * opts.c (parse_no_sanitize_attribute): New function.
9231 (common_handle_option): Use renamed enum value.
9232 * opts.h (parse_no_sanitize_attribute): Declare.
9233 * tree.c (sanitize_flags_p): New function.
9234 * tree.h: Declared here.
9235 * tsan.c: Use sanitize_flags_p.
9236 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9237 (instrument_mem_ref): Likewise.
9238 (instrument_bool_enum_load): Likewise.
9239 (do_ubsan_in_current_function): Remove the function.
9240 (pass_ubsan::execute): Use sanitize_flags_p.
9241 * ubsan.h: Remove do_ubsan_in_current_function
9242 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9243 (dump_function_to_file): Use it here.
9245 2017-06-13 Martin Jambor <mjambor@suse.cz>
9247 PR tree-optimization/80803
9248 PR tree-optimization/81063
9249 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9250 (propagate_subaccesses_across_link): Enqueue subtree whenever
9251 necessary instead of relying on the caller.
9253 2017-06-13 Martin Jambor <mjambor@suse.cz>
9255 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9256 that have a first_link.
9257 (sort_and_splice_var_accesses): Do not check first_link before
9259 (subtree_mark_written_and_enqueue): Likewise.
9260 (propagate_all_subaccesses): Likewise and do not stop at first
9261 parent with a first_link.
9263 2017-06-13 Martin Jambor <mjambor@suse.cz>
9265 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9268 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9270 * match.pd: New pattern.
9272 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9274 * tree-vrp.c (is_masked_range_test): New function.
9275 (register_edge_assert_for): Determine ranges for
9278 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9280 PR tree-optimization/67328
9281 * fold-const.c (maskable_range_p): New function.
9282 (build_range_check): Generate bittests if possible.
9284 2017-06-13 Martin Liska <mliska@suse.cz>
9286 * gimple-pretty-print.c (dump_probability): Add new argument.
9287 (dump_edge_probability): Dump both probability and count.
9288 (dump_gimple_label): Likewise.
9289 (dump_gimple_bb_header): Likewise.
9291 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9294 * config/avr/avr-devices.c: Fix indentation.
9295 * config/avr/gen-avr-mmcu-specs.c: Dito.
9297 2017-06-13 Richard Biener <rguenther@suse.de>
9299 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9300 instead get vector type from stmt_info.
9301 (vectorizable_reduction): Adjust. Remove dead code.
9303 2017-06-13 Richard Biener <rguenther@suse.de>
9306 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9307 case of C * (x * C2 + C3).
9308 (fold_addr_of_array_ref_difference): Properly fold index difference.
9310 2017-06-12 David S. Miller <davem@davemloft.net>
9313 * config/sparc/sparc.md (return expander): Emit frame blockage if
9314 function uses alloca.
9316 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9318 * combine.c (make_field_assignment): Check len rather than the mode
9319 precision when calling force_to_mode.
9321 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9323 Support multilibs and devices that see flash in RAM address range.
9326 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9327 (avr_mcu_t) <flash_pm_offset>: New field.
9328 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9329 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9330 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9331 (AVR_TINY_PM_OFFSET): Remove macro.
9332 * config/avr/avr.opt (-mshort-calls): New option.
9333 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9334 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9335 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9336 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9337 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9338 instead of avr_arch->have_jmp_call.
9339 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9340 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9341 avr_arch->flash_pm_offset to define.
9342 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9343 new field flash_pm_offset. Add entry for avrxmega3.
9344 (avr_texinfo): Add entry for avrxmega3.
9345 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9346 attiny212, attiny214,
9347 attiny412, attiny414, attiny416, attiny417,
9348 attiny814, attiny816, attiny817,
9349 attiny1614, attiny1616, attiny1617,
9350 attiny3214, attiny3216, attiny3217.
9351 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9352 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9353 (avr_print_operand_address) [AVR_TINY]: Same.
9354 (avr_asm_init_sections) <readonly_data_section>: Only patch
9355 callback if avr_arch->flash_pm_offset = 0.
9356 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9357 for rodata if avr_arch->flash_pm_offset != 0.
9358 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9359 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9360 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9361 (m_options): Append opt_rcall.
9362 (m_dirnames): Append dir_rcall.
9363 * config/avr/t-multilib: Regenerate.
9365 * configure.ac [target=avr]: Check whether avrxmega3 default
9366 linker description file works as needed.
9367 * configure: Regenerate.
9368 * doc/avr-mmcu.texi: Regenerate.
9369 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9370 <__AVR_ARCH__>: Document avrxmega3 and 103.
9371 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9372 <__AVR_SHORT_CALLS__>: Document it.
9373 <__AVR_PM_BASE_ADDRESS__>: Document it.
9374 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9375 (AVR Variable Attributes) <progmem>: Document this is
9376 not needed for avrxmega3.
9377 (AVR Named Address Spaces) <__flash>: Dito.
9379 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9381 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9383 2017-06-12 Doug Rupp <rupp@adacore.com>
9385 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9386 Append vxworks-stdint.h to the tm_file list.
9387 * config/vxworks-stdint.h: New file.
9389 2017-06-12 Martin Liska <mliska@suse.cz>
9391 PR tree-optimization/81041
9392 * tree-profile.c (gimple_gen_ic_func_profiler):
9393 Create an extra BB in profile-generate
9394 (gimple_gen_time_profiler): Likewise.
9396 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9398 PR tree-optimization/81003
9399 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9400 (update_range_test): Use it instead of force_gimple_operand_gsi.
9402 2017-06-12 Richard Biener <rguenther@suse.de>
9404 PR tree-optimization/81053
9405 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9406 with backedge value not defined in loop. Simplify def stmt
9409 2017-06-11 Tom de Vries <tom@codesourcery.com>
9412 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9414 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9415 nvptx_cannot_force_const_mem.
9417 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9419 * opts.c (finish_options): Move test for flag_split_stack after
9420 it has been initialized.
9422 2017-06-11 Jason Merrill <jason@redhat.com>
9424 * tree.h (id_equal): New.
9425 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9426 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9427 instead of strcmp of IDENTIFIER_POINTER.
9429 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9431 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9432 (mark_all_inlined_calls_cdtor): Fix formating.
9433 (inline_transform): Rescale profile before inlining.
9435 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9437 * cgraph.h (cgraph_edge::clone): Update prototype.
9438 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9439 (cgraph_node::create_clone): Update.
9440 (cgraph_node::create_version_clone): Update.
9441 * tree-inline.c (copy_bb): Update.
9442 (expand_call_inline): Update.
9444 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9446 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9447 factored out from ...
9448 (rs6000_emit_prologue): ... here.
9450 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9452 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9453 factored out from ...
9454 (rs6000_emit_prologue): ... here.
9456 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9458 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9460 (handle_missing_profiles): Fix computation of tp_first_run.
9461 (counts_to_freqs): Do not touch freqs when count is 0.
9463 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9465 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9468 2017-06-10 Tom de Vries <tom@codesourcery.com>
9470 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9471 attributes): Document signal effective target.
9473 2017-06-10 Tom de Vries <tom@codesourcery.com>
9475 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9476 Document effective target stack_size.
9478 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9480 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9481 to the edit_context if they can be auto-applied.
9483 2017-06-9 Ian Lance Taylor <iant@golang.org>
9485 * opts.c (finish_options): If -fsplit-stack, disable implicit
9486 -forder-blocks-and-partition.
9487 * doc/invoke.texi (Optimize Options): Document that when using
9488 -fsplit-stack -forder-blocks-and-partition is not implicitly
9491 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9493 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9494 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9495 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9496 * builtins.def (abort, trap, unreachable): Declare cold.
9497 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9498 * tree-core.h (ECF_COLD): New.
9499 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9500 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9502 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9504 * predict.c (unlikely_executed_stmt_p): Cleanup.
9506 2017-06-09 Richard Biener <rguenther@suse.de>
9508 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9509 model if the ref is always written to.
9511 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9513 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9515 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9517 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9520 2017-06-09 Tom de Vries <tom@codesourcery.com>
9523 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9524 "target cannot support label values" when encountering LABEL_REF.
9526 2017-06-09 Martin Liska <mliska@suse.cz>
9528 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9529 (gimple_gen_ic_func_profiler): Emit direct comparison
9530 of __gcov_indirect_call_callee with NULL.
9531 (gimple_gen_time_profiler): Change probability from
9532 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9534 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9536 * profile.c (edge_gcov_counts): Turn to pointer.
9537 (compute_branch_probabilities, compute_branch_probabilities): Update.
9538 (branch_prob): Do not clear edge_gcov_count.
9539 * profile.h (edge_gcov_counts): Turn to pointer.
9540 (edge_gcov_count): Update.
9542 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9544 * gimple.h (gimple_check_failed): Mark cold.
9546 2017-06-09 Richard Biener <rguenther@suse.de>
9548 PR tree-optimization/66623
9549 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9550 refactor check_reduction into two parts, properly computing
9551 whether we have to check reduction validity for outer loop
9554 2017-06-09 Richard Biener <rguenther@suse.de>
9556 PR tree-optimization/79483
9557 * graphite-scop-detection.c (order): New global.
9558 (get_order): Compute bb to order mapping that satisfies code
9559 generation constraints.
9560 (cmp_pbbs): New helper.
9561 (build_scops): Start domwalk at entry block, sort generated
9564 2017-06-09 Richard Biener <rguenther@suse.de>
9567 * ipa-polymorphic-call.c
9568 (ipa_polymorphic_call_context::restrict_to_inner_class):
9569 Skip FIELD_DECLs with error_mark_node type.
9570 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9573 2017-06-09 Martin Liska <mliska@suse.cz>
9575 * predict.c (struct branch_predictor): New struct.
9576 (test_prediction_value_range): New test.
9577 (predict_c_tests): New function.
9578 * selftest-run-tests.c (selftest::run_tests): Run the function.
9579 * selftest.h: Declare new tests.
9581 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9584 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9585 gen_add3_insn did not fail.
9586 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9587 r0, construct that number in a temporary reg and add that reg to r0.
9588 If asked to put the result in r0 as well, fail.
9590 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9592 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9593 for early expansion of vec_eqv.
9595 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9598 * ubsan.c (instrument_null): Avoid pointless code temporary.
9599 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9601 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9605 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
9608 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
9611 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9613 * predict.c (unlikely_executed_edge_p): Move ahead.
9614 (probably_never_executed_edge_p): Use it.
9616 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9619 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
9620 gimple_call_builtin_p call.
9622 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9624 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
9625 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
9626 rtl_check_failed_type2, rtl_check_failed_code1,
9627 rtl_check_failed_code2, rtl_check_failed_code_mode,
9628 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
9629 rtvec_check_failed_bounds, rtl_check_failed_flag,
9630 _fatal_insn_not_found, _fatal_insn): Likewise.
9631 * tree.h (tree_contains_struct_check_failed,
9632 tree_check_failed, tree_not_check_failed,
9633 tree_class_check_failed, tree_range_check_failed,
9634 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
9635 tree_vec_elt_check_failed, phi_node_elt_check_failed,
9636 tree_operand_check_failed, omp_clause_check_failed,
9637 omp_clause_operand_check_failed, omp_clause_range_check_failed):
9640 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9642 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
9643 flag_branch_probabilities.
9644 * ipa-inline.c (edge_badness): Likewise.
9645 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9646 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
9647 * predict.c (maybe_hot_frequency_p): Likewise.
9648 (probably_never_executed): Likewise.
9649 * sched-ebb.c (schedule_ebbs): Likewise.
9650 * sched-rgn.c (find_single_block_region): Likewise.
9651 * tracer.c (tail_duplicate): Likewise.
9653 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9655 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
9656 longer requires x_flag_profile_use.
9658 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9660 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
9661 instead of flag_reorder_blocks_and_partition.
9662 * dbxout.c (dbxout_function_end): Likewise.
9663 * dwarf2out.c (gen_subprogram_die): Likewise.
9664 * haifa-sched.c (sched_create_recovery_edges): Likewise.
9665 * hw-doloop.c (reorg_loops): Likewise.
9666 * varasm.c (assemble_start_function,
9667 assemble_end_function): Likewise.
9668 (decide_function_section): Do not check for
9669 flag_reorder_blocks_and_partition.
9671 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9673 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9675 (chkp_get_hard_register_fake_addr_expr): Ditto.
9676 (chkp_build_addr_expr): Add check for hard reg case.
9677 (chkp_parse_array_and_component_ref): Ditto.
9678 (chkp_find_bounds_1): Ditto.
9679 (chkp_process_stmt): Don't generate bounds store for
9682 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9684 * predict.c (maybe_hot_bb_p): Do not check profile status.
9685 (maybe_hot_edge_p): Likewise.
9686 (probably_never_executed): Check for zero counts even if profile
9688 (unlikely_executed_edge_p): New function.
9689 (unlikely_executed_stmt_p): New function.
9690 (unlikely_executed_bb_p): New function.
9691 (set_even_probabilities): Use unlikely predicates.
9692 (combine_predictions_for_bb): Likewise.
9693 (predict_paths_for_bb): Likewise.
9694 (predict_paths_leading_to_edge): Likewise.
9695 (determine_unlikely_bbs): New function.
9696 (estimate_bb_frequencies): Use it.
9697 (compute_function_frequency): Use zero counts even if profile is
9699 * profile-count.h: Fix typo.
9701 2017-08-08 Julia Koval <julia.koval@intel.com>
9703 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
9704 _mm512_mask_cvtsepi16_storeu_epi8,
9705 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
9706 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
9707 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
9708 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
9709 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
9710 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
9711 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
9712 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
9713 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
9714 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
9715 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
9716 __builtin_ia32_pmovuswb256mem_mask,
9717 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
9718 __builtin_ia32_pmovwb512mem_mask): New builtins.
9720 2017-08-08 Julia Koval <julia.koval@intel.com>
9722 PR target/73350,80862
9723 * config/i386/subst.md (round): Fix round pattern.
9724 * config/i386/i386.c (ix86_erase_embedded_rounding):
9725 Fix erasing rounding for the fixed pattern.
9727 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9729 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
9731 2017-06-08 Martin Liska <mliska@suse.cz>
9733 PR gcov-profile/80911
9734 * gcov.c (block_info::block_info): New constructor.
9736 2017-06-07 Carl Love <cel@us.ibm.com>
9738 * config/rs6000/rs6000-c: The return type of the following
9739 built-in functions was implemented as int not long long. Fix sign
9740 of return value for the unsigned version of vec_mulo and vec_mule.
9741 vector unsigned long long vec_bperm (vector unsigned long long,
9742 vector unsigned char)
9743 vector signed long long vec_mule (vector signed int,
9745 vector unsigned long long vec_mule (vector unsigned int,
9746 vector unsigned int)
9747 vector signed long long vec_mulo (vector signed int,
9749 vector unsigned long long vec_mulo (vector unsigned int,
9750 vector unsigned int)
9751 * doc/extend.texi: Fix the documentation for the built-in
9754 2017-06-07 Carl Love <cel@us.ibm.com>
9757 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
9760 2017-06-07 Carl Love <cel@us.ibm.com>
9762 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
9763 support, Generate doublehv for signed int/float for BE case only.
9765 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
9767 * doc/invoke.texi (mcx16): Rewrite.
9769 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9771 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
9772 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
9773 *mov<mode>_softfloat, and an anonymous splitter): Use
9774 nonimmediate_operand instead of rs6000_nonimmediate_operand.
9776 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9778 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
9780 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9781 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
9782 (rs6000_debug_reg_global): Adjust.
9783 (rs6000_init_hard_regno_mode_ok): Adjust.
9784 (rs6000_dbx_register_number): Adjust.
9785 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
9786 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9787 Remove SPE_ACC and SPEFSCR.
9788 (REG_ALLOC_ORDER): Ditto.
9789 (FRAME_POINTER_REGNUM): Change to 111.
9790 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
9791 (REG_CLASS_NAMES): Ditto.
9792 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
9793 (REGISTER_NAMES): Ditto.
9794 (ADDITIONAL_REG_NAMES): Ditto.
9795 (rs6000_reg_names): Ditto.
9796 * config/rs6000/rs6000.md: Renumber some register number
9799 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9801 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
9803 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9804 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
9806 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
9807 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9808 Delete the SPE high registers.
9809 (REG_ALLOC_ORDER): Ditto.
9810 (enum reg_class): Remove SPE_HIGH_REGS.
9811 (REG_CLASS_NAMES): Ditto.
9812 (REG_CLASS_CONTENTS): Delete the SPE high registers.
9813 (REGISTER_NAMES): Ditto.
9814 (rs6000_reg_names): Ditto.
9815 * doc/tm.texi.in: Remove SPE as example.
9816 * doc/tm.texi: Regenerate.
9818 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9820 * config/rs6000/8540.md (ppc8540_brinc): Delete.
9821 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
9822 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
9823 * config/rs6000/rs6000.md (type): Remove "brinc".
9825 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9827 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
9828 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
9829 * config/rs6000/linuxspe.h: Delete file.
9830 * config/rs6000/rs6000.md: Don't include spe.md.
9831 * config/rs6000/spe.h: Delete file.
9832 * config/rs6000/spe.md: Delete file.
9833 * config/rs6000/t-rs6000: Remove spe.md.
9835 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9837 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
9838 (reg_or_none500mem_operand): Delete.
9839 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
9840 instead of reg_or_none500mem_operand.
9842 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9844 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
9845 handling of SPE flags.
9846 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
9848 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9850 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
9852 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
9853 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
9854 paired_divv2sf3): Similar.
9855 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
9856 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
9857 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
9859 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
9860 Rename the paired_* instruction patterns.
9861 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
9863 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
9864 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
9865 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
9866 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
9867 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
9869 (struct machine_function): Delete field spe_insn_chain_scanned_p.
9870 (spe_func_has_64bit_regs_p): Delete.
9871 (spe_expand_predicate_builtin): Delete.
9872 (spe_expand_evsel_builtin): Delete.
9873 (TARGET_DWARF_REGISTER_SPAN): Do not define.
9874 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
9875 (invalid_e500_subreg): Delete.
9876 (rs6000_legitimize_address): Always force_reg op2 as well, for
9877 paired single memory accesses.
9878 (rs6000_member_type_forces_blk): Delete.
9879 (rs6000_spe_function_arg): Delete.
9880 (rs6000_expand_unop_builtin): Delete SPE handling.
9881 (rs6000_expand_binop_builtin): Ditto.
9882 (spe_expand_stv_builtin): Delete.
9883 (bdesc_2arg_spe): Delete.
9884 (spe_expand_builtin): Delete.
9885 (spe_expand_predicate_builtin): Delete.
9886 (spe_expand_evsel_builtin): Delete.
9887 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
9888 (spe_init_builtins): Delete.
9889 (spe_func_has_64bit_regs_p): Delete.
9890 (savres_routine_name): Delete "info" parameter. Adjust callers.
9891 (rs6000_emit_stack_reset): Ditto.
9892 (rs6000_dwarf_register_span): Delete.
9893 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
9894 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
9895 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
9897 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
9899 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
9900 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
9901 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
9902 mulv2sf3, divv2sf3): Delete expanders.
9904 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9906 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
9908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9910 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
9911 * config/rs6000/rs6000.c: Ditto.
9913 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9915 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
9916 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
9917 comparison_operator.
9919 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9921 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
9922 * config/rs6000/rs6000.opt: Ditto.
9923 * config/rs6000/t-rtems: Ditto.
9925 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9927 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
9928 TARGET_E500_SINGLE by 0, simplify.
9929 * config/rs6000/rs6000.c: Ditto.
9930 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
9931 (spe_build_register_parallel): Delete.
9932 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
9933 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
9934 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
9935 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
9936 (E500_CONVERT): Delete.
9937 * config/rs6000/spe.md: Remove many patterns and all define_constants.
9939 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9941 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
9942 * config/rs6000/dfp.md: Ditto.
9943 (negdd2, *negdd2_fpr): Merge.
9944 (absdd2, *absdd2_fpr): Merge.
9945 (negtd2, *negtd2_fpr): Merge.
9946 (abstd2, *abstd2_fpr): Merge.
9947 * config/rs6000/e500.h: Delete file.
9948 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
9949 TARGET_FPRS by 1 and simplify.
9950 * config/rs6000/rs6000-c.c: Ditto.
9951 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
9953 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
9955 * config/rs6000/rs6000.md: Ditto.
9956 (floatdidf2, *floatdidf2_fpr): Merge.
9957 (move_from_CR_gt_bit): Delete.
9958 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
9959 (E500_CR_IOR_COMPARE): Delete.
9960 (All patterns that require !TARGET_FPRS): Delete.
9961 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
9963 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9965 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
9967 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9969 * graphds.c (add_edge): Intitialize edge's attached data.
9970 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
9971 pointer parameter. Call pointed function on each edge during
9972 graph traversing. Skip traversing the edge when the function
9974 (graphds_dfs, graphds_scc): Ditto.
9975 (for_each_edge): New parameter. Pass the new parameter to callback
9977 * graphds.h (skip_edge_callback): New function pointer type.
9978 (graphds_dfs, graphds_scc): New function pointer parameter.
9979 (graphds_edge_callback, for_each_edge): New parameter.
9981 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9983 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
9984 out code checking if runtime alias check is possible to below ...
9985 Call the new function.
9986 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
9987 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
9989 2017-06-07 Marek Polacek <polacek@redhat.com>
9992 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
9993 TYPE_OVERFLOW_WRAPS check.
9995 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9997 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
9998 if versioning is required.
9999 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10000 peeling with the check for versioning.
10002 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10004 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10005 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10006 Set true to new parameter if new ssa variable is defined.
10007 (vect_gen_vector_loop_niters): Refactor. Set range information
10008 for the new vector loop bound variable.
10009 (vect_do_peeling): Ditto.
10011 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10013 * tree-affine.c (ssa.h): Include header file.
10014 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10015 has wrapping overflow behavior.
10017 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10019 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10021 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10023 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10024 (tree_to_aff_combination): ... here.
10026 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10028 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10029 reg_pressure model function.
10030 (ivopts_global_cost_for_size): Delete.
10031 (determine_set_costs, iv_ca_recount_cost): Call new model function
10032 ivopts_estimate_reg_pressure.
10034 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10036 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10037 expensive than udiv. Remove floating point cases from mod.
10039 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10041 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10042 Increase idiv cost.
10044 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10046 * config/aarch64/aarch64.md
10047 (copysignsf3): Fix mask generation.
10049 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10051 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10053 (class dump_manager): Add register_dumps method.
10054 * dumpfile.c: Include langhooks.h.
10055 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10056 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10057 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10058 (dump_manager::dump_register): Start with 512 entries instead of 32.
10059 (dump_manager::register_dumps): New method.
10060 * toplev.c (general_init): Instead of invoking register_dumps
10061 langhook, invoke register_dumps method on the dump manager.
10062 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10065 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10067 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10068 Say that # requires an associated define_split to exist, and that
10069 the define_split must be suitable for use after register allocation.
10071 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10073 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10074 (compute_outgoing_frequencies): Also initialize zero counts.
10075 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10076 around loops; preserve more of profile when nothing changes.
10078 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10080 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10082 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10083 * config/arm/arm-cpu-cdata.h: Regenerate.
10084 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10085 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10086 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10087 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10088 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10090 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10092 * config/arm/t-rmprofile: Likewise.
10093 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10095 2017-06-06 David S. Miller <davem@davemloft.net>
10098 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10099 blockage if function uses alloca.
10101 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10103 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10104 New "uid" fields to hold pretty-print IDs of group and ref.
10105 Memory references are now identified as <group_id>:<ref_id>
10106 instead of using [random] addresses.
10107 (dump_mem_details): Simplify, no functional change.
10108 (dump_mem_ref): Simplify and make output more concise.
10109 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10110 (find_or_create_group): Initialize group uid.
10111 (record_ref): Initialize ref uid. Improve debug output.
10112 (prune_group_by_reuse, should_issue_prefetch_p,)
10113 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10114 (mark_nontemporal_store, determine_loop_nest_reuse):
10115 Improve debug output.
10117 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10119 * dbgcnt.def (prefetch): New debug counter.
10120 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10121 (schedule_prefetches): Stop issueing prefetches if debug counter
10124 2017-06-06 Tom de Vries <tom@codesourcery.com>
10126 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10127 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10129 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10131 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10132 Use aarch64_reg_or_zero predicate for operand 4.
10133 (aarch64_compare_and_swap<mode> define_insn_and_split):
10134 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10135 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10137 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10139 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10140 (arm_compute_save_core_reg_mask): This.
10141 (thumb1_compute_save_reg_mask): Rename into ...
10142 (thumb1_compute_save_core_reg_mask): This.
10143 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10144 (arm_compute_frame_layout): Likewise.
10146 2017-06-06 Richard Biener <rguenther@suse.de>
10148 PR tree-optimization/80974
10149 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10150 keep or clear leaders SSA info.
10152 2017-06-06 Tom de Vries <tom@codesourcery.com>
10154 * config/nvptx/nvptx.c (split_mode_p): New function.
10155 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10157 2017-06-06 Tom de Vries <tom@codesourcery.com>
10159 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10161 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10164 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10167 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10169 * shrink-wrap.c (handle_simple_exit): Update profile.
10170 (try_shrink_wrapping): Upate profile.
10172 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10174 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10175 (tree_guess_outgoing_edge_probabilities): New.
10176 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10177 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10179 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10181 * ipa-split.c (split_function): Initialize return bb profile.
10183 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10185 * profile.c (compute_branch_probabilities): Also initialize
10186 EXIT_BLOCK profile.
10188 2017-06-06 Richard Biener <rguenther@suse.de>
10190 PR tree-optimization/80928
10191 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10192 (vect_analyze_loop_operations): Properly guard analysis for
10194 (vect_transform_loop): Likewise.
10195 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10196 (vect_model_induction_cost): Do not cost for pure SLP.
10197 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10198 of induction in inner loop vectorization.
10199 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10200 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10201 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10203 (vect_analyze_slp_cost_1): Cost induction.
10204 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10205 (vect_get_slp_vect_defs): Likewise.
10206 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10207 (vect_transform_stmt): Handle SLP reductions.
10208 * tree-vectorizer.h (vectorizable_induction): Adjust.
10210 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10212 * config/rs6000/rs6000.c (make_resolver_func): Update
10213 init_lowered_empty_function call.
10215 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10217 * doc/invoke.texi: Document the -fprofile-abs-path option.
10218 * common.opt (fprofile-abs-path): New option.
10219 * gcov-io.h (gcov_write_filename): Declare.
10220 * gcov-io.c (gcov_write_filename): New function.
10221 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10222 * profile.c (output_location): Likewise.
10224 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10226 * shring-wrap.c: Revert accidental commit.
10228 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10230 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10232 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10234 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10236 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10237 profile in callgraph edge.
10238 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10239 (apply_scale): Likewise.
10240 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10241 Also scale profile when inlining function with zero profile.
10242 (initialize_cfun): Update exit block profile even when it is zero.
10243 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10244 when profile is read.
10246 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10248 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10249 (CLONE_*): New constants to define the processors we can generate
10250 code for with the target_clone attribute.
10251 (rs6000_clone_map): New array to identify which clone processors
10252 the current program is running on.
10253 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10254 target_clone attribute.
10255 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10256 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10257 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10258 (cpu_expand_builtin): Add support for target_clone attribute.
10259 (rs6000_valid_attribute_p): Allow "default" attribute.
10260 (get_decl_name): New debug function to simplify printing the
10261 current function name in debugging statements.
10262 (rs6000_clone_priority): New functions to support the target_clone
10263 attribute, and be able to generate code to switch between ISA 2.05
10264 through ISA 3.0 (power6 through power9).
10265 (rs6000_compare_version_priority): Likewise.
10266 (rs6000_get_function_versions_dispatcher): Likewise.
10267 (make_resolver_func): Likewise.
10268 (add_condition_to_bb): Likewise.
10269 (dispatch_function_versions): Likewise.
10270 (rs6000_generate_version_dispatcher_body): Likewise.
10271 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10272 (fusion_gpr_load_p): Fix a spacing issue.
10273 * doc/extend.texi (Common Function Attributes): Document that the
10274 PowerPC supports the target_clone attribute.
10276 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10278 * config/arm/arm.h: explain F symbol found in description of ARM
10279 register allocation in its legend.
10281 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10283 * config/mips/frame-header-opt.c: Include profile-count.h.
10284 * config/riscv/riscv.c: Include profile-count.h
10286 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10288 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10290 (sm_set_flag_if_changed): Add bbs field.
10291 (execute_sm_if_changed_flag_set): Pass BBS.
10292 (execute_sm): Update.
10294 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10296 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10299 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10301 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10302 (peephole2): New peephole2 to emit the above.
10303 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10305 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10307 * config/aarch64/aarch64.c (define_peephole2 above
10308 *sub_<shift>_<mode>): New peephole.
10310 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10312 * config/i386/i386.c (make_resolver_func): Update.
10313 * Makefile.in: Add profile-count.h and profile-count.o
10314 * auto-profile.c (afdo_indirect_call): Update to new API.
10315 (afdo_set_bb_count): Update.
10316 (afdo_propagate_edge): Update.
10317 (afdo_propagate_circuit): Update.
10318 (afdo_calculate_branch_prob): Update.
10319 (afdo_annotate_cfg): Update.
10320 * basic-block.h: Include profile-count.h
10321 (struct edge_def): Turn count to profile_count.
10322 (struct basic_block_def): Likewie.
10323 (REG_BR_PROB_BASE): Move to profile-count.h
10324 (RDIV): Move to profile-count.h
10325 * bb-reorder.c (max_entry_count): Turn to profile_count.
10326 (find_traces): Update.
10327 (rotate_loop):Update.
10328 (connect_traces):Update.
10329 (sanitize_hot_paths):Update.
10330 * bt-load.c (migrate_btr_defs): Update.
10331 * cfg.c (RDIV): Remove.
10332 (init_flow): Use alloc_block.
10333 (alloc_block): Uninitialize count.
10334 (unchecked_make_edge): Uninitialize count.
10335 (check_bb_profile): Update.
10336 (dump_edge_info): Update.
10337 (dump_bb_info): Update.
10338 (update_bb_profile_for_threading): Update.
10339 (scale_bbs_frequencies_int): Update.
10340 (scale_bbs_frequencies_gcov_type): Update.
10341 (scale_bbs_frequencies_profile_count): New.
10342 * cfg.h (update_bb_profile_for_threading): Update.
10343 (scale_bbs_frequencies_profile_count): Declare.
10344 * cfgbuild.c (compute_outgoing_frequencies): Update.
10345 (find_many_sub_basic_blocks): Update.
10346 * cfgcleanup.c (try_forward_edges): Update.
10347 (try_crossjump_to_edge): Update.
10348 * cfgexpand.c (expand_gimple_tailcall): Update.
10349 (construct_exit_block): Update.
10350 * cfghooks.c (verify_flow_info): Update.
10351 (dump_bb_for_graph): Update.
10352 (split_edge): Update.
10353 (make_forwarder_block): Update.
10354 (duplicate_block): Update.
10355 (account_profile_record): Update.
10356 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10357 (get_estimated_loop_iterations): Update.
10358 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10359 (single_likely_exit): Update.
10360 * cfgloopmanip.c (scale_loop_profile): Update.
10362 (set_zero_probability): Update.
10363 (lv_adjust_loop_entry_edge): Update.
10364 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10365 (purge_dead_edges): Update.
10366 (rtl_account_profile_record): Update.
10367 * cgraph.c (cgraph_node::create): Uninitialize count.
10368 (symbol_table::create_edge): Uninitialize count.
10369 (cgraph_update_edges_for_call_stmt_node): Update.
10370 (cgraph_edge::dump_edge_flags): Update.
10371 (cgraph_node::dump): Update.
10372 (cgraph_edge::maybe_hot_p): Update.
10373 * cgraph.h: Include profile-count.h
10374 (create_clone), create_edge, create_indirect_edge): Update.
10375 (cgraph_node): Turn count to profile_count.
10376 (cgraph_edge0: Likewise.
10377 (make_speculative, clone): Update.
10378 (create_edge): Update.
10379 (init_lowered_empty_function): Update.
10380 * cgraphclones.c (cgraph_edge::clone): Update.
10381 (duplicate_thunk_for_node): Update.
10382 (cgraph_node::create_clone): Update.
10383 * cgraphunit.c (cgraph_node::analyze): Update.
10384 (cgraph_node::expand_thunk): Update.
10385 * final.c (dump_basic_block_info): Update.
10386 * gimple-streamer-in.c (input_bb): Update.
10387 * gimple-streamer-out.c (output_bb): Update.
10388 * graphite.c (print_global_statistics): Update.
10389 (print_graphite_scop_statistics): Update.
10390 * hsa-brig.c: Include basic-block.h.
10391 * hsa-dump.c: Include basic-block.h.
10392 * hsa-gen.c (T sum_slice): Update.
10393 (convert_switch_statements):Update.
10394 * hsa-regalloc.c: Include basic-block.h.
10395 * ipa-chkp.c (chkp_produce_thunks): Update.
10396 * ipa-cp.c (struct caller_statistics): Update.
10397 (init_caller_stats): Update.
10398 (gather_caller_stats): Update.
10399 (ipcp_cloning_candidate_p): Update.
10400 (good_cloning_opportunity_p): Update.
10401 (get_info_about_necessary_edges): Update.
10402 (dump_profile_updates): Update.
10403 (update_profiling_info): Update.
10404 (update_specialized_profile): Update.
10405 (perhaps_add_new_callers): Update.
10406 (decide_about_value): Update.
10407 (ipa_cp_c_finalize): Update.
10408 * ipa-devirt.c (struct odr_type_warn_count): Update.
10409 (struct decl_warn_count): Update.
10410 (struct final_warning_record): Update.
10411 (possible_polymorphic_call_targets): Update.
10412 (ipa_devirt): Update.
10413 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10414 * ipa-icf.c (sem_function::merge): Update.
10415 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10416 * ipa-inline.c (compute_uninlined_call_time): Update.
10417 (compute_inlined_call_time): Update.
10418 (want_inline_small_function_p): Update.
10419 (want_inline_self_recursive_call_p): Update.
10420 (edge_badness): Update.
10421 (lookup_recursive_calls): Update.
10422 (recursive_inlining): Update.
10423 (inline_small_functions): Update.
10424 (dump_overall_stats): Update.
10425 (dump_inline_stats): Update.
10426 * ipa-profile.c (ipa_profile_generate_summary): Update.
10427 (ipa_propagate_frequency): Update.
10428 (ipa_profile): Update.
10429 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10430 * ipa-utils.c (ipa_merge_profiles): Update.
10431 * loop-doloop.c (doloop_modify): Update.
10432 * loop-unroll.c (report_unroll): Update.
10433 (unroll_loop_runtime_iterations): Update.
10434 * lto-cgraph.c (lto_output_edge): Update.
10435 (lto_output_node): Update.
10436 (input_node): Update.
10437 (input_edge): Update.
10438 (merge_profile_summaries): Update.
10439 * lto-streamer-in.c (input_cfg): Update.
10440 * lto-streamer-out.c (output_cfg): Update.
10441 * mcf.c (create_fixup_graph): Update.
10442 (adjust_cfg_counts): Update.
10443 (sum_edge_counts): Update.
10444 * modulo-sched.c (sms_schedule): Update.
10445 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10446 * predict.c (maybe_hot_count_p): Update.
10447 (probably_never_executed): Update.
10448 (dump_prediction): Update.
10449 (combine_predictions_for_bb): Update.
10450 (propagate_freq): Update.
10451 (handle_missing_profiles): Update.
10452 (counts_to_freqs): Update.
10453 (rebuild_frequencies): Update.
10454 (force_edge_cold): Update.
10455 * predict.h: Include profile-count.h
10456 (maybe_hot_count_p, counts_to_freqs): UPdate.
10457 * print-rtl-function.c: Do not include cfg.h
10458 * print-rtl.c: Include basic-block.h
10459 * profile-count.c: New file.
10460 * profile-count.h: New file.
10461 * profile.c (is_edge_inconsistent): Update.
10462 (correct_negative_edge_counts): Update.
10463 (is_inconsistent): Update.
10464 (set_bb_counts): Update.
10465 (read_profile_edge_counts): Update.
10466 (compute_frequency_overlap): Update.
10467 (compute_branch_probabilities): Update; Initialize and deinitialize
10469 (branch_prob): Update.
10470 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10471 (edge_gcov_count): New.
10472 (bb_gcov_count): New.
10473 * shrink-wrap.c (try_shrink_wrapping): Update.
10474 * tracer.c (better_p): Update.
10475 * trans-mem.c (expand_transaction): Update.
10476 (ipa_tm_insert_irr_call): Update.
10477 (ipa_tm_insert_gettmclone_call): Update.
10478 * tree-call-cdce.c: Update.
10479 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10480 (gimple_duplicate_sese_tail): Update.
10481 (gimple_account_profile_record): Update.
10482 (execute_fixup_cfg): Update.
10483 * tree-inline.c (copy_bb): Update.
10484 (copy_edges_for_bb): Update.
10485 (initialize_cfun): Update.
10486 (freqs_to_counts): Update.
10487 (copy_cfg_body): Update.
10488 (expand_call_inline): Update.
10489 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10490 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10491 (try_unroll_loop_completely): Update.
10492 (try_peel_loop): Update.
10493 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10494 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10495 * tree-ssa-loop-split.c (connect_loops): Update.
10496 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10497 * tree-ssa-reassoc.c (branch_fixup): Update.
10498 * tree-ssa-tail-merge.c (replace_block_by): Update.
10499 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10500 (compute_path_counts): Update.
10501 (update_profile): Update.
10502 (recompute_probabilities): Update.
10503 (update_joiner_offpath_counts): Update.
10504 (estimated_freqs_path): Update.
10505 (freqs_to_counts_path): Update.
10506 (clear_counts_path): Update.
10507 (ssa_fix_duplicate_block_edges): Update.
10508 (duplicate_thread_path): Update.
10509 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10510 (struct switch_conv_info): Update.
10511 * tree-tailcall.c (decrease_profile): Update.
10512 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10513 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10514 * value-prof.c (check_counter): Update.
10515 (gimple_divmod_fixed_value): Update.
10516 (gimple_mod_pow2): Update.
10517 (gimple_mod_subtract): Update.
10518 (gimple_ic_transform): Update.
10519 (gimple_stringop_fixed_value): Update.
10520 * value-prof.h (gimple_ic): Update.
10522 2017-06-02 Carl Love <cel@us.ibm.com>
10524 * config/rs6000/rs6000-c: Add support for built-in functions
10525 vector double vec_doublee (vector signed int);
10526 vector double vec_doublee (vector unsigned int);
10527 vector double vec_doublee (vector float);
10528 vector double vec_doubleh (vector signed int);
10529 vector double vec_doubleh (vector unsigned int);
10530 vector double vec_doubleh (vector float);
10531 vector double vec_doublel (vector signed int);
10532 vector double vec_doublel (vector unsigned int);
10533 vector double vec_doublel (vector float);
10534 vector double vec_doubleo (vector signed int);
10535 vector double vec_doubleo (vector unsigned int);
10536 vector double vec_doubleo (vector float);.
10537 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10538 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10540 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10541 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10542 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10544 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10545 vec_doublel, vec_doubleh.
10546 * doc/extend.texi: Update the built-in documentation file for the
10547 new built-in functions.
10549 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10552 * ipa-inline-analysis.c (free_growth_caches): Set
10553 edge_removal_hook_holder to NULL after removing it.
10555 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10557 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10558 comparision with zero.
10560 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10561 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10562 for early expansion of vec_min and vec_max builtins.
10563 (builtin_function_type): Add min/max unsigned variants to those
10564 identified as having unsigned arguments.
10566 2017-06-02 Olivier Hainque <hainque@adacore.com>
10568 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10570 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10572 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10573 Use VALL_F16 iterator rather than VALL.
10575 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10577 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10578 Emit CBNZ inside loop when doing a strong exchange and comparing
10579 against zero. Generate the CC flags after the loop.
10581 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10583 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10584 (dl_section_ref): New.
10585 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10586 On AIX, append an expression to subtract the size of the
10587 section length to dl_section_ref.
10589 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10591 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10592 for early expansion of vector absolute builtins.
10594 2017-06-02 Richard Biener <rguenther@suse.de>
10596 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10597 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10599 2017-06-02 Richard Biener <rguenther@suse.de>
10601 PR tree-optimization/80948
10602 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10603 stmt order as well.
10605 2017-06-02 Richard Biener <rguenther@suse.de>
10607 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
10609 * tree-vect-stmts.c (process_use): Do not mark backedge defs
10610 for inductions as relevant.
10612 2017-06-02 Richard Biener <rguenther@suse.de>
10614 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
10615 (vectorizable_induction): ... this. Remove dead code.
10617 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
10619 * builtins. (expand_builtin_alloca): Remove second parameter and
10620 infer its value from the first parameter instead.
10621 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
10623 2017-06-02 Jakub Jelinek <jakub@redhat.com>
10625 PR rtl-optimization/80903
10626 * loop-doloop.c (add_test): Unshare sequence.
10628 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
10630 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
10632 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
10634 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
10636 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
10637 xlogue_layout::get_instance, logue_layout::xlogue_layout,
10638 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
10639 (xlogue_layout::get_stub_rtx): Make static.
10640 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
10641 (xlogue_layout::compute_stub_managed_regs): Rename to...
10642 (xlogue_layout::count_stub_managed_regs): ...this.
10643 (xlogue_layout::is_stub_managed_reg): New function.
10644 (xlogue_layout::m_stub_names): Rename to...
10645 (xlogue_layout::s_stub_names): ...this, make static.
10646 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
10647 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
10648 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
10649 xlogue_layout::s_stub_names): Instantiate statics.
10650 (stub_managed_regs): Remove.
10651 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
10652 (disable_call_ms2sysv_xlogues): Rename to...
10653 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
10654 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
10656 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
10657 change after reload_completed.
10658 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
10660 (ix86_expand_prologue): Likewise.
10661 (ix86_expand_epilogue): Likewise.
10662 (ix86_expand_split_stack_prologue): Likewise.
10663 (ix86_compute_frame_layout): Remove frame parameter ...
10664 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
10665 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
10667 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
10668 (ix86_frame): Move from here ...
10669 * config/i386/i386.h (ix86_frame): ... to here.
10670 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
10671 complete ix86_frame data structure instead. Remove some_ld_name.
10673 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10675 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
10676 symbols that hold a DECL_VALUE_EXPR.
10678 2017-06-01 Martin Jambor <mjambor@suse.cz>
10680 PR tree-optimization/80898
10681 * tree-sra.c (process_subtree_disqualification): Removed.
10682 (disqualify_candidate): Do not acll
10683 process_subtree_disqualification.
10684 (subtree_mark_written_and_enqueue): New function.
10685 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
10686 RHS has been disqualified and re-queue LHS if necessary. Apart
10687 from that, ignore disqualified RHS.
10689 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10691 * config/s390/s390.c (s390_emit_epilogue): Disable early return
10692 address fetch for z10 or later.
10694 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10696 * config/arc/arc.md (tst_movb): Add guard when splitting.
10698 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10700 * config/arc/arc.c (arc_can_eliminate): Test against
10701 arc_frame_pointer_needed.
10703 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10705 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
10706 to prevent store reordering.
10707 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
10708 (type): Add block type.
10709 (stack_tie): Define special instruction to be used in
10712 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10714 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
10715 constraint. It is not valid for the pattern.
10716 (noncommutative_binary_comparison): Likewise.
10718 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10720 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
10723 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10725 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
10726 be used by the reg-alloc.
10728 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10730 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
10731 reg-alloc when having mul64 or mul32x16 instructions.
10732 (mulsidi3): Likewise.
10733 (umulsidi3): Likewise.
10734 (mulsi32x16): New pattern.
10735 (mulsi64): Likewise.
10736 (mulsidi64): Likewise.
10737 (umulsidi64): Likewise.
10738 (MUL32x16_REG): Define.
10739 (mul64_600): Use MUL32x16_REG.
10740 (mac64_600): Likewise.
10741 (umul64_600): Likewise.
10742 (umac64_600): Likewise.
10744 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10746 * config/arc/arc.md (mulsi3_700): Make it commutative.
10748 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
10750 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
10752 (*sign_extendsidi2_insn): Likewise for movstosw.
10754 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10756 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
10757 the type of the input discriminant value. Convert the
10758 discriminant value of signedness vary.
10760 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
10762 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
10763 Add to -Wall section.
10765 2017-06-01 Richard Biener <rguenther@suse.de>
10767 PR middle-end/66313
10768 * fold-const.c (fold_plusminus_mult_expr): If the factored
10769 factor may be zero use a wrapping type for the inner operation.
10770 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
10771 and handle moved defs.
10772 (process_assignment): Properly guard the unary op case. Return a
10773 tri-state indicating that moving the stmt before the call may allow
10774 to continue. Pass through to_move.
10775 (find_tail_calls): Handle moving unrelated defs before
10778 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10781 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
10782 splitter result in the canonical way.
10784 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
10786 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
10787 also for 32bit target. Update insn attributes.
10788 (zero-extendsidi2 splitter): Allow all registers for operand 1.
10790 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
10792 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
10793 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
10794 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
10795 (_mm_maskz_min_ss): New intrinsics.
10797 2017-05-31 Martin Liska <mliska@suse.cz>
10799 * tree-vect-loop.c (vect_create_epilog_for_reduction):
10800 Change comment style to one we normally use.
10801 (vectorizable_reduction): Likewise.
10802 (vectorizable_induction): Likewise.
10803 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10804 (vectorizable_call): Likewise.
10805 (vectorizable_simd_clone_call): Likewise.
10806 (vectorizable_conversion): Likewise.
10807 (vectorizable_assignment): Likewise.
10808 (vectorizable_shift): Likewise.
10809 (vectorizable_operation): Likewise.
10810 (vectorizable_store): Likewise.
10811 (vectorizable_load): Likewise.
10812 * tree-vectorizer.h: Likewise.
10814 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
10816 * passes.c (emergency_dump_function): New.
10817 * tree-pass.h (emergency_dump_function): Declare.
10818 * plugin.c (plugins_internal_error_function): Remove.
10819 * plugin.h (plugins_internal_error_function): Remove declaration.
10820 * toplev.c (internal_error_function): New static function. Use it...
10821 (general_init): ...here.
10823 2017-05-31 Graham Markall <graham.markall@embecosm.com>
10825 * config/arc/arc.c (arc_print_operand): Handle constant operands.
10826 (arc_rtx_costs): Add costs for new patterns.
10827 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
10828 * config/arc/predicates.md: Add _1_2_3_operand predicate.
10830 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10832 * tree-ssa-strlen.c (get_next_strinfo): New function.
10833 (get_stridx_plus_constant): Use it.
10834 (zero_length_string): Likewise.
10835 (adjust_related_strinfos): Likewise.
10836 (adjust_last_stmt): Likewise.
10838 2017-05-31 Richard Biener <rguenther@suse.de>
10841 * config/i386/i386.c (ix86_expand_builtin): Remove assert
10842 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
10844 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10846 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
10847 loop_vinfo argument and use of dependence distance vectors.
10848 Check instead whether the two references differ only in their
10849 initial value and assume that they have the same alignment if the
10850 difference is a multiple of the vector alignment.
10851 (vect_analyze_data_refs_alignment): Update call accordingly.
10853 2017-05-31 Martin Liska <mliska@suse.cz>
10856 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
10858 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10860 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
10861 (create_intersect_range_checks): Move from ...
10862 * tree-data-ref.c (create_intersect_range_checks_index)
10863 (create_intersect_range_checks): ... to here.
10864 (create_runtime_alias_checks): New function factored from ...
10865 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
10866 here. Call above function.
10867 * tree-data-ref.h (create_runtime_alias_checks): New function.
10869 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10871 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
10872 segment length for dr_b and compute it in wide_int.
10874 2017-05-31 Richard Biener <rguenther@suse.de>
10876 PR tree-optimization/80906
10877 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
10878 and pass through iv_map.
10879 (copy_bb_and_scalar_dependences): Adjust.
10880 (translate_pending_phi_nodes): Likewise.
10881 (copy_loop_close_phi_args): Handle code-generating IVs instead
10884 2017-05-30 David Malcolm <dmalcolm@redhat.com>
10886 * diagnostic-color.c (color_dict): Add "type-diff".
10887 (parse_gcc_colors): Update comment.
10888 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10889 -fdiagnostics-show-template-tree and -fno-elide-type.
10890 (GCC_COLORS): Add type-diff to example.
10892 (-fdiagnostics-show-template-tree): New.
10893 (-fno-elide-type): New.
10894 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
10895 the pp_format_decoder callback. Call any m_format_postprocessor's
10897 (pretty_printer::pretty_printer): Initialize
10898 m_format_postprocessor.
10899 (pretty_printer::~pretty_printer): Delete any
10900 m_format_postprocessor.
10901 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
10902 (class format_postprocessor): New class.
10903 (struct pretty_printer::format_decoder): Document the new parameters.
10904 (struct pretty_printer::m_format_postprocessor): New field.
10905 * tree-diagnostic.c (default_tree_printer): Update for new
10906 bool and const char ** params.
10907 * tree-diagnostic.h (default_tree_printer): Likewise.
10909 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
10911 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
10912 (lwa_operand): Delete rs6000_gen_cell_microcode test.
10913 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10914 rs6000_gen_cell_microcode code.
10915 (rs6000_final_prescan_insn): Delete.
10916 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
10917 "warn-cell-microcode" entries.
10918 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
10919 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
10920 throughout. Change cc_reg_not_micro_cr0_operand to
10921 cc_reg_not_cr0_operand throughout.
10922 (*extendhi<mode>2_noload): Delete.
10923 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
10924 (mwarn-cell-microcode): Delete.
10925 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
10926 -mgen-cell-microcode and -mwarn-cell-microcode.
10928 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
10931 * config/i386/constraints.md (Yd): New constraint.
10933 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
10934 and (?Yd, r) alternatives. Update insn attributes.
10935 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
10936 and (?*Yd, r) alternatives. Update insn attributes.
10937 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
10939 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
10941 * gimplify.c (gimplify_modify_expr): Don't create a
10942 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
10945 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10947 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
10949 2017-05-30 Richard Biener <rguenther@suse.de>
10951 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
10952 and reduc_def fields.
10953 (STMT_VINFO_REDUC_TYPE): New define.
10954 (STMT_VINFO_REDUC_DEF): Likewise.
10955 (vect_force_simple_reduction): Adjust prototype.
10956 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
10957 (vect_is_simple_reduction): Remove check_reduction argument.
10958 (vect_force_simple_reduction): Adjust and set
10959 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
10960 (vectorizable_reduction): Do not re-do reduction analysis
10961 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
10962 * tree-parloops.c (gather_scalar_reductions): Adjust.
10964 2017-05-30 Richard Biener <rguenther@suse.de>
10966 PR middle-end/80901
10967 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
10970 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
10972 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
10973 Introduce unknown_misalignment parameter and remove vf.
10974 (vect_peeling_hash_get_lowest_cost):
10975 Pass unknown_misalignment parameter.
10976 (vect_enhance_data_refs_alignment):
10977 Fix unsupportable data ref treatment.
10979 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10981 * tree-vect-data-refs.c (vect_get_data_access_cost):
10982 Workaround for SLP handling.
10983 (vect_enhance_data_refs_alignment):
10984 Compute costs for doing no peeling at all, compare to the best
10985 peeling costs so far and avoid peeling if cheaper.
10987 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10989 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
10990 Return peeling info and set costs to zero for unlimited cost
10992 (vect_enhance_data_refs_alignment): Also inspect all datarefs
10993 with unknown misalignment. Compute and costs for unknown
10994 misalignment, compare them to the costs for known misalignment
10995 and choose the cheapest for peeling.
10997 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10999 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11000 (vect_get_peeling_costs_all_drs): Create function.
11001 (vect_peeling_hash_get_lowest_cost):
11002 Use vect_get_peeling_costs_all_drs.
11003 (vect_peeling_supportable): Create function.
11004 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11006 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11008 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11009 DR_HAS_NEGATIVE_STEP.
11010 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11011 (vect_enhance_data_refs_alignment): Use.
11012 (vect_duplicate_ssa_name_ptr_info): Use.
11013 * tree-vectorizer.h (dr_misalignment): Use.
11014 (known_alignment_for_access_p): Use.
11016 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11019 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11021 (has_section_name): New function.
11023 2017-05-30 Martin Liska <mliska@suse.cz>
11026 * auto-profile.c (get_function_decl_from_block): Fix
11029 2017-05-30 Richard Biener <rguenther@suse.de>
11031 PR middle-end/80876
11032 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11034 2017-05-30 Martin Liska <mliska@suse.cz>
11036 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11037 * dumpfile.h (struct dump_file_info): Remove ctors.
11039 2017-05-30 Martin Liska <mliska@suse.cz>
11041 * predict.def: Fix GNU coding style.
11043 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11045 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11046 Mark 'to' argument with ATTRIBUTE_UNUSED.
11048 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11050 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11051 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11052 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11055 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11057 * doc/install.texi (Options specification): Restore entry of
11058 --enable-sjlj-exceptions.
11060 2017-05-27 Michael Eager <eager@eagercon.com>
11063 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11065 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11067 * config/microblaze/microblaze.h
11068 (FIXED_REGISTERS): Update in macro.
11069 (CALL_USED_REGISTERS): Update in macro.
11071 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11073 * doc/install.texi: Add links to macOS binary distributions.
11075 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11079 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11081 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11083 2017-05-26 Martin Liska <mliska@suse.cz>
11085 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11087 2017-05-26 Martin Liska <mliska@suse.cz>
11089 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11090 always leading ';; '.
11091 (dump_bb_info): Likewise.
11092 (brief_dump_cfg): Likewise.
11093 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11094 * dumpfile.c: Remove usage of TDF_VERBOSE.
11095 * dumpfile.h (enum dump_kind): Likewise.
11096 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11097 * print-tree.c (debug_verbose): Remove.
11098 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11099 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11100 * tree-diagnostic.c (default_tree_printer): Replace
11101 TDF_DIAGNOSTIC with TDF_SLIM.
11103 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11105 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11106 in parameter loop, rather than loop_vinfo.
11107 (create_intersect_range_checks): Ditto.
11108 (vect_create_cond_for_alias_checks): Update call to above functions.
11110 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11112 PR tree-optimization/80815
11113 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11114 for merging runtime alias checks. Handle negative DR_STEPs.
11116 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11118 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11120 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11121 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11122 out code pruning runtime alias checks.
11123 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11124 factored out from above.
11125 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11127 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11129 (prune_runtime_alias_test_list): New decalaration.
11131 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11133 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11134 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11135 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11136 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11137 (operator==, comp_dr_with_seg_len_pair): Ditto.
11138 (vect_prune_runtime_alias_test_list): Ditto.
11140 2017-05-26 Martin Liska <mliska@suse.cz>
11143 * params.def: Bound partial-inlining-entry-probability param.
11145 2017-05-26 Marek Polacek <polacek@redhat.com>
11148 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11151 2017-05-26 Richard Biener <rguenther@suse.de>
11153 PR tree-optimization/80842
11154 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11157 2017-05-26 Richard Biener <rguenther@suse.de>
11159 PR tree-optimization/80844
11160 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11162 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11164 * doc/md.texi (Machine Constraints): Update x86 family
11165 machine constraints section to match 'config/i386/constraints.md'.
11167 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11169 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11171 2017-05-25 Nathan Sidwell <nathan@acm.org>
11173 * doc/invoke.texi (--enable-languages): Update documentation.
11175 2017-05-25 Martin Liska <mliska@suse.cz>
11177 * dumpfile.c: Add TDF_FOLDING.
11178 * dumpfile.h (enum dump_kind): Likewise.
11179 * genmatch.c (dt_simplify::gen_1): Use it.
11181 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11183 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11185 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11187 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11188 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11190 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11192 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11193 * match.pd (X == C): Rewrite it here.
11194 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11195 with_certain_nonzero_bits2): New predicates.
11196 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11198 2017-05-24 Nathan Sidwell <nathan@acm.org>
11200 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11203 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11206 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11208 * config/powerpcspe: New port. Files are copied from the rs6000
11209 port, with "rs6000" in filenames replaced by "powerpcspe".
11211 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11213 PR rtl-optimization/80754
11214 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11216 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11218 * config/sparc/sparc.md (length): Return the correct value for -mflat
11219 sibcalls to match output_sibcall.
11221 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11225 * config/rs6000/rs6000.c (struct machine_function): Add new field
11227 (rs6000_get_separate_components): Init that field, use it.
11228 (rs6000_components_for_bb): Use the field.
11230 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11232 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11234 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11236 PR middle-end/80823
11237 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11239 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11242 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11243 address against address_operand predicate.
11244 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11246 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11248 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11249 parameters passed indirectly.
11251 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11253 * config/i386/i386.md (*movdi_internal): Remove SSE4
11254 alternative 18 (?r, *v). Update insn attributes.
11255 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11256 Update insn attributes.
11257 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11258 Update insn attributes.
11259 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11260 alternative 1 (r, v). Remove isa attribute.
11261 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11262 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11263 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11265 2017-05-23 Tom de Vries <tom@codesourcery.com>
11267 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11270 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11272 * cgraphunit.c (symbol_table::process_new_functions): Update.
11273 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11274 (inline_generate_summary): Rename to ...
11275 (ipa_fn_summary_generate): ... this one.
11276 (inline_read_summary): Rename to ...
11277 (ipa_fn_summary_read): ... this one.
11278 (inline_write_summary): Rename to ...
11279 (ipa_fn_summary_write): ... this one.
11280 (inline_free_summary): Rename to ...
11281 (ipa_free_fn_summary): ... this one.
11282 (pass_data_local_fn_summary, pass_local_fn_summary,
11283 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11284 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11285 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11286 make_pass_ipa_fn_summary): New.
11287 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11288 inline_write_summary, inline_free_summary): Remove.
11289 (ipa_free_fn_summary) : New.
11290 * ipa-inline.c (ipa_inline): Update.
11291 (pass_ipa_inline): Do not generate summaries.
11292 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11294 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11295 and add pass_ipa_fn_summary.
11296 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11298 (make_pass_inline_parameters): Remove.
11300 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11302 * omp-low.c (struct omp_context): Remove "default_kind" member.
11305 * omp-offload.c (execute_oacc_device_lower): Remove the
11306 parallelism dimensions function attributes for unparallelized
11307 OpenACC kernels constructs.
11309 2017-05-23 Martin Liska <mliska@suse.cz>
11311 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11313 (cgraph_edge::make_speculative): Likewise.
11314 (cgraph_edge::resolve_speculation): Likewise.
11315 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11316 (cgraph_node::dump): Likewise.
11317 * cgraph.h: Likewise.
11318 * cgraphunit.c (analyze_functions): Likewise.
11319 (symbol_table::compile): Likewise.
11320 * ipa-cp.c (print_all_lattices): Likewise.
11321 (determine_versionability): Likewise.
11322 (initialize_node_lattices): Likewise.
11323 (ipcp_verify_propagated_values): Likewise.
11324 (estimate_local_effects): Likewise.
11325 (update_profiling_info): Likewise.
11326 (create_specialized_node): Likewise.
11327 (perhaps_add_new_callers): Likewise.
11328 (decide_about_value): Likewise.
11329 (decide_whether_version_node): Likewise.
11330 (identify_dead_nodes): Likewise.
11331 (ipcp_store_bits_results): Likewise.
11332 * ipa-devirt.c (dump_targets): Likewise.
11333 (ipa_devirt): Likewise.
11334 * ipa-icf.c (sem_item::dump): Likewise.
11335 (sem_function::equals): Likewise.
11336 (sem_variable::equals): Likewise.
11337 (sem_item_optimizer::read_section): Likewise.
11338 (sem_item_optimizer::execute): Likewise.
11339 (congruence_class::dump): Likewise.
11340 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11341 (dump_inline_summary): Likewise.
11342 (estimate_node_size_and_time): Likewise.
11343 (inline_analyze_function): Likewise.
11344 * ipa-inline-transform.c (inline_call): Likewise.
11345 * ipa-inline.c (report_inline_failed_reason): Likewise.
11346 (want_early_inline_function_p): Likewise.
11347 (edge_badness): Likewise.
11348 (update_edge_key): Likewise.
11349 (inline_small_functions): Likewise.
11350 * ipa-profile.c (ipa_profile): Likewise.
11351 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11352 (ipa_make_edge_direct_to_target): Likewise.
11353 (remove_described_reference): Likewise.
11354 (ipa_impossible_devirt_target): Likewise.
11355 (propagate_controlled_uses): Likewise.
11356 (ipa_print_node_params): Likewise.
11357 (ipcp_transform_function): Likewise.
11358 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11359 (propagate_pure_const): Likewise.
11360 * ipa-reference.c (generate_summary): Likewise.
11361 (read_write_all_from_decl): Likewise.
11362 (propagate): Likewise.
11363 (ipa_reference_read_optimization_summary): Likewise.
11364 * ipa-utils.c (ipa_merge_profiles): Likewise.
11365 * ipa.c (walk_polymorphic_call_targets): Likewise.
11366 (symbol_table::remove_unreachable_nodes): Likewise.
11367 (ipa_single_use): Likewise.
11368 * passes.c (execute_todo): Likewise.
11369 * predict.c (drop_profile): Likewise.
11370 * symtab.c (symtab_node::get_dump_name): New function.
11371 (symtab_node::dump_name): Likewise.
11372 (symtab_node::dump_asm_name): Likewise.
11373 (symtab_node::dump_references): Likewise.
11374 (symtab_node::dump_referring): Likewise.
11375 (symtab_node::dump_base): Likewise.
11376 (symtab_node::debug_symtab): Likewise.
11377 * tree-sra.c (convert_callers_for_node): Likewise.
11378 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11379 * value-prof.c (init_node_map): Likewise.
11381 2017-05-23 Martin Liska <mliska@suse.cz>
11383 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11384 and symtab_node::debug_symtab to symbol_table::debug.
11385 * cgraphunit.c (analyze_functions): Use the renamed function.
11386 (symbol_table::compile): Likewise.
11387 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11388 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11389 * passes.c (execute_todo): Likewise.
11390 * symtab.c (symbol_table::dump): New function.
11391 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11393 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11395 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11396 that nonconst implies exec.
11398 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11400 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11401 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11402 (inline_edge_summary_vec): Turn into ...
11403 (ipa_call_summaries): ... this one.
11404 (redirect_to_unreachable, edge_set_predicate,
11405 evaluate_properties_for_edge, inline_summary_alloc,
11406 reset_ipa_call_summary, reset_inline_summary,
11407 inline_summary_t::duplicate): Update.
11408 (inline_edge_duplication_hook): Turn to ...
11409 (ipa_call_summary_t::duplicate): ... this one.
11410 (inline_edge_removal_hook): Turn to ...
11411 (ipa_call_summary_t::remove): ... this one.
11412 (dump_inline_edge_summary): Turn to ...
11413 (dump_ipa_call_summary): ... this one.
11414 (estimate_function_body_sizes): Update.
11415 (inline_update_callee_summaries): Update.
11416 (remap_edge_change_prob): Update.
11417 (remap_edge_summaries): Update.
11418 (inline_merge_summary): Update.
11419 (do_estimate_edge_time): Update.
11420 (inline_generate_summary): Update.
11421 (inline_read_section): Update.
11422 (inline_read_summary): Update.
11423 (inline_free_summary): Update.
11424 * ipa-inline.c (can_inline_edge_p): Update.
11425 (compute_inlined_call_time): Update.
11426 (want_inline_small_function_p): Update.
11427 (edge_badness): Update.
11428 (early_inliner): Update.
11429 * ipa-inline.h (inline_edge_summary): Turn to ...
11430 (ipa_call_summary): ... this one.
11431 (ipa_call_summary_t): New class.
11432 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11433 (ipa_call_summaries): New.
11434 (inline_edge_summary): Remove.
11435 (estimate_edge_growth): Update.
11436 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11437 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11438 * ipa-split.c (execute_split_functions): Update.
11439 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11441 2017-05-23 Tom de Vries <tom@codesourcery.com>
11443 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11444 attributes): Document rdrand effective target.
11446 2017-05-23 Tom de Vries <tom@codesourcery.com>
11448 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11449 attributes): Sort alphabetically.
11451 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11453 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11455 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11458 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11459 V2DF/V2DI splat into two separate patterns, one that handles
11460 registers, and the other that only handles memory. Drop support
11461 for splatting from a GPR on ISA 2.07 and then splitting the
11462 splat into direct move and splat.
11463 (vsx_splat_<mode>_reg): Likewise.
11464 (vsx_splat_<mode>_mem): Likewise.
11466 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11468 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11470 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11472 PR middle-end/80809
11473 * omp-low.c (finish_taskreg_remap): New function.
11474 (finish_taskreg_scan): If unit size of ctx->record_type
11475 is non-constant, unshare the size expression and replace
11476 decls in it with possible outer var refs.
11478 PR middle-end/80809
11479 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11480 GOVD_SHARED rather than GOVD_PRIVATE with it.
11481 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11482 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11484 PR middle-end/80853
11485 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11486 as last argument to build_outer_var_ref for pointer bases of array
11487 section reductions.
11489 2017-05-19 Martin Sebor <msebor@redhat.com>
11491 * print-tree.c (print_node): Print DECL_READ_P flag.
11493 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11495 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11496 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11497 * cgraph.c: Likewise.
11498 * cgraphunit.c: Likewise.
11499 * gengtype.c: Likewise.
11500 * ipa-cp.c: Likewise.
11501 * ipa-devirt.c: Likewise.
11502 * ipa-icf.c: Likewise.
11503 * ipa-predicate.c: Likewise.
11504 * ipa-profile.c: Likewise.
11505 * ipa-prop.c: Likewise.
11506 * ipa-split.c: Likewise.
11508 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11509 edge_predicate_pool, dump_inline_hints,
11510 inline_summary::account_size_time, redirect_to_unreachable,
11511 edge_set_predicate, set_hint_predicate,
11512 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11513 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11514 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11515 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11516 ipa_call_summary_t::remove, initialize_growth_caches,
11517 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11518 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11519 mark_modified, unmodified_parm_1, unmodified_parm,
11520 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11521 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11522 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11523 will_be_nonconstant_predicate, record_modified_bb_info,
11524 get_minimal_bb, record_modified, param_change_prob,
11525 phi_result_unknown_predicate, predicate_for_phi_result,
11526 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11527 estimate_function_body_sizes, compute_inline_parameters,
11528 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11529 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11530 inline_update_callee_summaries, remap_edge_change_prob,
11531 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11532 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11533 inline_analyze_function, inline_summary_t::insert,
11534 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11535 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11536 inline_free_summary): Move to ipa-fnsummary.h
11537 (predicate_t): Remove.
11538 * ipa-fnsummary.c: New file.
11539 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11540 (enum inline_hints_vals, inline_hints, agg_position_info,
11541 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11542 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11543 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11544 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11545 inline_read_summary, inline_write_summary, inline_free_summary,
11546 inline_analyze_function, initialize_inline_failed,
11547 inline_merge_summary, inline_update_overall_summary,
11548 compute_inline_parameters): Move to ipa-fnsummary.h
11549 * ipa-fnsummary.h: New file.
11550 * ipa-inline-transform.h: Include ipa-inline.h.
11551 * ipa-inline.c: LIkewise.
11553 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11555 * ipa-inline.c (edge_badness): Use inlined_time instead of
11556 inline_summaries->get.
11558 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11560 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11562 2017-05-22 Nathan Sidwell <nathan@acm.org>
11564 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11565 (fdump-lang): Document 'raw' option.
11566 * dumpfile.h (TDI_tu): Delete.
11567 * dumpfile.c (dump_files): Remove translation-unit.
11568 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11570 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11572 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11573 command option from $(AWK) call.
11574 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11575 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11576 [FORMAT]: Remove handling of variable.
11577 * config/avr/t-multilib: Regenerate.
11579 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11581 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11583 (dump_inline_summary): Do not print self_time.
11584 (estimate_function_body_sizes): Do not set self_time.
11585 (compute_inline_parameters): Likewise.
11586 (inline_read_section, inline_write_summary): Do not stream self_time.
11587 * ipa-inline.h (inline_summary): Drop self_time.
11589 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11591 * ipa-inline-analysis.c (account_size_time): Rename to ...
11592 (inline_summary::account_size_time): ... this one.
11593 (reset_ipa_call_summary): Turn to ...
11594 (ipa_call_summary::reset): ... this one.
11595 (reset_inline_summary): Turn to ...
11596 (inline_summary::reset): ... this one.
11597 (inline_summary_t::remove): Update.
11598 (inline_summary_t::duplicate): Update.
11599 (ipa_call_summary_t::remove): Update.
11600 (dump_inline_summary): Update.
11601 (estimate_function_body_sizes): Update.
11602 (compute_inline_parameters): Update.
11603 (estimate_node_size_and_time): Update.
11604 (inline_merge_summary): Update.
11605 (inline_update_overall_summary): Update.
11606 (inline_read_section): Update.
11607 (inline_write_summary): Update.
11608 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
11609 add account_size_time and reset member functions.
11610 (ipa_call_summary): Add reset function.
11611 * ipa-predicate.h (predicate::operator &): Constify.
11613 2017-05-22 Richard Biener <rguenther@suse.de>
11615 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
11617 2017-05-19 Jason Merrill <jason@redhat.com>
11619 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
11621 2017-05-19 Marek Polacek <polacek@redhat.com>
11624 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
11625 TYPE_OVERFLOW_WRAPS checks.
11627 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
11629 * tree-core.h (enum omp_clause_default_kind): Add
11630 "OMP_CLAUSE_DEFAULT_PRESENT".
11631 * tree-pretty-print.c (dump_omp_clause): Handle it.
11632 * gimplify.c (enum gimplify_omp_var_data): Add
11633 "GOVD_MAP_FORCE_PRESENT".
11634 (gimplify_adjust_omp_clauses_1): Map it to
11635 "GOMP_MAP_FORCE_PRESENT".
11636 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
11638 * gimplify.c (oacc_default_clause): Clarify.
11640 2017-05-19 Nathan Sidwell <nathan@acm.org>
11642 LANG_HOOK_REGISTER_DUMPS
11643 * toplev.c (general_init): Call register dump lang hook.
11644 * doc/invoke.texi: Document -fdump-lang option family.
11645 * dumpfile.c (dump_files): Remove class dump here.
11646 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
11647 * dumpfile.h (tree_dump_index): Remove TDI_class.
11648 * langhooks-def.h (lhd_register_dumps): Declare.
11649 (LANG_HOOKS_REGISTER_DUMPS): Define.
11650 (LANG_HOOKS_INITIALIZER): Add it.
11651 * langhooks.c (lhd_register_dumps): Define.
11652 * langhooks.h (struct lang_hooks): Add register_dumps.
11654 2017-05-19 Nathan Sidwell <nathan@acm.org>
11656 * context.h (context::set_passes): New.
11657 * context.c (context::context): Do not create pass manager.
11658 * toplev.c (general_init): Create pass manager here.
11660 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
11662 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
11663 use this splitter if two add or or instructions would also work for
11664 the constant we want to generate.
11666 2017-05-19 Richard Biener <rguenther@suse.de>
11669 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
11670 predicate evaluation.
11672 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11674 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
11676 * ipa-inline.c (want_inline_small_function_p): Do not cast to
11679 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11681 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11682 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11683 (inline_edge_summary_vec): Turn into ...
11684 (ipa_call_summaries): ... this one.
11685 (redirect_to_unreachable, edge_set_predicate,
11686 evaluate_properties_for_edge, inline_summary_alloc,
11687 reset_ipa_call_summary, reset_inline_summary,
11688 inline_summary_t::duplicate): Update.
11689 (inline_edge_duplication_hook): Turn to ...
11690 (ipa_call_summary_t::duplicate): ... this one.
11691 (inline_edge_removal_hook): Turn to ...
11692 (ipa_call_summary_t::remove): ... this one.
11693 (dump_inline_edge_summary): Turn to ...
11694 (dump_ipa_call_summary): ... this one.
11695 (estimate_function_body_sizes): Update.
11696 (inline_update_callee_summaries): Update.
11697 (remap_edge_change_prob): Update.
11698 (remap_edge_summaries): Update.
11699 (inline_merge_summary): Update.
11700 (do_estimate_edge_time): Update.
11701 (inline_generate_summary): Update.
11702 (inline_read_section): Update.
11703 (inline_read_summary): Update.
11704 (inline_free_summary): Update.
11705 * ipa-inline.c (can_inline_edge_p): Update.
11706 (compute_inlined_call_time): Update.
11707 (want_inline_small_function_p): Update.
11708 (edge_badness): Update.
11709 (early_inliner): Update.
11710 * ipa-inline.h (inline_edge_summary): Turn to ...
11711 (ipa_call_summary): ... this one.
11712 (ipa_call_summary_t): New class.
11713 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11714 (ipa_call_summaries): New.
11715 (inline_edge_summary): Remove.
11716 (estimate_edge_growth): Update.
11717 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11718 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11719 * ipa-split.c (execute_split_functions): Update.
11720 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11722 2017-05-19 Richard Biener <rguenther@suse.de>
11724 PR middle-end/80764
11725 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
11727 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
11729 * config/rs6000/rs6000.c (struct machine_function): Add field
11730 fpr_is_wrapped_separately.
11731 (rs6000_get_separate_components): Use 64 components. Handle the
11732 new FPR components.
11733 (rs6000_components_for_bb): Handle the FPR components.
11734 (rs6000_emit_prologue_components): Handle the FPR components.
11735 (rs6000_emit_epilogue_components): Handle the FPR components.
11736 (rs6000_set_handled_components): Handle the FPR components.
11737 (rs6000_emit_prologue): Don't output prologue code for those FPRs
11738 that are already separately shrink-wrapped.
11739 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
11740 that are already separately shrink-wrapped.
11742 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11745 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
11748 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
11749 (define_peephole2 for Altivec d-form load): Add peepholes to catch
11750 cases where the register allocator uses a move and an offsettable
11751 memory operation to/from a FPR register on ISA 2.06/2.07.
11752 (define_peephole2 for Altivec d-form store): Likewise.
11754 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
11757 * config/i386/mmx.md (*mov<mode>_internal): Enable
11758 alternatives 11, 12, 13 and 14 also for 32bit targets.
11759 Remove alternatives 15, 16, 17 and 18.
11760 * config/i386/sse.md (vec_concatv2di): Change
11761 alternative (!x, *y) to (x, ?!*Yn).
11763 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
11765 * dumpfile.h (enum dump_kind): Remove stray comma.
11767 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11769 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
11770 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
11771 predicate::num_conditions
11772 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
11773 (CHANGED): turn into predicate::changed.
11774 (agg_position_info): Move to ipa-predicate.h
11775 (add_condition, predicate::add_clause, predicate::operator &=,
11776 predicate::or_with, predicate::evaluate, predicate::probability,
11777 dump_condition, dump_clause, predicate::dump,
11778 predicate::remap_after_duplication, predicate::remap_after_inlining,
11779 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
11780 (evaluate_conditions_for_known_args): Update.
11781 (set_cond_stmt_execution_predicate): Update.
11782 * ipa-inline.h: Include ipa-predicate.h
11783 (condition, inline_param_summary, conditions, agg_position_info,
11784 predicate): Move to ipa-predicate.h
11785 * ipa-predicate.c: New file.
11786 * ipa-predicate.h: New file.
11788 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
11790 * final.c (leaf_function_p): Check we are not in a sequence.
11792 2017-05-18 Martin Liska <mliska@suse.cz>
11794 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
11795 * dumpfile.c (dump_register): Use new enum dump_kind.
11796 (get_dump_file_name): Likewise.
11797 (dump_enable_all): Likewise.
11798 (dump_switch_p_1): Likewise.
11799 (enable_rtl_dump_file): Remove usage of TDF_RTL.
11800 * dumpfile.h (enum dump_kind): New enum type.
11801 (struct dump_file_info): Create constructor and
11802 format fields and comments.
11803 * passes.c (pass_manager::register_one_dump_file):
11805 * statistics.c (statistics_early_init): Likewise.
11806 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
11807 TDF_TREE with TDF_SLIM.
11808 (gather_memory_references_ref): Likewise.
11810 2017-05-18 Martin Liska <mliska@suse.cz>
11812 * vec.h (struct vnull): Use it.
11814 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11816 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
11817 (true_predicate, false_predicate, true_predicate_p,
11818 false_predicate_p): Remove.
11819 (single_cond_predicate, not_inlined_predicate): Turn to member function
11821 (add_condition): Update.
11822 (add_clause): Turn to...
11823 (predicate::add_clause): ... this one; update; allow passing NULL
11825 (and_predicates): Turn to ...
11826 (predicate::operator &=): ... this one.
11827 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
11828 (or_predicates): Turn to ...
11829 (predicate::or_with): ... this one.
11830 (evaluate_predicate): Turn to ...
11831 (predicate::evaluate): ... this one.
11832 (predicate_probability): Turn to ...
11833 (predicate::probability): ... this one.
11834 (dump_condition): Update.
11835 (dump_predicate): Turn to ...
11836 (predicate::dump): ... this one.
11837 (account_size_time): Update.
11838 (edge_set_predicate): Update.
11839 (set_hint_predicate): UPdate.
11840 (evaluate_conditions_for_known_args): Update.
11841 (evaluate_properties_for_edge): Update.
11842 (remap_predicate_after_duplication): Turn to...
11843 (predicate::remap_after_duplication): ... this one.
11844 (remap_hint_predicate_after_duplication): Update.
11845 (inline_summary_t::duplicate): UPdate.
11846 (dump_inline_edge_summary): Update.
11847 (dump_inline_summary): Update.
11848 (set_cond_stmt_execution_predicate): Update.
11849 (set_switch_stmt_execution_predicate): Update.
11850 (compute_bb_predicates): Update.
11851 (will_be_nonconstant_expr_predicate): Update.
11852 (will_be_nonconstant_predicate): Update.
11853 (phi_result_unknown_predicate): Update.
11854 (predicate_for_phi_result): Update.
11855 (array_index_predicate): Update.
11856 (estimate_function_body_sizes): Update.
11857 (estimate_node_size_and_time): Update.
11858 (estimate_ipcp_clone_size_and_time): Update.
11859 (remap_predicate): Rename to ...
11860 (predicate::remap_after_inlining): ... this one.
11861 (remap_hint_predicate): Update.
11862 (inline_merge_summary): Update.
11863 (inline_update_overall_summary): Update.
11864 (estimate_size_after_inlining): Update.
11865 (read_predicate): Rename to ...
11866 (predicate::stream_in): ... this one.
11867 (read_inline_edge_summary): Update.
11868 (write_predicate): Rename to ...
11869 (predicate::stream_out): ... this one.
11870 (write_inline_edge_summary): Update.
11871 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
11872 (clause_t): Turn to uint32_t
11873 (predicate): Turn to class; implement constructor and operators
11875 (size_time_entry): Update.
11876 (inline_summary): Update.
11877 (inline_edge_summary): Update.
11879 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
11881 * fold-const.c (fold_binary_loc): Move transformation...
11882 * match.pd (C - X CMP X): ... here.
11884 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
11886 * config/sparc/sparc.c (sparc_option_override): Set function
11887 alignment for -mcpu=niagara7 to 64 to match the I$ line.
11888 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
11890 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
11892 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
11894 2017-05-18 Marek Polacek <polacek@redhat.com>
11897 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
11898 (pass_ubsan::execute): Call gimple_assign_single_p instead of
11899 gimple_assign_load_p.
11901 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
11903 PR middle-end/80692
11904 * real.c (do_compare): Give decimal_do_compare preference over
11905 comparing just the signs.
11907 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
11909 * doc/md.texi (Canonicalization of Instructions): Describe the
11910 canonical form of instructions that inherently set a condition
11913 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
11915 PR middle-end/80775
11916 * tree-cfg.c: Move deletion of unreachable case statements to after
11917 the merging of consecutive case labels.
11919 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
11921 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
11922 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
11923 restoring of callee-saved registers.
11925 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
11927 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
11928 * config/visium/visium.c (single_set_and_flags): Likewise.
11929 * config/visium/visium.md (Substitutions): Likewise.
11931 2017-05-17 Martin Liska <mliska@suse.cz>
11933 * cfg.c: Introduce dump_flags_t type and
11934 use it instead of int type.
11936 * cfghooks.c: Likewise.
11937 * cfghooks.h (struct cfg_hooks): Likewise.
11938 * cfgrtl.c: Likewise.
11939 * cfgrtl.h: Likewise.
11940 * cgraph.c (cgraph_node::get_body): Likewise.
11941 * coretypes.h: Likewise.
11942 * domwalk.c: Likewise.
11943 * domwalk.h: Likewise.
11944 * dumpfile.c (struct dump_option_value_info): Likewise.
11945 (dump_enable_all): Likewise.
11946 (dump_switch_p_1): Likewise.
11947 (opt_info_switch_p): Likewise.
11948 * dumpfile.h (enum tree_dump_index): Likewise.
11949 (struct dump_file_info): Likewise.
11950 * genemit.c: Likewise.
11951 * generic-match-head.c: Likewise.
11952 * gengtype.c (open_base_files): Likewise.
11953 * gimple-pretty-print.c: Likewise.
11954 * gimple-pretty-print.h: Likewise.
11955 * graph.c (print_graph_cfg): Likewise.
11956 * graphite-scop-detection.c (dot_all_sese): Likewise.
11957 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
11958 * loop-unroll.c (report_unroll): Likewise.
11959 * passes.c (pass_manager::register_one_dump_file): Likewise.
11960 * print-tree.c: Likewise.
11961 * statistics.c: Likewise.
11962 * tree-cfg.c: Likewise.
11963 * tree-cfg.h: Likewise.
11964 * tree-dfa.c: Likewise.
11965 * tree-dfa.h: Likewise.
11966 * tree-dump.c (dump_function): Likewise.
11967 * tree-dump.h (struct dump_info): Likewise.
11968 * tree-pretty-print.c: Likewise.
11969 * tree-pretty-print.h: Likewise.
11970 * tree-ssa-live.c: Likewise.
11971 * tree-ssa-live.h: Likewise.
11972 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
11973 * tree-vect-loop.c: Likewise.
11974 * tree-vect-slp.c: Likewise.
11976 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
11977 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11979 PR tree-optimization/80457
11980 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
11981 of all arguments to a statement as scalar_to_vec operations.
11982 (vectorizable_call): Adjust call to vect_model_simple_cost for
11984 (vectorizable_conversion): Likewise.
11985 (vectorizable_assignment): Likewise.
11986 (vectorizable_shift): Likewise.
11987 (vectorizable_operation): Likewise.
11988 (vectorizable_comparison): Likewise.
11989 (vect_is_simple_cond): Record the def types for operands.
11990 (vectorizable_condition): Likewise, call vect_model_simple_cost.
11991 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
11992 for statement argument count.
11994 2017-05-16 Carl Love <cel@us.ibm.com>
11996 * config/rs6000/rs6000-c: Add support for built-in functions
11997 vector unsigned long long vec_bperm (vector unsigned long long,
11998 vector unsigned char)
11999 vector signed long long vec_mule (vector signed int,
12001 vector unsigned long long vec_mule (vector unsigned int,
12002 vector unsigned int)
12003 vector signed long long vec_mulo (vector signed int,
12005 vector unsigned long long vec_mulo (vector unsigned int,
12006 vector unsigned int)
12007 vector signed char vec_sldw (vector signed char,
12008 vector signed char,
12010 vector unsigned char vec_sldw (vector unsigned char,
12011 vector unsigned char,
12013 vector signed short vec_sldw (vector signed short,
12014 vector signed short,
12016 vector unsigned short vec_sldw (vector unsigned short,
12017 vector unsigned short,
12019 vector signed int vec_sldw (vector signed int,
12022 vector unsigned int vec_sldw (vector unsigned int,
12023 vector unsigned int,
12025 vector signed long long vec_sldw (vector signed long long,
12026 vector signed long long,
12028 vector unsigned long long vec_sldw (vector unsigned long long,
12029 vector unsigned long long,
12031 * config/rs6000/rs6000-c: Add support for built-in functions
12032 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12033 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12034 * doc/extend.texi: Update the built-in documentation for the
12035 new built-in functions.
12037 2017-05-16 Marek Polacek <polacek@redhat.com>
12041 * tree.c (save_expr): Don't fold the expression.
12043 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12045 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12046 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12047 and (?*y,m). Update insn attributes.
12049 2017-05-16 Martin Liska <mliska@suse.cz>
12051 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12052 flags argument of print_gimple_stmt, print_gimple_expr,
12053 print_generic_stmt and print_generic_expr.
12054 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12055 * coretypes.h: Likewise.
12056 * except.c (dump_eh_tree): Likewise.
12057 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12058 * gimple-pretty-print.h: Likewise.
12059 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12060 (backprop::push_to_worklist): Likewise.
12061 (backprop::pop_from_worklist): Likewise.
12062 (backprop::process_use): Likewise.
12063 (backprop::intersect_uses): Likewise.
12064 (note_replacement): Likewise.
12065 * gimple-ssa-store-merging.c
12066 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12067 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12068 (pass_store_merging::execute): Likewise.
12069 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12070 (ssa_base_cand_dump_callback): Likewise.
12071 (dump_incr_vec): Likewise.
12072 (replace_refs): Likewise.
12073 (replace_mult_candidate): Likewise.
12074 (create_add_on_incoming_edge): Likewise.
12075 (create_phi_basis): Likewise.
12076 (insert_initializers): Likewise.
12077 (all_phi_incrs_profitable): Likewise.
12078 (introduce_cast_before_cand): Likewise.
12079 (replace_one_candidate): Likewise.
12080 * gimplify.c (gimplify_expr): Likewise.
12081 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12082 (set_rename): Likewise.
12083 (rename_uses): Likewise.
12084 (copy_loop_phi_nodes): Likewise.
12085 (add_close_phis_to_merge_points): Likewise.
12086 (copy_loop_close_phi_args): Likewise.
12087 (copy_cond_phi_args): Likewise.
12088 (graphite_copy_stmts_from_block): Likewise.
12089 (translate_pending_phi_nodes): Likewise.
12090 * graphite-poly.c (print_pdr): Likewise.
12091 (dump_gbb_cases): Likewise.
12092 (dump_gbb_conditions): Likewise.
12093 (print_scop_params): Likewise.
12094 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12095 (build_cross_bb_scalars_use): Likewise.
12096 (gather_bbs::before_dom_children): Likewise.
12097 * hsa-dump.c (dump_hsa_immed): Likewise.
12098 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12099 (get_replacement_map): Likewise.
12100 * ipa-inline-analysis.c (dump_condition): Likewise.
12101 (estimate_function_body_sizes): Likewise.
12102 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12103 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12104 * ipa-prop.c (ipa_dump_param): Likewise.
12105 (ipa_print_node_jump_functions_for_edge): Likewise.
12106 (ipa_modify_call_arguments): Likewise.
12107 (ipa_modify_expr): Likewise.
12108 (ipa_dump_param_adjustments): Likewise.
12109 (ipa_dump_agg_replacement_values): Likewise.
12110 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12111 * ipa-pure-const.c (check_stmt): Likewise.
12112 (pass_nothrow::execute): Likewise.
12113 * ipa-split.c (execute_split_functions): Likewise.
12114 * omp-offload.c (dump_oacc_loop_part): Likewise.
12115 (dump_oacc_loop): Likewise.
12116 * trans-mem.c (tm_log_emit): Likewise.
12117 (tm_memopt_accumulate_memops): Likewise.
12118 (dump_tm_memopt_set): Likewise.
12119 (dump_tm_memopt_transform): Likewise.
12120 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12121 (print_loop): Likewise.
12122 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12123 (chkp_gather_checks_info): Likewise.
12124 (chkp_get_check_result): Likewise.
12125 (chkp_remove_check_if_pass): Likewise.
12126 (chkp_use_outer_bounds_if_possible): Likewise.
12127 (chkp_reduce_bounds_lifetime): Likewise.
12128 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12129 (chkp_mark_completed_bounds): Likewise.
12130 (chkp_register_incomplete_bounds): Likewise.
12131 (chkp_mark_invalid_bounds): Likewise.
12132 (chkp_maybe_copy_and_register_bounds): Likewise.
12133 (chkp_build_returned_bound): Likewise.
12134 (chkp_get_bound_for_parm): Likewise.
12135 (chkp_build_bndldx): Likewise.
12136 (chkp_get_bounds_by_definition): Likewise.
12137 (chkp_generate_extern_var_bounds): Likewise.
12138 (chkp_get_bounds_for_decl_addr): Likewise.
12139 * tree-chrec.c (chrec_apply): Likewise.
12140 * tree-data-ref.c (dump_data_reference): Likewise.
12141 (dump_subscript): Likewise.
12142 (dump_data_dependence_relation): Likewise.
12143 (analyze_overlapping_iterations): Likewise.
12144 * tree-inline.c (expand_call_inline): Likewise.
12145 (tree_function_versioning): Likewise.
12146 * tree-into-ssa.c (dump_defs_stack): Likewise.
12147 (dump_currdefs): Likewise.
12148 (dump_names_replaced_by): Likewise.
12149 (dump_update_ssa): Likewise.
12150 (update_ssa): Likewise.
12151 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12152 * tree-parloops.c (build_new_reduction): Likewise.
12153 (try_create_reduction_list): Likewise.
12154 (ref_conflicts_with_region): Likewise.
12155 (oacc_entry_exit_ok_1): Likewise.
12156 (oacc_entry_exit_single_gang): Likewise.
12157 * tree-pretty-print.h: Likewise.
12158 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12159 (get_scalar_evolution): Likewise.
12160 (add_to_evolution): Likewise.
12161 (get_loop_exit_condition): Likewise.
12162 (analyze_evolution_in_loop): Likewise.
12163 (analyze_initial_condition): Likewise.
12164 (analyze_scalar_evolution): Likewise.
12165 (instantiate_scev): Likewise.
12166 (number_of_latch_executions): Likewise.
12167 (gather_chrec_stats): Likewise.
12168 (final_value_replacement_loop): Likewise.
12169 (scev_const_prop): Likewise.
12170 * tree-sra.c (dump_access): Likewise.
12171 (disqualify_candidate): Likewise.
12172 (create_access): Likewise.
12173 (reject): Likewise.
12174 (maybe_add_sra_candidate): Likewise.
12175 (create_access_replacement): Likewise.
12176 (analyze_access_subtree): Likewise.
12177 (analyze_all_variable_accesses): Likewise.
12178 (sra_modify_assign): Likewise.
12179 (initialize_constant_pool_replacements): Likewise.
12180 (find_param_candidates): Likewise.
12181 (decide_one_param_reduction): Likewise.
12182 (replace_removed_params_ssa_names): Likewise.
12183 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12184 * tree-ssa-copy.c (dump_copy_of): Likewise.
12185 (copy_prop_visit_cond_stmt): Likewise.
12186 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12187 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12188 (record_equivalences_from_stmt): Likewise.
12189 * tree-ssa-dse.c (compute_trims): Likewise.
12190 (delete_dead_call): Likewise.
12191 (delete_dead_assignment): Likewise.
12192 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12193 (forward_propagate_into_cond): Likewise.
12194 (pass_forwprop::execute): Likewise.
12195 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12196 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12198 (move_computations_worker): Likewise.
12199 (execute_sm): Likewise.
12200 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12201 (remove_exits_and_undefined_stmts): Likewise.
12202 (remove_redundant_iv_tests): Likewise.
12203 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12204 (adjust_iv_update_pos): Likewise.
12205 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12206 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12207 (value_replacement): Likewise.
12208 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12209 * tree-ssa-pre.c (print_pre_expr): Likewise.
12210 (get_representative_for): Likewise.
12211 (create_expression_by_pieces): Likewise.
12212 (insert_into_preds_of_block): Likewise.
12213 (eliminate_insert): Likewise.
12214 (eliminate_dom_walker::before_dom_children): Likewise.
12215 (eliminate): Likewise.
12216 (remove_dead_inserted_code): Likewise.
12217 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12218 * tree-ssa-reassoc.c (get_rank): Likewise.
12219 (eliminate_duplicate_pair): Likewise.
12220 (eliminate_plus_minus_pair): Likewise.
12221 (eliminate_not_pairs): Likewise.
12222 (undistribute_ops_list): Likewise.
12223 (eliminate_redundant_comparison): Likewise.
12224 (update_range_test): Likewise.
12225 (optimize_range_tests_var_bound): Likewise.
12226 (optimize_vec_cond_expr): Likewise.
12227 (rewrite_expr_tree): Likewise.
12228 (rewrite_expr_tree_parallel): Likewise.
12229 (linearize_expr): Likewise.
12230 (break_up_subtract): Likewise.
12231 (linearize_expr_tree): Likewise.
12232 (attempt_builtin_powi): Likewise.
12233 (attempt_builtin_copysign): Likewise.
12234 (transform_stmt_to_copy): Likewise.
12235 (transform_stmt_to_multiply): Likewise.
12236 (dump_ops_vector): Likewise.
12237 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12238 (print_scc): Likewise.
12239 (set_ssa_val_to): Likewise.
12240 (visit_reference_op_store): Likewise.
12241 (visit_use): Likewise.
12242 (sccvn_dom_walker::before_dom_children): Likewise.
12243 (run_scc_vn): Likewise.
12244 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12246 (expr_hash_elt::print): Likewise.
12247 (const_and_copies::pop_to_marker): Likewise.
12248 (const_and_copies::record_const_or_copy_raw): Likewise.
12249 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12250 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12251 (dump_predicates): Likewise.
12252 (find_uninit_use): Likewise.
12253 (warn_uninitialized_phi): Likewise.
12254 (pass_late_warn_uninitialized::execute): Likewise.
12255 * tree-ssa.c (verify_vssa): Likewise.
12256 (verify_ssa): Likewise.
12257 (maybe_optimize_var): Likewise.
12258 * tree-vrp.c (dump_value_range): Likewise.
12259 (dump_all_value_ranges): Likewise.
12260 (dump_asserts_for): Likewise.
12261 (register_edge_assert_for_2): Likewise.
12262 (vrp_visit_cond_stmt): Likewise.
12263 (vrp_visit_switch_stmt): Likewise.
12264 (vrp_visit_stmt): Likewise.
12265 (vrp_visit_phi_node): Likewise.
12266 (simplify_cond_using_ranges_1): Likewise.
12267 (fold_predicate_in): Likewise.
12268 (evrp_dom_walker::before_dom_children): Likewise.
12269 (evrp_dom_walker::push_value_range): Likewise.
12270 (evrp_dom_walker::pop_value_range): Likewise.
12271 (execute_early_vrp): Likewise.
12273 2017-05-16 Richard Biener <rguenther@suse.de>
12275 * dwarf2out.c (loc_list_from_tree_1): Do not create
12276 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12278 2017-05-16 Richard Biener <rguenther@suse.de>
12280 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12282 (note_variable_value_in_expr): If we resolved the decl ref
12283 do not push to the stack.
12285 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12287 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12288 operations in fast-math mode.
12289 (vaddq_f16): Likewise.
12290 (vmul_f16): Likewise.
12291 (vmulq_f16): Likewise.
12292 (vsub_f16): Likewise.
12293 (vsubq_f16): Likewise.
12294 * config/arm/neon.md (add<mode>3): New.
12296 (fma:<VH:mode>3): New. Also remove outdated comment.
12299 2017-05-16 Martin Liska <mliska@suse.cz>
12303 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12304 (odr_types_equivalent_p): Likewise.
12306 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12308 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12310 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12313 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12314 non-interunit SSE move alternatives with '?'.
12315 (zero-extendsidi peephole2): New peephole to skip intermediate
12316 general register in SSE zero-extend sequence.
12318 2017-05-15 Jeff Law <law@redhat.com>
12320 * reorg.c (relax_delay_slots): Create a new variable to hold
12321 the temporary target rather than clobbering TARGET_LABEL.
12323 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12324 missing argument to extract_bit_field call.
12325 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12327 2017-05-15 Martin Liska <mliska@suse.cz>
12330 * gcc.c (process_command): Do not allow empty argument of -o option.
12332 2017-05-15 Renlin Li <renlin.li@arm.com>
12334 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12335 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12336 * config/aarch64/constraints.md (Usf): Add long call check.
12337 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12338 (call_value): Likewise.
12339 (sibcall): Likewise.
12340 (sibcall_value): Likewise.
12342 (call_value_insn): New.
12343 (sibcall_insn): Update rtx pattern.
12344 (sibcall_value_insn): Likewise.
12345 (call_internal): Remove.
12346 (call_value_internal): Likewise.
12347 (sibcall_internal): Likewise.
12348 (sibcall_value_internal): Likewise.
12349 (call_reg): Likewise.
12350 (call_symbol): Likewise.
12351 (call_value_reg): Likewise.
12352 (call_value_symbol): Likewise.
12354 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12357 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12359 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12361 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12362 compatible with CCGOCmode and with CCZmode.
12364 2017-05-14 Martin Sebor <msebor@redhat.com>
12366 PR middle-end/77671
12367 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12368 (gimple_fold_builtin_snprintf): Same.
12369 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12370 (gimple_fold_builtin_snprintf): Same.
12371 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12372 of character types.
12373 (is_call_safe): New function.
12374 (try_substitute_return_value): Call it.
12375 (try_simplify_call): New function.
12376 (pass_sprintf_length::handle_gimple_call): Call it.
12378 2017-05-14 Martin Sebor <msebor@redhat.com>
12380 PR middle-end/80669
12381 * builtins.c (expand_builtin_stpncpy): Simplify.
12383 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12385 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12386 * config/i386/i386.h
12387 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12388 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12389 (struct machine_function): Add new members call_ms2sysv,
12390 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12391 (struct machine_frame_state): New fields sp_realigned and
12392 sp_realigned_offset.
12393 * config/i386/i386.c
12394 (enum xlogue_stub): New enum.
12395 (enum xlogue_stub_sets): New enum.
12396 (class xlogue_layout): New class.
12397 (struct ix86_frame): New fields stack_realign_allocate_offset,
12398 stack_realign_offset and outlined_save_offset. Modify comments to
12399 detail stack layout when using out-of-line stubs.
12400 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12401 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12402 -mcall-ms2sysv-xlogues.
12403 (stub_managed_regs): New static variable.
12404 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12405 registers managed by out-of-line stub.
12406 (disable_call_ms2sysv_xlogues): New function.
12407 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12408 m->call_ms2sysv when appropriate and compute frame layout for
12410 (sp_valid_at, fp_valid_at): New inline functions.
12411 (choose_basereg): New function.
12412 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12414 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12415 Use align parameter of choose_baseaddr to generated aligned SSE movs
12417 (pro_epilogue_adjust_stack): Modify to track
12418 machine_frame_state::sp_realigned.
12419 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12420 (ix86_nsaved_sseregs): Likewise.
12421 (ix86_emit_save_regs): Likewise.
12422 (ix86_emit_save_regs_using_mov): Likewise.
12423 (ix86_emit_save_sse_regs_using_mov): Likewise.
12424 (get_scratch_register_on_entry): Likewise.
12425 (gen_frame_set): New function.
12426 (gen_frame_load): Likewise.
12427 (gen_frame_store): Likewise.
12428 (emit_outlined_ms2sysv_save): Likewise.
12429 (emit_outlined_ms2sysv_restore): Likewise.
12430 (ix86_expand_prologue): Modify stack re-alignment code and call
12431 emit_outlined_ms2sysv_save when appropriate.
12432 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12433 parameter rtx_insn *insn, which allows the function to be used to only
12434 generate the notes.
12435 (ix86_expand_epilogue): Modify validity checks of frame and stack
12436 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12437 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12438 * config/i386/predicates.md
12439 (save_multiple): New predicate.
12440 (restore_multiple): Likewise.
12441 * config/i386/sse.md
12442 (save_multiple<mode>): New pattern.
12443 (save_multiple_realign<mode>): Likewise.
12444 (restore_multiple<mode>): Likewise.
12445 (restore_multiple_and_return<mode>): Likewise.
12446 (restore_multiple_leave_return<mode>): Likewise.
12447 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12449 2017-05-14 Julia Koval <julia.koval@intel.com>
12451 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12452 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12453 (__builtin_ia32_xsetbv): New builtins.
12454 * config/i386/i386.c (ix86_expand_special_args_builtin):
12456 (ix86_expand_builtin): Special expand for new intrinsics.
12457 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12458 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12459 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12461 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12463 * cfganal.c (inverted_post_order_compute): Change argument type
12465 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12466 * df-core.c (rest_of_handle_df_initialize): Adjust.
12467 (rest_of_handle_df_finish): Likewise.
12468 (df_analyze_1): Likewise.
12469 (df_analyze): Likewise.
12470 (loop_inverted_post_order_compute): Change argument to be a vec *.
12471 (df_analyze_loop): Adjust.
12472 (df_get_n_blocks): Likewise.
12473 (df_get_postorder): Likewise.
12474 * df.h (struct df_d): Change field to be a vec.
12475 * lcm.c (compute_laterin): Adjust.
12476 (compute_available): Likewise.
12477 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12478 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12479 * tree-ssa-pre.c (compute_antic): Likewise.
12481 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12483 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12484 (depth_first_search::depth_first_search): Change structure init
12485 function to this constructor.
12486 (depth_first_search::add_bb): Rename function to this member.
12487 (depth_first_search::execute): Likewise.
12488 (flow_dfs_compute_reverse_finish): Adjust.
12490 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12492 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12493 (longest_simple_path): Likewise.
12494 * shrink-wrap.c (spread_components): Likewise.
12495 (disqualify_problematic_components): Likewise.
12496 (emit_common_heads_for_components): Likewise.
12497 (emit_common_tails_for_components): Likewise.
12498 (insert_prologue_epilogue_for_components): Likewise.
12500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12502 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12505 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12507 * df-core.c (df_set_blocks): Start using auto_bitmap.
12508 (df_compact_blocks): Likewise.
12509 * df-problems.c (df_rd_confluence_n): Likewise.
12510 * df-scan.c (df_insn_rescan_all): Likewise.
12511 (df_process_deferred_rescans): Likewise.
12512 (df_update_entry_block_defs): Likewise.
12513 (df_update_exit_block_uses): Likewise.
12514 (df_entry_block_bitmap_verify): Likewise.
12515 (df_exit_block_bitmap_verify): Likewise.
12516 (df_scan_verify): Likewise.
12517 * lra-constraints.c (lra_constraints): Likewise.
12518 (undo_optional_reloads): Likewise.
12519 (lra_undo_inheritance): Likewise.
12520 * lra-remat.c (calculate_gen_cands): Likewise.
12521 (do_remat): Likewise.
12522 * lra-spills.c (assign_spill_hard_regs): Likewise.
12523 (spill_pseudos): Likewise.
12524 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12525 (bitmap_set_subtract_values): Likewise.
12527 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12529 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12530 management with auto_bitmap.
12531 (fix_inter_tick): Likewise.
12532 (fix_recovery_deps): Likewise.
12533 * ira.c (add_store_equivs): Likewise.
12534 (find_moveable_pseudos): Likewise.
12535 (split_live_ranges_for_shrink_wrap): Likewise.
12536 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12537 (rtx_reuse_manager::seen_def_p): Likewise.
12538 (rtx_reuse_manager::set_seen_def): Likewise.
12539 * print-rtl.h (class rtx_reuse_manager): Likewise.
12541 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12543 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12545 (migrate_btr_def): Likewise.
12546 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12547 * df-core.c (loop_post_order_compute): Likewise.
12548 (loop_inverted_post_order_compute): Likewise.
12549 * hsa-common.h: Likewise.
12550 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12551 * init-regs.c (initialize_uninitialized_regs): Likewise.
12552 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12553 (inline_small_functions): Likewise.
12554 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12555 * ira.c (combine_and_move_insns): Likewise.
12556 (build_insn_chain): Likewise.
12557 * loop-invariant.c (find_invariants): Likewise.
12558 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12559 * predict.c (tree_predict_by_opcode): Likewise.
12560 (predict_paths_leading_to): Likewise.
12561 (predict_paths_leading_to_edge): Likewise.
12562 (estimate_loops_at_level): Likewise.
12563 (estimate_loops): Likewise.
12564 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12565 (spread_components): Likewise.
12566 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12567 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12568 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12569 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12570 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12571 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12572 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12573 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12574 (mark_threaded_blocks): Likewise.
12575 (thread_through_all_blocks): Likewise.
12576 * tree-ssa.c (verify_ssa): Likewise.
12577 (execute_update_addresses_taken): Likewise.
12578 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12580 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12582 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12584 (post_order_compute): Likewise.
12585 (inverted_post_order_compute): Likewise.
12586 (pre_and_rev_post_order_compute_fn): Likewise.
12588 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12590 * genrecog.c (int_set::int_set): Explicitly construct our
12591 auto_vec base class.
12592 * vec.h (auto_vec::auto_vec): New constructor.
12594 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12596 * bitmap.h (class auto_bitmap): New constructor taking
12597 bitmap_obstack * argument.
12599 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12601 * bitmap.h (class auto_bitmap): Change type of m_bits to
12602 bitmap_head, and adjust ctor / dtor and member operators.
12604 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12606 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12607 when returned register mode doesn't match original mode.
12609 2017-05-12 Jeff Law <law@redhat.com>
12610 Jakub Jelinek <jakub@redhat.com>
12612 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
12613 we look for cc setter after the compare-elim changes.
12614 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
12615 within the vector to match what compare-elim now expects.
12616 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
12617 (xorsi3_flags, one_cmplsi2_flags): Likewise.
12619 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
12620 after the compare-elim changes.
12621 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
12622 the vector to match what compare-elim now expects.
12623 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
12624 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
12625 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
12626 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
12627 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
12629 * config/visium/visium.c (single_set_and_flags): Fix where
12630 we look for cc setter after the compare-elim changes.
12631 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
12632 with the vector to match what compare-elim now expects.
12633 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
12634 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
12635 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
12636 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
12637 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
12638 (neg<mode>2_insn_set_overflow): Likewise.
12640 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
12642 PR middle-end/79794
12643 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
12644 maybe_expand_insn call, set ops[0].target. If still set after call,
12645 set alt_rtl. Add extra arg to recursive calls.
12646 (extract_bit_field): Add alt_rtl argument. Pass to
12648 * expmed.h (extract_bit_field): Fix prototype.
12649 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
12650 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
12651 to extract_bit_field_calls.
12652 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
12653 Pass alt_rtl to extract_bit_field calls.
12654 * calls.c (store_unaligned_arguments_into_psuedos)
12655 load_register_parameters): Pass extra NULL to extract_bit_field calls.
12656 * optabs.c (maybe_legitimize_operand): Clear op->target when call
12658 * optabs.h (struct expand_operand): Add target bitfield.
12660 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12662 * compare-elim.c (try_eliminate_compare): Canonicalize
12663 operation with embedded compare to
12664 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
12665 (set (reg) (operation)].
12667 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
12669 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12672 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
12673 cost of adding a carry flag for ADC instruction.
12674 [case MINUS]: Ignore the cost of subtracting a carry flag
12675 for SBB instruction.
12677 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
12679 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
12681 * config/rs6000/bmiintrin.h: New file.
12682 * config/rs6000/bmi2intrin.h: New file.
12683 * config/rs6000/x86intrin.h: New file.
12685 2017-05-12 Jeff Law <law@redhat.com>
12687 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
12690 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
12692 PR middle-end/80707
12693 * tree-cfg.c: Remove cfg edges of unreachable case statements.
12695 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12697 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12698 early expansion of vector divide builtins.
12699 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
12700 builtins identified as having unsigned arguments.
12702 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12704 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
12705 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
12706 expansion of vector logical operations (and, andc, or, xor,
12709 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12711 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
12712 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
12714 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12717 early GIMPLE expansion of vector multiplies.
12719 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12721 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
12722 TARGET_HAVE_MOVT conditional.
12723 (movt splitter): Likewise.
12725 2017-05-12 Richard Biener <rguenther@suse.de>
12727 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
12728 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12729 Fold all stmts not inplace.
12731 2017-05-12 Richard Biener <rguenther@suse.de>
12733 PR tree-optimization/80713
12734 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
12735 inserted_exprs bit for not removed stmts.
12737 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
12739 PR middle-end/69921
12740 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
12741 parallelized" attribute for parallelized OpenACC kernels.
12742 * omp-offload.c (execute_oacc_device_lower): Use it.
12744 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
12745 Set "oacc kernels" attribute.
12746 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
12747 parameter. Adjust all users.
12748 (oacc_fn_attrib_kernels_p): Remove function.
12749 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
12750 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
12751 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
12752 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
12753 assert "oacc kernels" attribute is set.
12755 2017-05-11 Carl Love <cel@us.ibm.com>
12757 * config/rs6000/rs6000-c: Add support for built-in functions
12758 vector unsigned char vec_popcnt (vector signed char)
12759 vector unsigned char vec_popcnt (vector unsigned char)
12760 vector unsigned short vec_popcnt (vector signed short)
12761 vector unsigned short vec_popcnt (vector unsigned short)
12762 vector unsigned int vec_popcnt (vector signed int)
12763 vector unsigned int vec_popcnt (vector unsigned int)
12764 vector unsigned long long vec_popcnt (vector signed long long)
12765 vector unsigned long long vec_popcnt (vector unsigned long long)
12766 vector signed long long vec_slo (vector signed long long,
12767 vector signed char)
12768 vector signed long long vec_slo (vector signed long long,
12769 vector unsigned char)
12770 vector unsigned long long vec_slo (vector unsigned long long,
12771 vector signed char)
12772 vector unsigned long long vec_slo (vector unsigned long long,
12773 vector unsigned char)
12774 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
12775 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
12776 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
12777 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
12778 * doc/extend.texi: Update the built-in documentation file for the
12779 new built-in functions.
12781 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12783 * attribs.h (sorted_attr_string): Move machine independent
12784 functions for target clone support from the i386 port to common
12785 code. Rename ix86_function_versions to common_function_versions.
12786 Rename make_name to make_unique_name.
12787 (common_function_versions): Likewise.
12788 (make_unique_name): Likewise.
12789 (make_dispatcher_decl): Likewise.
12790 (is_function_default_version): Likewise.
12791 * attribs.c (attr_strcmp): Likewise.
12792 (sorted_attr_string): Likewise.
12793 (common_function_versions): Likewise.
12794 (make_unique_name): Likewise.
12795 (make_dispatcher_decl): Likewise.
12796 (is_function_default_version): Likewise.
12797 * config/i386/i386.c (attr_strcmp): Likewise.
12798 (sorted_attr_string): Likewise.
12799 (ix86_function_versions): Likewise.
12800 (make_name): Likewise.
12801 (make_dispatcher_decl): Likewise.
12802 (is_function_default_version): Likewise.
12803 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
12805 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12808 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12809 Account for direct move costs for vec_construct of integer
12812 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
12815 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
12816 (UNSPEC_STX_ATOMIC): Ditto.
12817 (loaddi_via_sse): New insn.
12818 (storedi_via_sse): Ditto.
12819 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
12820 Update corresponding peephole2 patterns.
12821 (atomic_storedi_fpu): Ditto.
12823 2017-05-11 Julia Koval <julia.koval@intel.com>
12825 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
12826 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
12828 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
12829 (__builtin_ia32_rsqrt14ss_mask): New builtins.
12830 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
12832 2017-05-11 Nathan Sidwell <nathan@acm.org>
12834 * graphite-poly.c: Include dumpfile.h.
12836 * dumpfle.h (dump_function): Declare here ...
12837 * tree-dump.h (dump_function): ... not here.
12838 * dumpfile.c: #include tree-cfg.h.
12839 (dump_function): Move here from ...
12840 * tree-dump.c (dump_function): ... here.
12841 * gimplify.c: #include splay-tree.h, not tree-dump.h.
12842 * graphite-poly.c: Don't include tree-dump.h.
12843 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
12844 * print-tree.c: Likewise.
12845 * stor-layout.c: Likewise.
12846 * tree-nested.c: Likewise.
12848 * dumpfile.c (dump_start): Use TDF_FLAGS.
12849 (dump_enable_all): Fix TDF_KIND check thinko.
12851 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12853 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12854 array entries to represent two legal parameterizations of the
12855 overloaded __builtin_cmpb function, as represented by the
12856 P6_OV_BUILTIN_CMPB constant.
12857 (altivec_resolve_overloaded_builtin): Add special case handling
12858 for the __builtin_cmpb function, as represented by the
12859 P6_OV_BUILTIN_CMPB constant.
12860 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
12861 (BU_P6_64BIT_2): New macro.
12862 (BU_P6_OVERLOAD_2): New macro
12863 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
12864 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
12865 (CMPB): Add overload support to represent both 32-bit and 64-bit
12866 compare-bytes function.
12867 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12868 support for TARGET_CMPB.
12869 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
12870 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12871 documentation of the __builtin_cmpb overloaded built-in function.
12873 2017-05-11 Richard Biener <rguenther@suse.de>
12875 PR tree-optimization/80705
12876 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
12877 bases are not vectorizable.
12879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12881 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
12882 when counting register pressure.
12884 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12886 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
12887 (iv_ca_more_deps): Renamed to ...
12888 (iv_ca_compare_deps): ... this.
12889 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
12891 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12893 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
12895 (determine_group_iv_costs): ... here.
12896 (find_inv_vars_cb): Record inv var if it's not recorded before.
12898 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12900 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
12901 (get_shiftadd_cost): Ditto.
12903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12905 * tree-ssa-address.c: Include header file.
12906 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
12908 (add_to_parts): Refactor.
12909 (addr_to_parts): New parameter. Update use of move_hint_to_base.
12910 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
12913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12915 PR tree-optimization/53090
12916 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
12918 (extract_cond_operands): Detect condition with IV on both sides
12919 and return COMP_IV_EXPR_2.
12920 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
12921 (rewrite_use_compare): Simplify by removing call to function
12922 extract_cond_operands.
12924 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12926 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
12927 (extract_cond_operands): Detect condition comparing against non-
12928 invariant bound and return appropriate enum value.
12929 (find_interesting_uses_cond): Update use of extract_cond_operands.
12930 Handle its return value accordingly.
12931 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
12933 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12935 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
12936 nonlinear iv_use computation in loop invariant sensitive way.
12938 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12940 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
12941 (find_iv_candidates): Call relate_compare_use_with_all_cands.
12943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12945 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
12946 (dump_cand): Support iv_cand.inv_exprs.
12947 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
12949 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
12952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12954 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
12956 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
12957 as local function. Include necessary header files.
12958 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
12960 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12962 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
12964 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12966 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
12967 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
12968 RSHIFT_EXPR and BIT_NOT_EXPR.
12970 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12972 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
12973 (adjust_setup_cost): New parameter supporting round up adjustment.
12974 (struct address_cost_data): Delete.
12975 (force_expr_to_var_cost): Don't bound cost with spill_cost.
12976 (split_address_cost, ptr_difference_cost): Delete.
12977 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
12978 (struct ainc_cost_data): New struct.
12979 (get_address_cost_ainc): New function.
12980 (get_address_cost, get_computation_cost): Reimplement.
12981 (determine_group_iv_cost_address): Record inv_expr for all uses of
12983 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
12984 (iv_ca_has_deps): Reimplemented to ...
12985 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
12987 (iv_ca_extend): Call iv_ca_more_deps.
12989 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12991 * tree-ssa-address.c (struct mem_address): Move to header file.
12992 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
12993 * tree-ssa-address.h (struct mem_address): Move from C file.
12994 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
12996 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12998 * tree-affine.h (aff_combination_type): New interface.
12999 (aff_combination_zero_p): Remove static.
13000 (aff_combination_const_p): New interface.
13001 (aff_combination_singleton_var_p): New interfaces.
13003 2017-05-11 Richard Biener <rguenther@suse.de>
13005 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13006 Skip unreachable blocks and destinations.
13007 (eliminate): Move stmt removal and fixup ...
13008 (fini_eliminate): ... here. Skip inserted exprs.
13009 (pass_pre::execute): Move fini_pre after fini_eliminate.
13010 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13011 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13012 PRE to get rid of dead code that has invalid SSA form and
13013 split critical edges again.
13015 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13017 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13019 2017-05-11 Richard Biener <rguenther@suse.de>
13021 * passes.c (execute_function_todo): Verify loops if they are
13022 said to be up-to-date.
13023 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13024 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13026 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13029 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13030 handle calling assemble_external ourself.
13033 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13034 modes with zero size. Enhance comment.
13036 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13038 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13039 built-ins for vec_xl and vec_xst with short and char pointer
13042 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13044 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13045 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13046 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13047 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13048 (_mm_maskz_min_round_ss): New intrinsics.
13049 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13050 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13051 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13052 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13053 (__builtin_ia32_minss_mask_round): New builtins.
13054 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13055 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13056 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13058 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13059 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13061 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13064 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13066 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13067 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13068 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13069 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13070 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13071 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13072 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13073 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13074 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13075 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13076 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13077 (__builtin_ia32_mulss_mask_round): New builtins.
13078 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13079 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13080 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13082 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13083 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13085 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13088 2017-05-10 Julia Koval <julia.koval@intel.com>
13090 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13091 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13092 (_mm256_setr_m128i): New intrinsics.
13094 2017-05-10 Julia Koval <julia.koval@intel.com>
13096 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13097 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13098 (_mm_maskz_rcp14_ss): New intrinsics.
13099 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13100 (__builtin_ia32_rcp14ss_mask): New builtins.
13101 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13103 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13105 PR tree-optimization/51513
13106 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13107 (assert_unreachable_fallthru_edge_p): Use it.
13108 (group_case_labels_stmt): Likewise.
13109 * tree-cfg.h: Prototype it.
13110 * stmt.c: Include cfghooks.h and tree-cfg.h.
13111 (emit_case_dispatch_table) <gap_label>: New local variable.
13112 Use it to fill dispatch table gaps.
13113 Test for default_label before updating probabilities.
13114 (expand_case) <default_label>: Remove unneeded initialization.
13115 Test for unreachable default case statement and remove its edge.
13116 Set default_label accordingly.
13117 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13119 2017-05-10 Carl Love <cel@us.ibm.com>
13121 * config/rs6000/rs6000-c: Add support for built-in functions
13122 vector signed char vec_neg (vector signed char)
13123 vector signed short int vec_neg (vector short int)
13124 vector signed int vec_neg (vector signed int)
13125 vector signed long long vec_neg (vector signed long long)
13126 vector float vec_neg (vector float)
13127 vector double vec_neg (vector double)
13128 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13130 * config/rs6000/altivec.h: Add define for vec_neg
13131 * doc/extend.texi: Update the built-in documentation for the
13132 new built-in functions.
13134 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13136 PR tree-optimization/77644
13137 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13139 2017-05-10 Nathan Sidwell <nathan@acm.org>
13141 * dumpfile.h (TDI_lang_all): New.
13142 (TDF_KIND): New. Renumber others
13143 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13145 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13147 (get_dump_file_name): Adjust suffix generation.
13148 (dump_enable_all): Use TDF_KIND.
13149 * doc/invoke.texi (-fdump-lang-all): Document.
13151 * dumpfile.h: Tabify.
13153 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13156 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13157 Move member access before delete.
13159 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13161 * tree-inline.c (expand_call_inline): Split block at stmt
13164 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13167 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13168 are now unused after splitting mov{sf,sd}_hardfloat.
13169 (f32_lr2): Likewise.
13170 (f32_lm): Likewise.
13171 (f32_lm2): Likewise.
13172 (f32_li): Likewise.
13173 (f32_li2): Likewise.
13174 (f32_lv): Likewise.
13175 (f32_sr): Likewise.
13176 (f32_sr2): Likewise.
13177 (f32_sm): Likewise.
13178 (f32_sm2): Likewise.
13179 (f32_si): Likewise.
13180 (f32_si2): Likewise.
13181 (f32_sv): Likewise.
13182 (f32_dm): Likewise.
13183 (f32_vsx): Likewise.
13184 (f32_av): Likewise.
13185 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13186 For movsf, order stores so the VSX stores occur before the GPR
13187 store which encourages the register allocator to use a traditional
13188 FPR instead of a GPR. For movsd, order the stores so that the GPR
13189 store comes before the VSX stores to allow the power6 to work.
13190 This is due to the power6 not having a 32-bit integer store
13191 instruction from a FPR.
13192 (movsf_hardfloat): Likewise.
13193 (movsd_hardfloat): Likewise.
13195 2017-05-09 Martin Sebor <msebor@redhat.com>
13197 PR translation/80280
13198 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13201 PR translation/80280
13202 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13203 data member added in r247778.
13204 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13206 2017-05-09 Nathan Sidwell <nathan@acm.org>
13208 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13210 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13213 2017-05-09 Marek Polacek <polacek@redhat.com>
13215 * doc/invoke.texi: Fix typo.
13217 2017-05-09 Richard Biener <rguenther@suse.de>
13219 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13220 (vrp_val_is_min): Likewise.
13221 (set_value_range_to_value): Likewise.
13222 (set_value_range_to_nonnegative): Likewise.
13223 (gimple_assign_nonzero_p): Likewise.
13224 (gimple_stmt_nonzero_p): Likewise.
13225 (vrp_int_const_binop): Likewise. Remove unreachable case.
13226 (adjust_range_with_scev): Adjust comments.
13227 (compare_range_with_value): Likewise.
13228 (extract_range_from_phi_node): Likewise.
13229 (test_for_singularity): Likewise.
13231 2017-05-09 Richard Biener <rguenther@suse.de>
13233 * tree-vrp.c (get_single_symbol): Add assert that we don't
13234 get overflowed constants as invariant part.
13235 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13236 checks. Use wi::cmp instead of recursing for integer constants.
13237 (compare_values): Just ignore whether we assumed undefined
13238 overflow instead of failing the compare.
13239 (extract_range_for_var_from_comparison_expr): Add comment before the
13240 TREE_NO_WARNING sets.
13241 (test_for_singularity): Likewise.
13242 (extract_range_from_comparison): Do not disable optimization
13243 when we assumed undefined overflow.
13244 (extract_range_basic): Remove init of unused var.
13246 2017-05-09 Richard Biener <rguenther@suse.de>
13248 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13249 (extract_range_from_multiplicative_op_1): Adjust.
13250 (extract_range_from_binary_expr_1): Use int_const_binop.
13252 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13255 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13256 rs6000_store_data_bypass_p in seven define_bypass directives and
13257 in several comments.
13258 * config/rs6000/rs6000-protos.h: Add prototype for
13259 rs6000_store_data_bypass_p function.
13260 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13261 function implements slightly different (rs6000-specific) semantics
13262 than store_data_bypass_p, returning false rather than aborting
13263 with assertion error when arguments do not satisfy the
13264 requirements of store data bypass.
13265 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13266 rs6000_store_data_bypass_p.
13268 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13270 * config/xtensa/xtensa-protos.h
13271 (xtensa_initial_elimination_offset): New declaration.
13272 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13273 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13274 macro definition, add case for FRAME_POINTER_REGNUM when
13275 FRAME_GROWS_DOWNWARD.
13276 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13277 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13278 xtensa_initial_elimination_offset.
13280 2017-05-08 Nathan Sidwell <nathan@acm.org>
13282 * doc/invoke.texi: Alphabetize -fdump options.
13284 2017-05-08 Martin Sebor <msebor@redhat.com>
13286 PR translation/80280
13287 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13289 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13291 * target.def (compute_frame_layout): New optional target hook.
13292 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13293 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13294 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13296 * reload1.c (verify_initial_elim_offsets): Likewise.
13297 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13298 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13299 (arm_get_frame_offsets): Split up into this ...
13300 (arm_compute_frame_layout): ... and this function.
13302 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13304 * config/aarch64/constraints.md (Usa): New constraint.
13305 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13307 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13309 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13310 with_multilib_list after it has been checked.
13312 2017-05-08 Richard Biener <rguenther@suse.de>
13314 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13315 (bitmap_set_subtract_values): Likewise.
13317 2017-05-08 Richard Biener <rguenther@suse.de>
13319 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13320 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13322 (gimple_stmt_nonzero_warnv_p): Rename to ...
13323 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13325 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13326 (extract_range_basic): Adjust, do not disable propagation on
13327 strict overflow sensitive simplification.
13328 (vrp_visit_cond_stmt): Likewise.
13330 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13332 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13333 body size unconditionally.
13335 2017-05-07 Jeff Law <law@redhat.com>
13338 2017-05-06 Jeff Law <law@redhat.com>
13339 PR tree-optimization/78496
13340 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13343 PR tree-optimization/78496
13344 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13345 (simplify_stmt_using_ranges): Call it.
13346 (vrp_dom_walker::before_dom_children): Extract equivalences
13347 from an ASSERT_EXPR with an equality comparison against a
13350 2017-05-06 Jeff Law <law@redhat.com>
13352 PR tree-optimization/78496
13353 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13356 PR tree-optimization/78496
13357 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13358 (simplify_stmt_using_ranges): Call it.
13359 (vrp_dom_walker::before_dom_children): Extract equivalences
13360 from an ASSERT_EXPR with an equality comparison against a
13363 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13365 * lra-constraints.c (lra_copy_reg_equiv): New function.
13366 (split_reg): Use it to copy equivalence information from the
13367 original register to the spill register.
13369 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13371 PR rtl-optimization/75964
13372 * simplify-rtx.c (simplify_const_relational_operation): Remove
13373 invalid handling of comparisons of integer ABS.
13375 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13377 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13378 initialize to zero.
13379 (init_regs): Remove declaration.
13380 (function_arg_advance_32): Initialize error_p as boolean variable.
13382 2017-05-05 Nathan Sidwell <nathan@acm.org>
13384 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13385 lines. Use for (;;).
13387 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13389 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13390 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13391 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13392 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13393 VF=2 that require versioning.
13395 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13397 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13400 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13402 * diagnostic.h (diagnostic_override_option_index): Convert from
13403 macro to inline function.
13405 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13407 * diagnostic.c (last_module_changed_p): New function.
13408 (set_last_module): New function.
13409 (diagnostic_report_current_module): Convert macro usage to
13410 the above functions.
13411 * diagnostic.h (diagnostic_context::last_module): Strengthen
13412 from const line_map * to const line_map_ordinary *.
13413 (diagnostic_last_module_changed): Delete macro.
13414 (diagnostic_set_last_module): Delete macro.
13416 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13418 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13419 with diagnostic_report_diagnostic.
13420 (diagnostic_n_impl_richloc): Likewise.
13421 * diagnostic.h (report_diagnostic): Delete macro.
13422 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13423 with diagnostic_report_diagnostic.
13424 * substring-locations.c (format_warning_va): Likewise.
13426 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13428 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13429 save/restor of format_spec. Move option-printing code to...
13430 (print_option_information): ...this new function, and
13431 reimplement by simply printing to the pretty_printer,
13432 rather than appending to the format string.
13434 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13436 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13437 handling logic into...
13438 (update_effective_level_from_pragmas): ...this new function.
13440 2017-05-04 Andrew Waterman <andrew@sifive.com>
13442 * config/riscv/riscv.opt (mstrict-align): New option.
13443 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13444 (SLOW_UNALIGNED_ACCESS): Define.
13445 (riscv_slow_unaligned_access): Declare.
13446 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13448 (riscv_slow_unaligned_access): New variable.
13449 (rocket_tune_info): Set slow_unaligned_access to true.
13450 (optimize_size_tune_info): Set slow_unaligned_access to false.
13451 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13452 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13453 (riscv_option_override): Set riscv_slow_unaligned_access.
13454 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13456 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13458 * config/riscv/riscv.md: Unify indentation.
13460 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13465 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13467 (extendsi<mode>2): Add support for doing sign extension via
13468 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13469 don't have ISA 3.0 instructions.
13470 (extendsi<mode>2 splitter): Likewise.
13471 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13472 generate the normal insns since SImode can now go in vector
13473 registers. Disallow the special UNSPECs needed for previous
13474 machines to hide SImode being used. Add new insns
13475 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13476 (fix_trunc<mode>si2_stfiwx): Likewise.
13477 (fix_trunc<mode>si2_internal): Likewise.
13478 (fixuns_trunc<mode>si2): Likewise.
13479 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13480 (fctiw<u>z_<mode>_smallint): Likewise.
13481 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13482 of floating point to 32-bit integer from doing a direct move to
13483 the GPR registers to do a store.
13484 (fctiwz_<mode>): Break long line.
13486 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13488 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13489 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13490 (addr_list, addr_offset_valid_p): New.
13491 (split_address_groups): Check offset validity with above function.
13492 (gt-tree-ssa-loop-ivopts.h): Include header file.
13494 2017-05-05 Nathan Sidwell <nathan@acm.org>
13496 * config.gcc (arm*-*-*): Add missing 'fi'.
13498 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13500 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13501 included in -fopt-info does not matter.
13502 * doc/optinfo.texi (-fopt-info): Fix description of default
13503 behavour. Explicitly say order of options included in -fopt-info
13506 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13508 * config.gcc: Allow combinations of aprofile and rmprofile values for
13509 --with-multilib-list.
13510 * config/arm/t-multilib: New file.
13511 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13512 variables. Remove setting of ISA and floating-point ABI in
13513 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13514 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13515 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13516 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13518 * config/arm/t-rmprofile: Likewise except for the matches changes.
13519 * doc/install.texi (--with-multilib-list): Document the combination of
13520 aprofile and rmprofile values and warn about pitfalls in doing that.
13522 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13524 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13525 (movdi_aarch64): Likewise.
13527 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13529 PR tree-optimization/80632
13530 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13532 (build_arrays): Initialize it for virtual phis.
13533 (fix_phi_nodes): Use it for virtual phis.
13535 PR tree-optimization/80558
13536 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13537 [x, y] op z into [x op, y op z] for op & or | if conditions
13540 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13541 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13544 * config/arm/arm.md (use_literal_pool): Remove.
13545 (64-bit immediate split): No longer takes cost into consideration
13546 if arm_disable_literal_pool is enabled.
13547 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13548 used when arm_disable_literal_pool is enabled.
13549 (arm_max_const_double_inline_cost): Remove use of
13550 arm_disable_literal_pool.
13551 (push_minipool_fix): Add assert.
13552 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13553 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13554 (no_literal_pool_sf_immediate): New.
13556 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13558 PR tree-optimization/80613
13559 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13560 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13562 2017-05-05 Richard Biener <rguenther@suse.de>
13564 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13566 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13568 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13569 of this flag from insn conditions due to removal from r247495.
13571 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13573 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13575 (arm_early_store_addr_dep_ptr): Likewise.
13576 * config/arm/aarch-common-protos.h
13577 (arm_early_load_addr_dep_ptr): Add prototype.
13578 (arm_early_store_addr_dep_ptr): Likewise.
13579 * config/arm/cortex-a53.md: Add new bypasses.
13581 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13583 * tree.c (next_type_uid): Change type to unsigned.
13584 (type_hash_canon): Decrement back next_type_uid if
13585 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13586 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13589 2017-05-04 Martin Sebor <msebor@redhat.com>
13591 * builtins.c: Fix a trivial typo in a comment.
13593 PR middle-end/79234
13594 * builtins.c (check_sizes): Adjust to handle reading past the end.
13595 Avoid printing excessive upper bound of ranges. Use %E to print
13596 tree nodes instead of converting them to %wu.
13597 (expand_builtin_memchr): New function.
13598 (compute_dest_size): Rename...
13599 (compute_objsize): ...to this.
13600 (expand_builtin_memcpy): Adjust.
13601 (expand_builtin_mempcpy): Adjust.
13602 (expand_builtin_strcat): Adjust.
13603 (expand_builtin_strcpy): Adjust.
13604 (check_strncat_sizes): Adjust.
13605 (expand_builtin_strncat): Adjust.
13606 (expand_builtin_strncpy): Adjust and simplify.
13607 (expand_builtin_memset): Adjust.
13608 (expand_builtin_bzero): Adjust.
13609 (expand_builtin_memcmp): Adjust.
13610 (expand_builtin): Handle memcmp.
13611 (maybe_emit_chk_warning): Check strncat just once.
13613 2017-05-04 Martin Sebor <msebor@redhat.com>
13615 PR preprocessor/79214
13616 PR middle-end/79222
13617 PR middle-end/79223
13618 * builtins.c (check_sizes): Add inlining context and issue
13619 warnings even when -Wno-system-headers is set.
13620 (check_strncat_sizes): Same.
13621 (expand_builtin_strncat): Same.
13622 (expand_builtin_memmove): New function.
13623 (expand_builtin_stpncpy): Same.
13624 (expand_builtin): Handle memmove and stpncpy.
13626 2017-05-04 Bin Cheng <bin.cheng@arm.com>
13628 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
13629 which is not used any more.
13631 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13633 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
13635 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13637 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
13638 (cortexa53_tunings): Likewise.
13639 (cortexa57_tunings): Likewise.
13640 (cortexa72_tunings): Likewise.
13641 (cortexa73_tunings): Likewise.
13643 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13645 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
13646 Set loop alignment to 8.
13648 2017-05-04 Martin Sebor <msebor@redhat.com>
13650 PR translation/80280
13651 * builtins.c (expand_builtin_object_size): Add missing quoting to
13652 %D and like directives.
13653 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
13654 (hsa_type_for_tree_type): Same.
13655 (verify_function_arguments): Same.
13656 * symtab.c (symbol_table::change_decl_assembler_name): Same.
13657 * varasm.c (get_section): Same.
13660 2017-05-04 Martin Sebor <msebor@redhat.com>
13662 PR translation/80280
13663 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
13665 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13667 * config/aarch64/aarch64.c (generic_addrcost_table):
13668 Change HI/TI mode setting.
13670 2017-05-04 Martin Jambor <mjambor@suse.cz>
13672 PR tree-optimization/80622
13673 * tree-sra.c (comes_initialized_p): New function.
13674 (build_accesses_from_assign): Only set write lazily when
13675 comes_initialized_p is false.
13676 (analyze_access_subtree): Use comes_initialized_p.
13677 (propagate_subaccesses_across_link): Assert !comes_initialized_p
13678 instead of testing for PARM_DECL.
13680 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13682 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
13683 constraint on operand 0 to allow more general addressing modes.
13684 Adjust output template.
13685 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
13687 * config/aarch64/aarch64-protos.h
13688 (aarch64_address_valid_for_prefetch_p): Declare prototype.
13689 * config/aarch64/constraints.md (Dp): New address constraint.
13690 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
13693 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
13695 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
13696 update use of estimate_ipcp_clone_size_and_time.
13697 (estimate_local_effects): Update use of
13698 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
13699 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
13700 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
13701 Return nonspecialized time.
13703 2017-05-04 Richard Biener <rguenther@suse.de>
13705 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
13706 for the last VUSE which def dominates the PHI. Directly call
13708 (get_continuation_for_phi_1): Remove.
13710 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
13712 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
13713 to explain the use of truncating division. Cap the number of
13714 iterations to the maximum given by nb_iterations_upper_bound,
13717 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13719 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
13720 * configure: Regenerate.
13721 * config.in: Regenerate.
13722 * config/i386/driver-mingw32.c: new file.
13723 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
13724 * config.host: Link driver-mingw32.o on MinGW host.
13725 * doc/install.texi: Document new --enable-mingw-wildcard configure
13728 2017-05-04 Marek Polacek <polacek@redhat.com>
13730 PR tree-optimization/80612
13731 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
13733 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13734 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
13736 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
13737 (movt splitter): Likewise.
13738 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
13739 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
13740 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
13741 block for Thumb-1 with MOVT.
13742 (thumb2_legitimate_address_p): Move code block ...
13743 (can_avoid_literal_pool_for_label_p): ... into this new function.
13744 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
13746 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
13747 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
13748 "M-profile targets with the MOVT instruction".
13750 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13752 * config/arm/arm-builtins.c (arm_init_builtins): Rename
13753 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
13754 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
13756 2017-05-04 Martin Liska <mliska@suse.cz>
13758 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
13759 variable cond_code.
13761 2017-05-04 Richard Biener <rguenther@suse.de>
13763 * tree.c (array_at_struct_end_p): Handle arrays at struct
13764 end with flexarrays more conservatively. Refactor and treat
13765 arrays of arrays or aggregates more strict. Fix
13766 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
13767 * tree.c (array_at_struct_end_p): Adjust prototype.
13768 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13769 * gimple-fold.c (get_range_strlen): Likewise.
13770 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
13772 2017-05-04 Richard Biener <rguenther@suse.de>
13774 PR tree-optimization/31130
13775 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
13777 (supports_overflow_infinity): Likewise.
13778 (is_negative_overflow_infinity): Likewise.
13779 (is_positive_overflow_infinity): Likewise.
13780 (is_overflow_infinity): Likewise.
13781 (stmt_overflow_infinity): Likewise.
13782 (overflow_infinity_range_p): Likewise.
13783 (usable_range_p): Remove as always returning true.
13784 (make_overflow_infinity): Remove.
13785 (negative_overflow_infinity): Likewise.
13786 (positive_overflow_infinity): Likewise.
13787 (avoid_overflow_infinity): Likewise.
13788 (set_value_range): Adjust accordingly.
13789 (set_value_range_to_nonnegative): Likewise, remove now unused
13790 overflow_infinity arg.
13791 (vrp_operand_equal_p): Adjust.
13792 (update_value_range): Likewise.
13793 (range_int_cst_singleton_p): Likewise.
13794 (operand_less_p): Likewise.
13795 (compare_values_warnv): Likewise.
13796 (extract_range_for_var_from_comparison_expr): Likewise.
13797 (vrp_int_const_binop): Likewise.
13798 (zero_nonzero_bits_from_vr): Likewise.
13799 (extract_range_from_multiplicative_op_1): Likewise.
13800 (extract_range_from_binary_expr_1): Likewise.
13801 (extract_range_from_unary_expr): Likewise.
13802 (extract_range_from_comparison): Likewise.
13803 (extract_range_basic): Likewise.
13804 (adjust_range_with_scev): Likewise.
13805 (compare_ranges): Likewise.
13806 (compare_range_with_value): Likewise.
13807 (dump_value_range): Likewise.
13808 (test_for_singularity): Likewise, remove strict_overflow_p parameter
13810 (simplify_cond_using_ranges): Adjust.
13812 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13814 * brig-builtins.def: Added a builtin for class_f64.
13815 * builtin-types.def: Added a builtin type needed by class_f64.
13817 2017-05-03 Jason Merrill <jason@redhat.com>
13819 * timevar.def: Add TV_CONSTEXPR.
13821 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13823 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
13825 2017-05-03 Martin Jambor <mjambor@suse.cz>
13827 * ipa-prop.c (ipa_update_after_lto_read): Removed.
13828 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
13829 * ipa-cp.c (ipcp_propagate_stage): Do not call
13830 ipa_update_after_lto_read.
13831 * ipa-inline.c (ipa_inline): Likewise.
13833 2017-05-03 Martin Jambor <mjambor@suse.cz>
13835 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
13836 tag. Added a default constructor and a destructor.
13837 (ipa_edge_args_sum_t): New class;
13838 (ipa_edge_args_sum): Declare.
13839 (ipa_edge_args_vector): Remove declaration.
13840 (IPA_EDGE_REF): Use ipa_edge_args_sum.
13841 (ipa_free_edge_args_substructures): Remove declaration.
13842 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
13843 (ipa_edge_args_info_available_for_edge_p): Likewise.
13844 * ipa-prop.c (ipa_edge_args_vector): Removed.
13845 (edge_removal_hook_holder): Likewise.
13846 (edge_duplication_hook_holder): Likewise.
13847 (ipa_edge_args_sum): New variable.
13848 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
13849 ipa_edge_args_vector.
13850 (ipa_free_edge_args_substructures): Likewise.
13851 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
13852 ipa_edge_args_vector.
13853 (ipa_edge_removal_hook): Turned into method
13854 ipa_edge_args_sum_t::remove.
13855 (ipa_edge_duplication_hook): Turned into method
13856 ipa_edge_args_sum_t::duplicate.
13857 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
13858 registering edge hooks.
13859 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
13860 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
13861 ipa_edge_args_sum instead of ipa_edge_args_vector.
13862 * ipa-profile.c (ipa_profile): Likewise.
13864 2017-05-03 Martin Jambor <mjambor@suse.cz>
13866 * symbol-summary.h (function_summary): New method exists.
13867 (function_summary::symtab_removal): Deallocate through release.
13868 (call_summary): New class.
13869 (gt_ggc_mx): New overload.
13870 (gt_pch_nx): Likewise.
13871 (gt_pch_nx): Likewise.
13873 2017-05-03 Jeff Law <law@redhat.com>
13875 PR tree-optimization/78496
13876 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
13877 from simplify_cond_using_ranges. Split off code to walk
13878 backwards through casts into ...
13879 (simplify_cond_using_ranges_2): New function.
13880 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
13881 (execute_vrp): After identifying jump threads, call
13882 simplify_cond_using_ranges_2.
13884 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
13887 * ipa-inline.h (inline_summary): Add ctor.
13888 (create_ggc): Do not use ggc_cleared_alloc.
13890 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
13891 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13893 * gcc.c (handle_braces): Support escaping in switch matching
13895 * doc/invoke.texi (Spec Files): Document it.
13896 Remove superfluous @code markup in items.
13898 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13900 * diagnostic-show-locus.c (struct column_range): New struct.
13901 (get_affected_columns): New function.
13902 (get_printed_columns): New function.
13903 (struct correction): New struct.
13904 (correction::ensure_capacity): New function.
13905 (correction::ensure_terminated): New function.
13906 (struct line_corrections): New struct.
13907 (line_corrections::~line_corrections): New dtor.
13908 (line_corrections::add_hint): New function.
13909 (layout::print_trailing_fixits): Reimplement in terms of the new
13911 (selftest::test_overlapped_fixit_printing): New function.
13912 (selftest::diagnostic_show_locus_c_tests): Call it.
13914 2017-05-03 Nathan Sidwell <nathan@acm.org>
13916 Canonicalize canonical type hashing
13917 * tree.h (type_hash_canon_hash): Declare.
13918 * tree.c (type_hash_list, attribute_hash_list): Move into
13919 type_hash_canon_hash.
13920 (build_type_attribute_qual_variant): Break out hash code calc into
13921 type_hash_canon_hash.
13922 (type_hash_canon_hash): New. Generic type hash computation.
13923 (build_range_type_1, build_array_type_1, build_function_type,
13924 build_method_type_directly, build_offset_type, build_complex_type,
13925 make_vector_type): Call it.
13927 2017-05-03 Richard Biener <rguenther@suse.de>
13929 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
13930 When all DRs have unknown misaligned do not always peel
13931 when there is a store but apply the same costing model as if
13932 there were only loads.
13934 2017-05-03 Richard Biener <rguenther@suse.de>
13937 PR tree-optimization/80492
13938 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
13939 compare_base_decls returning dont-know properly.
13941 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
13943 * config/arm/iterators.md (CCSI): New mode iterator.
13944 (arch): New mode attribute.
13945 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
13946 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
13947 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
13948 code iterator for success result mode.
13949 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
13950 the corresponding new insn generators.
13952 2017-05-03 Bin Cheng <bin.cheng@arm.com>
13955 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13956 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13958 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
13960 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
13961 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
13962 (DDR_A): Wrap DDR argument in brackets.
13963 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
13964 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
13965 (DDR_REVERSED_P): Likewise.
13967 2017-05-03 Jakub Jelinek <jakub@redhat.com>
13969 PR tree-optimization/79472
13970 * tree-switch-conversion.c (struct switch_conv_info): Add
13971 contiguous_range and default_case_nonstandard fields.
13972 (collect_switch_conv_info): Compute contiguous_range and
13973 default_case_nonstandard fields, don't clear final_bb if
13974 contiguous_range and only the default case doesn't have the required
13976 (check_all_empty_except_final): Set default_case_nonstandard instead
13977 of failing if contiguous_range and the default case doesn't have empty
13979 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
13980 and only the default case doesn't have the required constants. Skip
13982 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
13983 if default_case_nonstandard.
13984 (build_constructors): Build constant 1 just once. Assert that default
13985 values aren't inserted in between cases if contiguous_range. Skip
13987 (build_arrays): Skip virtual phis.
13988 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
13989 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
13990 Handle virtual phis.
13991 (gen_inbound_check): Handle default_case_nonstandard case.
13992 (process_switch): Adjust check_final_bb caller. Call
13993 gather_default_values with the first non-default case instead of
13994 default case if default_case_nonstandard.
13996 2017-05-02 Nathan Sidwell <nathan@acm.org>
13998 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
13999 check. Fix formatting.
14001 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14003 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14004 errors when comparing specialized and unspecialized times.
14006 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14008 * diagnostic-show-locus.c
14009 (layout::should_print_annotation_line_p): Make private.
14010 (layout::print_annotation_line): Make private.
14011 (layout::annotation_line_showed_range_p): Make private.
14012 (layout::show_ruler): Make private.
14013 (layout::print_source_line): Make private. Pass in line and
14014 line_width, rather than calling location_get_source_line. Drop
14016 (layout::print_leading_fixits): New method.
14017 (layout::print_any_fixits): Rename to...
14018 (layout::print_trailing_fixits): ...this, and make private.
14019 Don't print newline fixits.
14020 (diagnostic_show_locus): Move logic for printing one row into...
14021 (layout::print_line): ...this new function. Move the
14022 location_get_source_line call and error-handling from
14023 print_source_line to here. Call print_leading_fixits, and rename
14024 print_any_fixits to print_trailing_fixits.
14025 (selftest::test_fixit_insert_containing_newline): Update now that
14026 newlines are partially supported.
14027 (selftest::test_fixit_insert_containing_newline_2): New test.
14028 (selftest::test_fixit_replace_containing_newline): Update comments.
14029 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14030 * edit-context.c (class added_line): New class.
14031 (class edited_line): Describe newline handling in comment.
14032 (edited_line::actually_edited_p): New method.
14033 (edited_line::print_content): Delete redundant decl.
14034 (edited_line::m_predecessors): New field.
14035 (edited_file::print_content): Call edited_line::print_content.
14036 (edited_file::print_diff): Update to support newlines.
14037 (edited_file::print_diff_hunk): Likewise.
14038 (edited_file::print_run_of_changed_lines): New function.
14039 (edited_file::print_diff_line): Convert to...
14040 (print_diff_line): ...this.
14041 (edited_file::get_effective_line_count): New function.
14042 (edited_line::edited_line): Initialize new field m_predecessors.
14043 (edited_line::~edited_line): Clean up m_predecessors.
14044 (edited_line::apply_fixit): Handle newlines.
14045 (edited_line::get_effective_line_count): New function.
14046 (edited_line::print_content): New function.
14047 (edited_line::print_diff_lines): New function.
14048 (selftest::test_applying_fixits_insert_containing_newline): New
14050 (selftest::test_applying_fixits_replace_containing_newline): New
14052 (selftest::insert_line): New function.
14053 (selftest::test_applying_fixits_multiple_lines): Add example of
14055 (selftest::edit_context_c_tests): Call the new tests.
14057 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14059 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14060 parameter cand. Update dump information.
14061 (get_computation_cost): Update uses.
14063 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14065 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14066 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14067 (get_computation_at, rewrite_use_address): Update use of
14068 get_computation_aff.
14070 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14072 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14073 (get_computation): Delete.
14074 (get_computation_cost): Implement like get_computation_cost_at.
14075 Use get_computation_at.
14076 (get_computation_cost_at): Delete.
14077 (rewrite_use_nonlinear_expr): Use get_computation_at.
14078 (rewrite_use_compare, remove_unused_ivs): Ditto.
14080 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14082 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14084 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14086 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14087 (ivopts_global_cost_for_size): Rename parameter and update uses.
14088 (iv_ca_recount_cost): Update uses.
14089 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14090 candidates seperately in n_invs and n_cands.
14091 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14093 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14095 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14096 (find_inv_vars_cb): New.
14097 (find_depends): Renamed to ...
14098 (find_inv_vars): ... this.
14099 (add_candidate_1, force_var_cost): Call find_inv_vars.
14100 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14102 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14104 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14105 inv_vars. Add inv_exprs.
14106 (struct iv_cand): Rename depends_on to inv_vars.
14107 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14108 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14109 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14110 (dump_cand): Dump inv_vars.
14111 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14112 (record_invariant, find_depends, add_candidate_1): Ditto.
14113 (set_group_iv_cost, force_var_cost): Ditto.
14114 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14115 (get_computation_cost_at, get_computation_cost): Ditto.
14116 (determine_group_iv_cost_generic): Ditto.
14117 (determine_group_iv_cost_address): Ditto.
14118 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14119 (determine_group_iv_costs): Ditto.
14120 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14121 (iv_ca_set_remove_invariants): Renamed to ...
14122 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14123 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14124 (iv_ca_set_add_invariants): Renamed to ...
14125 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14126 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14127 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14128 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14129 (create_new_ivs): Remove useless dump.
14131 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14133 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14135 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14136 (iv_ca_set_no_cp, create_new_iv): Ditto.
14138 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14140 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14142 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14144 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14145 function tree_check2.
14147 2017-05-02 Martin Liska <mliska@suse.cz>
14149 * doc/gcov.texi: Add missing preposition.
14150 * gcov.c (function_info::function_info): Properly fill up
14151 all member variables.
14153 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14155 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14157 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14159 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14161 2017-05-02 Martin Liska <mliska@suse.cz>
14164 * lto-streamer-in.c (lto_read_tree_1): Remove
14165 LTO_STREAMER_DEBUG.
14166 * lto-streamer.c (struct tree_hash_entry): Likewise.
14167 (struct tree_entry_hasher): Likewise.
14168 (tree_entry_hasher::hash): Likewise.
14169 (tree_entry_hasher::equal): Likewise.
14170 (lto_streamer_init): Likewise.
14171 (lto_orig_address_map): Likewise.
14172 (lto_orig_address_get): Likewise.
14173 (lto_orig_address_remove): Likewise.
14174 * lto-streamer.h: Likewise.
14175 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14176 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14178 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14180 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14181 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14182 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14183 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14184 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14185 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14186 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14187 (mm_maskz_sub_ss): New intrinsics.
14188 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14189 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14190 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14191 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14192 (__builtin_ia32_subss_mask_round): New builtins.
14193 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14194 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14195 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14197 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14198 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14200 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14203 2017-05-02 Martin Jambor <mjambor@suse.cz>
14205 PR tree-optimization/78687
14206 * tree-sra.c (access): New field parent.
14207 (process_subtree_disqualification): New function.
14208 (disqualify_candidate): Call it.
14209 (build_accesses_from_assign): Reset write flag if creating an
14211 (build_access_subtree): Fill in parent field and also prpagate
14212 down grp_write flag.
14213 (create_artificial_child_access): New parameter set_grp_write, set
14214 grp_write to its value.
14215 (propagate_subaccesses_across_link): Also propagate grp_write flag
14217 (propagate_all_subaccesses): Push the closest parent back to work
14218 queue if add_access_to_work_queue returned true.
14220 2017-05-02 Richard Biener <rguenther@suse.de>
14222 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14223 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14224 -fstrict-overflow documentation.
14225 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14226 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14227 flag_strict_overflow.
14228 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14229 * lto-opts.c (lto_write_options): Do not stream it.
14230 * lto-wrapper.c (merge_and_complain): Do not handle it.
14231 * opts.c (default_options_table): Do not set -fstrict-overflow.
14232 (finish_options): Likewise do not clear it when sanitizing.
14233 * simplify-rtx.c (simplify_const_relational_operation): Do not
14234 test flag_strict_overflow.
14236 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14238 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14239 using enabled attribute.
14240 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14241 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14242 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14243 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14244 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14245 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14246 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14247 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14248 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14249 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14251 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14253 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14255 2017-05-02 Richard Biener <rguenther@suse.de>
14257 PR tree-optimization/80591
14259 2017-04-10 Richard Biener <rguenther@suse.de>
14261 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14264 2017-05-02 Richard Biener <rguenther@suse.de>
14266 PR tree-optimization/80549
14267 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14268 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14269 headers if they do not have a preheader.
14271 2017-05-02 Martin Liska <mliska@suse.cz>
14274 * common.opt: Fix typo.
14275 * doc/invoke.texi: Likewise.
14277 2017-05-01 Jan Beulich <jbeulich@suse.com>
14279 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14280 swapping, add (x,x,m,x,n) alternative.
14282 2017-05-01 Nathan Sidwell <nathan@acm.org>
14284 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14285 unnecessary unadjusted_alignment check.
14287 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14290 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14292 * gimplify.c (gimplify_cilk_detach): New function.
14293 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14294 * tree-core.h: Document EXPR_CILK_SPAWN.
14295 * tree.h (EXPR_CILK_SPAWN): Define.
14297 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14299 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14300 to use new fixit_hint representation, using the "replace" logic.
14301 (get_line_span_for_fixit_hint): Likewise.
14302 (layout::print_any_fixits): Likewise.
14303 (selftest::test_one_liner_many_fixits): Rename to...
14304 (selftest::test_one_liner_many_fixits_1): ...this, and update
14305 comment and expected output to reflect that the multiple fix-it
14306 hints are now consolidated into one insertion.
14307 (selftest::test_one_liner_many_fixits_2): New test.
14308 (selftest::test_diagnostic_show_locus_one_liner): Update for
14310 (selftest::test_fixit_consolidation): Update for fix-it API
14312 * diagnostic.c (print_parseable_fixits): Likewise.
14313 * edit-context.c (edited_line::m_line_events): Convert from
14314 auto_vec <line_event *> to auto_vec <line_event>.
14315 (class line_event): Convert from abstract base class to a concrete
14316 class, taking over the role of replace_event.
14317 (class insert_event): Delete.
14318 (class replace_event): Rename to class line_event. Convert to
14320 (edit_context::add_fixits): Reimplement.
14321 (edit_context::apply_insert): Delete.
14322 (edit_context::apply_replace): Rename to...
14323 (edit_context::apply_fixit): ...this. Convert to half-open range.
14324 (edited_file::apply_insert): Delete.
14325 (edited_file::apply_replace): Rename to...
14326 (edited_file::apply_fixit): ...this.
14327 (edited_line::~edited_line): Drop deletion of events.
14328 (edited_line::apply_insert): Delete.
14329 (edited_line::apply_replace): Rename to...
14330 (edited_line::apply_fixit): ...this. Convert to half-open range.
14331 Update for change to type of m_line_events.
14332 * edit-context.h (edit_context::apply_insert): Delete.
14333 (edit_context::apply_replace): Rename to...
14334 (edit_context::apply_fixit): ...this.
14336 2017-05-01 Martin Sebor <msebor@redhat.com>
14338 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14341 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14344 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14345 __get_cpuid_max returns 0.
14346 (__get_cpuid_count): Ditto.
14348 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14350 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14351 replacement expression is another instance of one of its arguments.
14353 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14356 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14357 check for stack push/pop autoinc.
14358 * config/i386/i386.c (ix86_agi_dependent): Return false
14359 if the only reason why modified_in_p returned true is that
14360 addr is SP based and set_insn is a push or pop.
14362 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14364 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14367 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14370 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14371 (account_size_time): Use two predicates - exec_pred and
14373 (evaluate_conditions_for_known_args): Compute both clause and
14375 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14376 (inline_summary_t::duplicate): Update.
14377 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14379 (compute_inline_parameters): Likewise.
14380 (estimate_edge_size_and_time): Update caluclation of time.
14381 (estimate_node_size_and_time): Compute both time and nonspecialized
14383 (estimate_ipcp_clone_size_and_time): Update.
14384 (inline_merge_summary): Update.
14385 (do_estimate_edge_time): Update.
14386 (do_estimate_edge_size): Update.
14387 (do_estimate_edge_hints): Update.
14388 (inline_read_section, inline_write_summary): Stream both new predicates.
14389 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14391 (compute_inlined_call_time): Cleanup.
14392 (big_speedup_p): Update.
14393 (edge_badness): Update.
14394 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14395 (size_time_entry): Replace predicate by exec_predicate and
14396 nonconst_predicate.
14397 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14398 (estimate_edge_time): Return also nonspec_time.
14399 (reset_edge_growth_cache): Update.
14401 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14403 PR rtl-optimization/80491
14404 * ifcvt.c (noce_process_if_block): When looking for x setter
14405 with missing else_bb, don't check only the insn right before
14406 cond_earliest, but look for the last insn that x is modified in
14407 within the same bb.
14409 PR rtl-optimization/80491
14410 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14412 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14414 PR tree-optimization/80487
14415 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14417 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14419 PR tree-optimization/79697
14420 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14421 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14422 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14424 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14425 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14427 2017-04-28 Martin Sebor <msebor@redhat.com>
14429 PR tree-optimization/80523
14430 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14431 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14433 (maybe_warn, format_directive, parse_directive): Use new functions.
14434 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14436 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14438 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14440 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14442 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14443 target_header_dir): Set correctly.
14444 * configure: Regenerated.
14445 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14446 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14447 instead of SYSTEM_HEADER_DIR.
14449 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14451 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14452 (estimate_local_effects): Likewise.
14453 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14454 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14455 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14456 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14457 do_estimate_edge_time, estimate_edge_time): Likewise.
14458 * ipa-inline-analysis.c (estimate_node_size_and_time,
14459 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14460 (estimate_time_after_inlining): Remove.
14462 2017-04-28 Martin Liska <mliska@suse.cz>
14464 * doc/gcov.texi: Enhance documentation of gcov.
14466 2017-04-28 Martin Liska <mliska@suse.cz>
14468 * doc/gcov.texi: Sort options in alphabetic order.
14469 * doc/gcov-dump.texi: Likewise.
14470 * doc/gcov-tool.texi: Likewise.
14471 * gcov.c (print_usage): Likewise.
14472 * gcov-dump.c (print_usage): Likewise.
14473 * gcov-tool.c (print_merge_usage_message): Likewise.
14474 (print_rewrite_usage_message): Likewise.
14475 (print_overlap_usage_message): Likewise.
14477 2017-04-28 Martin Liska <mliska@suse.cz>
14479 PR gcov-profile/53915
14480 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14482 2017-04-28 Martin Liska <mliska@suse.cz>
14484 PR gcov-profile/79891
14485 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14486 is marked by compiler as living on a line.
14487 (get_cycles_count): Remove usage of the union.
14488 (output_intermediate_file): Likewise.
14489 (find_source): Fix GNU coding style.
14490 (accumulate_line_counts): Remove old non-all block mode.
14491 (output_lines): Remove usage of the union.
14492 * profile.c (output_location): Include all BBs, even if
14493 belonging to a same line (and file) as a previous BB.
14495 2017-04-28 Martin Liska <mliska@suse.cz>
14497 * gcov.c (process_args): Handle new argument 'w'.
14498 (read_graph_file): Assign ID to BBs.
14499 (output_branch_count): Display BB # if verbose flag is set.
14500 (output_lines): Likewise for arcs.
14501 (print_usage): Add '--verbose' option help.
14502 * doc/gcov.texi: Document --verbose (-w) option.
14504 2017-04-28 Martin Liska <mliska@suse.cz>
14506 * gcov.c (struct block_location_info): New struct.
14507 (process_file): Fill up the new structure.
14508 (read_graph_file): Replace usage of encoding by the newly added
14510 (add_line_counts): Likewise.
14511 (accumulate_line_counts): Remove usage of the union.
14512 (function_info::function_info): New function.
14513 (function_info::~function_info): Likewise.
14514 (process_file): Call delete instead of release_function.
14515 (release_function): Release the function.
14516 (release_structures): Call delete instead of release_function.
14517 (solve_flow_graph): Replace usage of num_blocks.
14518 (find_exception_blocks): Likewise.
14519 (output_lines): Fix GNU coding style.
14521 2017-04-28 Martin Liska <mliska@suse.cz>
14524 * coverage.c (coverage_remove_note_file): New function.
14525 * coverage.h: Declare the function.
14526 * toplev.c (finalize): Clean if an error has been seen.
14528 2017-04-28 Martin Liska <mliska@suse.cz>
14530 PR gcov-profile/80031
14531 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14532 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14533 * gcov.c (read_graph_file): Read just number of blocks.
14534 * profile.c (branch_prob): Do not stream 0 flags per a basic
14537 2017-04-28 Martin Liska <mliska@suse.cz>
14539 * gcov-dump.c (tag_*): Add new argument to declarations.
14540 (dump_gcov_file): Likewise.
14541 (tag_blocks): Add and use new argument depth.
14542 (tag_arcs): Likewise.
14543 (tag_lines): Likewise.
14544 (tag_counters): Likewise.
14545 (tag_summary): Likewise.
14546 (dump_working_sets): Use depth to do a proper indentation.
14548 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14551 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14552 * symtab.c (symtab_node::debug_symtab): Move definition here.
14554 2017-04-28 Richard Biener <rguenther@suse.de>
14556 * lto-streamer.h (LTO_major_version): Bump to 7.
14558 2017-04-28 Richard Biener <rguenther@suse.de>
14560 * tree-vrp.c (assert_info): New struct.
14561 (add_assert_info): New helper.
14562 (register_edge_assert_for_2): Refactor to add asserts to a vector
14564 (register_edge_assert_for_1): Likewise.
14565 (register_edge_assert_for): Likewise.
14566 (finish_register_edge_assert_for): New helper actually registering
14567 asserts where live on edge.
14568 (find_conditional_asserts): Adjust.
14569 (find_switch_asserts): Likewise.
14570 (evrp_dom_walker::try_find_new_range): Generalize.
14571 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14573 2017-04-27 Marek Polacek <polacek@redhat.com>
14576 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14577 arg10 and arg11 to itype.
14579 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14581 * doc/extend.texi (Object Size Checking): Improve grammar.
14583 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14586 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14587 that the logic for permitting reciprocal estimates matches that
14590 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14593 * tree.c (type_cache_hasher::equal): Only compare
14594 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14595 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14596 non-aggregate element types.
14597 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14598 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14600 2017-04-27 Richard Biener <rguenther@suse.de>
14602 PR middle-end/80533
14603 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14604 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14605 keeping a reference to a trailing array.
14607 2017-04-27 Richard Biener <rguenther@suse.de>
14609 PR middle-end/80539
14610 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
14611 being in loop-closed SSA form conservatively.
14612 (chrec_fold_multiply_poly_poly): Likewise.
14614 2017-04-27 Tamar Christina <tamar.christina@arm.com>
14616 PR middle-end/79665
14617 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
14618 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
14620 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14623 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
14624 (aarch64_function_arg_alignment): Return unsigned int again, but still
14625 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
14626 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14627 Don't emit -Wpsabi note.
14628 (aarch64_function_arg_boundary): Likewise.
14629 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14632 2017-04-26 Nathan Sidwell <nathan@acm.org>
14634 * tree.h (crc32_unsigned_n): Declare.
14635 (crc32_unsigned, crc32_unsigned): Make inline.
14636 * tree.c (crc32_unsigned_bits): Replace with ...
14637 (crc32_unsigned_n): ... this.
14638 (crc32_unsigned, crc32_byte): Remove.
14639 (crc32_string): Remove unnecessary braces.
14641 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14643 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
14644 * ipa-inline-analysis.c (MAX_TIME): Remove.
14645 (account_size_time): Use sreal for time.
14646 (dump_inline_summary): Update.
14647 (estimate_function_body_sizes): Update.
14648 (estimate_edge_size_and_time): Update.
14649 (estimate_calls_size_and_time): Update.
14650 (estimate_node_size_and_time): Update.
14651 (inline_merge_summary): Update.
14652 (inline_update_overall_summary): Update.
14653 (estimate_time_after_inlining): Update.
14654 (inline_read_section): Update.
14655 (inline_write_summary): Update.
14656 * ipa-inline.c (compute_uninlined_call_time): Update.
14657 (compute_inlined_call_time): Update.
14658 (recursive_inlining): Update.
14659 (inline_small_functions): Update.
14660 (dump_overall_stats): Update.
14661 * ipa-inline.h: Include sreal.h.
14662 (size_time_entry): Turn time to sreal.
14663 (inline_summary): Turn self_time nad time to sreal.
14665 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14667 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
14669 (sreal::stream_out, sreal::stream_in): New.
14670 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
14672 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14674 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
14677 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
14680 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
14682 (dimode_scalar_chain::compute_convert_gain): Ditto.
14683 (dimode_scalar_chain::make_vector_copies): Ditto.
14684 (dimode_scalar_chain::convert_reg): Ditto.
14685 (dimode_scalar_chain::convert_insn): Ditto.
14686 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
14687 (VI248_AVX512BW_1): New mode iterator.
14688 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
14689 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
14692 2017-04-25 Martin Sebor <msebor@redhat.com>
14694 PR tree-optimization/80497
14695 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
14696 constants are representable in HOST_WIDE_INT.
14697 (parse_directive): Ditto.
14699 2017-04-25 Martin Sebor <msebor@redhat.com>
14702 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
14703 (new_zero_array): Adjust signature.
14704 (dom_info::dom_init): Used unsigned rather that size_t.
14705 (dom_info::dom_info): Same.
14707 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14708 Jakub Jelinek <jakub@redhat.com>
14711 * config/arm/arm.c: Include gimple.h.
14712 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14713 returns negative, increment ncrn only if it returned positive.
14714 (arm_needs_doubleword_align): Return int instead of bool,
14715 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
14716 members, but if there is any such non-FIELD_DECL
14717 > PARM_BOUNDARY aligned decl, return -1 instead of false.
14718 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14719 returns negative, increment nregs only if it returned positive.
14720 (arm_setup_incoming_varargs): Likewise.
14721 (arm_function_arg_boundary): Emit -Wpsabi note if
14722 arm_needs_doubleword_align returns negative, return
14723 DOUBLEWORD_ALIGNMENT only if it returned positive.
14725 2017-04-25 Marek Polacek <polacek@redhat.com>
14728 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
14729 first argument to type.
14731 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
14734 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
14735 type checks to test for compatibility instead of equality.
14737 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14738 Jakub Jelinek <jakub@redhat.com>
14741 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
14743 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
14744 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
14745 the alignment computation, but return their maximum in warn_alignment.
14746 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14747 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
14749 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
14750 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14753 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14755 * config/arc/simdext.md (dmpyh): Fix typo.
14757 2017-04-25 Richard Biener <rguenther@suse.de>
14759 PR tree-optimization/80492
14760 * alias.c (compare_base_decls): Handle registers with asm
14761 specification conservatively.
14762 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14763 compare_base_decls returning dont-know properly.
14765 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14767 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
14768 (legitimate_offset_address_p): New function.
14769 (arc_legitimate_address_p): Use above function.
14771 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14773 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
14775 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14777 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
14778 ACCH registers whenever they are available.
14780 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14782 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
14783 double regs fix when not used.
14785 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14787 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
14789 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
14790 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
14792 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14794 * config/arc/arc.c (arc_output_addsi): Check for h-register class
14795 when emitting short ADD instructions.
14797 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14799 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
14801 (cmpsi_cc_c_insn): Likewise.
14802 (cbranchsi4_scratch): Compute proper instruction length using
14803 compact_hreg_operand.
14804 * config/arc/predicates.md (compact_hreg_operand): New predicate.
14806 2017-04-25 Richard Biener <rguenther@suse.de>
14808 PR middle-end/80509
14809 * passes.c (pass_manager::pass_manager): Initialize
14810 m_name_to_pass_map.
14812 2017-04-25 Richard Biener <rguenther@suse.de>
14814 PR tree-optimization/79201
14815 * tree-ssa-sink.c (statement_sink_location): Handle calls.
14817 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14820 * config/s390/vector.md: Split MEM->GPR vector moves for
14821 non-s_operand addresses.
14823 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14826 * config/s390/predicates.md (reload_const_wide_int_operand): New
14828 * config/s390/s390.md ("movti"): Remove d/P alternative.
14829 ("movti_bigconst"): New pattern definition.
14831 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14834 * s390-protos.h (s390_expand_cs_hqi): Removed.
14835 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
14836 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
14837 modes as well as CCZ1mode and CCZmode.
14838 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
14839 signature of s390_emit_compare_and_swap.
14840 (s390_expand_cs_hqi): Likewise, make static.
14841 (s390_expand_cs_tdsi): Generate an explicit compare before trying
14842 compare-and-swap, in some cases.
14843 (s390_expand_cs): Wrapper function.
14844 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
14846 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
14847 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
14848 patterns for small and large integers. Forbid symref memory operands.
14849 Move expander to s390.c. Require cc register.
14850 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
14851 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
14852 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
14853 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
14854 symref memory operands. Remove CC mode and call s390_match_ccmode
14856 ("atomic_exchange<mode>"): Allow and implement all integer modes.
14858 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14860 * config/s390/s390.md (define_peephole2): New peephole to help
14861 combining the load-and-test pattern with volatile memory.
14863 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14865 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
14866 with CCZmode for TARGET_Z196.
14868 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14870 PR rtl-optimization/80501
14871 * combine.c (make_compound_operation_int): Set subreg_code to SET
14872 even for AND with mask of the sign bit of mode.
14874 PR rtl-optimization/80500
14875 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
14876 sum's initial value.
14878 2017-04-25 Julian Brown <julian@codesourcery.com>
14879 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14881 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
14883 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
14885 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
14887 2017-04-25 Julian Brown <julian@codesourcery.com>
14888 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14890 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
14891 (thunderx2t99_sha): New Reservation.
14893 2017-04-25 Julian Brown <julian@codesourcery.com>
14894 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14896 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
14897 type for 1-element load.
14899 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
14901 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
14903 2017-04-24 Martin Jambor <mjambor@suse.cz>
14905 PR tree-optimization/80293
14906 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
14907 char arrays not totally scalarizable if it is false.
14908 (analyze_all_variable_accesses): Pass correct value in the new
14909 parameter. Add a statistics counter.
14911 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
14913 PR middle-end/79931
14914 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
14916 2017-04-24 Richard Biener <rguenther@suse.de>
14918 PR tree-optimization/80494
14919 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
14920 out for complex types.
14922 2017-04-24 Richard Biener <rguenther@suse.de>
14924 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
14925 * tree-ssa-sccvn.c (print_scc): Print SCC size.
14926 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
14927 (DFS): Adjust and never fail.
14928 (sccvn_dom_walker::fail): Remove.
14929 (sccvn_dom_walker::before_dom_children): Adjust.
14930 (run_scc_vn): Likewise and never fail.
14931 * tree-ssa-pre.c (pass_pre::execute): Adjust.
14932 (pass_fre::execute): Likewise.
14934 2017-04-24 Richard Biener <rguenther@suse.de>
14936 PR tree-optimization/79725
14937 * tree-ssa-sink.c (statement_sink_location): Return whether
14938 failure reason was zero uses. Move that check later.
14939 (sink_code_in_bb): Deal with zero uses by removing the stmt
14942 2017-04-24 Richard Biener <rguenther@suse.de>
14945 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
14946 pointer-based references.
14948 2017-04-24 Richard Biener <rguenther@suse.de>
14951 * pass_manager.h (pass_manager::operator new): Remove.
14952 (pass_manager::operator delete): Likewise.
14953 * passes.c (pass_manager::operator new): Remove.
14954 (pass_manager::operator delete): Likewise.
14955 (pass_manager::pass_manager): Zero individual pass members.
14957 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
14960 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
14961 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
14962 Check "XEXP (src, 1)" operand here.
14963 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
14964 Check "XEXP (src, 1)" operand here.
14965 (dimode_scalar_chain::make_vector_copies): Detect count register
14966 of a shift instruction. Zero extend count register from QImode
14967 to DImode to satisfy vector shift pattern count operand predicate.
14968 Substitute vector shift count operand with a DImode copy.
14969 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
14972 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
14974 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
14975 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
14976 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
14977 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
14978 (UNSPEC_NOREX_MEM): Remove definition.
14980 2017-04-21 Richard Biener <rguenther@suse.de>
14982 PR tree-optimization/79547
14983 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14984 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
14985 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
14986 without any constraints.
14988 2017-04-21 Richard Biener <rguenther@suse.de>
14990 PR tree-optimization/78847
14991 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
14993 2017-04-21 Richard Biener <rguenther@suse.de>
14995 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
14996 (build_distinct_type_copy): Likewise.
14997 (build_variant_type_copy): Likewise.
14998 * tree.c (build_qualified_type): Pass down mem-stat info.
14999 (build_distinct_type_copy): Likewise.
15000 (build_variant_type_copy): Likewise.
15002 2017-04-21 Richard Biener <rguenther@suse.de>
15004 PR tree-optimization/80237
15005 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15007 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15008 for a simplified result.
15010 2016-04-21 Richard Biener <rguenther@suse.de>
15012 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15013 sth as strict as a simple_iv but a chrec without symbols and an
15014 operand defined in the loop we are peeling (and not some subloop).
15015 (propagate_constants_for_unrolling): Propagate all constants.
15017 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15020 * config/i386/i386.c (print_reg): Remove assert for disalowed
15021 regno values, call output_operand_lossage instead.
15023 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15026 * config/i386/constraints.md (Yc): New register constraint.
15027 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15028 Use Yc constraint for alternative 2 of operand 0. Remove
15029 preferred_for_speed attribute.
15031 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15033 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15034 lastprivate clauses in SIMT case.
15036 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15038 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15040 2017-04-20 Richard Biener <rguenther@suse.de>
15042 PR tree-optimization/57796
15043 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15044 as N scalar stores.
15045 (vect_model_load_cost): Cost gathers as N scalar loads.
15047 2017-04-20 Richard Biener <rguenther@suse.de>
15049 * ggc-page.c (ggc_allocated_p): Rename to ...
15050 (safe_lookup_page_table_entry): ... this and return the lookup
15052 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15054 2017-04-20 Richard Biener <rguenther@suse.de>
15056 PR tree-optimization/80453
15057 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15058 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15059 from the conditions.
15060 (vn_phi_eq): Pass them down.
15061 (vn_phi_lookup): Record them.
15062 (vn_phi_insert): Likewise.
15064 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15066 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15067 uninitialized variable warning to avoid buffer overrun.
15069 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15072 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15073 is suppressed for '{ 0 }' in C.
15075 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15077 * BASE-VER: Set to 8.0.0.
15079 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15081 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15082 priority .init_array and .fini_array section with SECTION_NOTYPE
15085 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15087 PR middle-end/80423
15088 * tree.h (build_array_type): Add typeless_storage default argument.
15089 * tree.c (type_cache_hasher::equal): Also compare
15090 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15091 (build_array_type): Add typeless_storage argument, set
15092 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15094 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15095 (build_array_type): Likewise. Add typeless_storage argument.
15097 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15098 Jakub Jelinek <jakub@redhat.com>
15100 PR tree-optimization/80426
15101 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15102 operation on symbolic operands, also compute the overflow for the
15103 invariant part when the operation degenerates into a negation.
15105 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15108 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15109 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15112 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15114 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15117 * config/avr/avr.c (tree.h): Include it.
15118 (cgraph.h): Include it.
15119 (avr_encode_section_info): Don't warn for uninitialized progmem
15120 variable if it's just an alias.
15122 2017-04-19 Richard Biener <rguenther@suse.de>
15125 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15126 when needed by AutoPGO.
15128 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15131 * doc/lto.texi: Remove an extra 'that'.
15133 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15135 PR rtl-optimization/80429
15136 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15137 are only used in debug insns.
15139 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15140 Vladimir Makarov <vmakarov@redhat.com>
15142 * config/sparc/predicates.md (input_operand): Add comment. Return
15143 true for any memory operand when LRA is in progress.
15144 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15146 2017-04-18 Jeff Law <law@redhat.com>
15149 * mips.md ({return,simple_return}_internal): Do not overwrite
15152 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15154 PR tree-optimization/80443
15155 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15156 instead of adding 1, subtract -1 and similarly instead of subtracting
15159 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15161 PR rtl-optimization/80357
15162 * haifa-sched.c (tmp_bitmap): New variable.
15163 (model_recompute): Handle duplicate use records.
15164 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15165 (free_global_sched_pressure_data): Free it.
15167 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15170 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15171 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15172 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15173 instead of SYSTEM_HEADER_DIR.
15175 2017-04-18 Jeff Law <law@redhat.com>
15177 PR middle-end/80422
15178 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15179 predecessors after walking up the insn chain.
15181 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15184 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15185 sizetype type into debug info.
15187 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15190 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15191 unneeded test for TARGET_UPPER_REGS_SF.
15192 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15194 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15197 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15198 instead of gsi_after_labels.
15200 2017-04-18 Jeff Law <law@redhat.com>
15202 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15206 2017-04-13 Jeff Law <law@redhat.com>
15207 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15208 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15210 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15213 * config/avr/avr.c (intl.h): Include it.
15214 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15216 2017-04-18 Martin Liska <mliska@suse.cz>
15218 PR gcov-profile/78783
15219 * gcov-tool.c (gcov_output_files): Validate that destination
15220 file is either removed by the tool or by a user.
15222 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15223 Guy Benyei <guybe@mellanox.com>
15225 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15226 block, and do not negate it, the stored id is already negative.
15228 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15230 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15232 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15235 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15236 masks of options that should be turned off if the VSX vector
15237 options are turned off.
15238 (OTHER_P8_VECTOR_MASKS): Likewise.
15239 (OTHER_VSX_VECTOR_MASKS): Likewise.
15240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15241 rs6000_disable_incompatible_switches to validate no type switches
15243 (rs6000_incompatible_switch): New function to disallow turning on
15244 other vector options if -mno-vsx, -mno-power8-vector, or
15245 -mno-power9-vector are specified.
15247 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15249 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15251 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15253 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15254 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15255 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15256 (ARG_POINTER_CFA_OFFSET): Likewise.
15258 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15260 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15261 conditions to take advantage of various optimizations.
15263 2017-04-13 Jeff Law <law@redhat.com>
15265 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15266 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15267 (zero_extendsidi2_dext): Likewise.
15269 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15272 * fold-const.c (fold_ternary_loc): Revert
15273 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15276 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15278 PR rtl-optimization/80343
15279 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15280 new scratch pseudo.
15282 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15285 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15286 to ubsan_encode_value.
15288 2017-04-13 Jeff Law <law@redhat.com>
15290 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15291 appearing in DEBUG_INSNs.
15293 2017-04-13 Martin Liska <mliska@suse.cz>
15295 PR gcov-profile/80413
15296 * gcov-io.c (gcov_write_string): Copy to buffer just when
15297 allocated size is greater than zero.
15299 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15302 * dwarf2out.c (decls_for_scope): Ignore declarations of
15303 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15305 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15308 * ipa-visibility.c (non_local_p): Fix typos.
15309 (localize_node): When localizing symbol in same comdat group,
15310 dissolve the group only when we know external symbols are going
15312 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15314 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15316 PR tree-optimization/79390
15317 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15318 order does not result in usable sequence, retry with reversed operand
15324 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15325 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15326 op0 instead of fold_convert_loc (loc, type, arg0).
15328 2017-04-12 Jeff Law <law@redhat.com>
15330 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15331 has a delay slot in the generated code.
15333 * config/cris/cris.md (cris_preferred_reload_class): Return
15334 GENNONACR_REGS rather than GENERAL_REGS.
15336 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15339 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15340 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15341 signedness of the result type.
15343 2017-04-12 Richard Biener <rguenther@suse.de>
15344 Jeff Law <law@redhat.com>
15346 PR tree-optimization/80359
15347 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15348 trim stores to TARGET_MEM_REFs.
15350 2017-04-12 Richard Biener <rguenther@suse.de>
15352 PR tree-optimization/79390
15353 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15354 threading case even more.
15356 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15359 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15360 for quad_address_p for TImode, instead of just not indexed_address.
15362 2017-04-12 Richard Biener <rguenther@suse.de>
15363 Bernd Edlinger <bernd.edlinger@hotmail.de>
15365 PR middle-end/79671
15366 * alias.c (component_uses_parent_alias_set_from): Handle
15367 TYPE_TYPELESS_STORAGE.
15368 (get_alias_set): Likewise.
15369 * tree-core.h (tree_type_common): Add typeless_storage flag.
15370 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15371 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15372 for types containing members with TYPE_TYPELESS_STORAGE.
15373 (place_field): Likewise.
15374 (layout_type): Likewise for ARRAY_TYPE.
15375 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15376 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15377 TYPE_TYPELESS_STORAGE.
15378 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15380 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15383 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15384 first argument to type.
15386 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15390 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15391 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15392 (rs6000_expand_binop_builtin): Likewise.
15393 (rs6000_expand_ternop_builtin): Likewise; also add missing
15394 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15395 vshasigma built-ins.
15396 * doc/extend.texi: Document that vec_xxpermdi's third argument
15397 must be a constant.
15399 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15401 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15402 Use shift_const cost parameter when calculating gain of STV shifts.
15404 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15406 PR rtl-optimization/70478
15407 * lra-constraints.c (process_alt_operands): Check memory for
15408 disfavoring memory insn operand.
15410 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15412 PR middle-end/80100
15413 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15414 left shift in unsigned HOST_WIDE_INT type.
15416 PR rtl-optimization/80385
15417 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15418 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15421 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15422 if they have any depend clauses.
15424 2017-04-11 Martin Liska <mliska@suse.cz>
15427 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15428 * ipa-split.c (split_function): Create a local comdat symbol
15429 if caller is in a comdat group.
15431 2017-04-11 Martin Liska <mliska@suse.cz>
15434 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15437 2017-04-11 Martin Sebor <msebor@redhat.com>
15439 PR middle-end/80364
15440 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15441 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15443 (directive::set_width, directive::set_precision, format_character):
15445 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15448 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15451 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15452 conflict, set target->arch_name instead of target->cpu_name.
15454 2017-04-11 Richard Biener <rguenther@suse.de>
15456 PR tree-optimization/80374
15457 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15458 build_zero_cst, remove fold_convertible_p check again.
15460 2017-04-11 Martin Liska <mliska@suse.cz>
15463 * ubsan.c (instrument_object_size): Do not instrument register
15466 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15469 * config/i386/i386-builtin-types.def
15470 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15471 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15472 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15473 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15474 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15475 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15476 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15477 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15478 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15479 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15480 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15481 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15482 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15483 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15484 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15485 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15486 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15487 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15488 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15489 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15490 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15491 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15492 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15493 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15494 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15495 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15496 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15497 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15498 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15499 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15500 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15501 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15502 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15503 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15504 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15505 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15506 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15507 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15508 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15509 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15510 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15511 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15512 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15513 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15514 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15515 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15517 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15518 flag to second_arg_count, handle 4 argument function type _COUNT
15519 aliases, handle second_arg_count on second argument rather than last.
15521 2017-04-10 Jeff Law <law@redhat.com>
15523 PR tree-optimization/80374
15524 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15525 record anything if we can not convert integer_zero_node to the
15528 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15531 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15532 Enhance special handling given to the TARGET_P9_MINMAX option in
15533 relation to certain other options.
15535 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15537 PR tree-optimization/80153
15538 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15539 remove POINTER_PLUS_EXPR's base part directly, rather than through
15542 2017-04-10 Richard Biener <rguenther@suse.de>
15543 Bin Cheng <bin.cheng@arm.com>
15545 PR tree-optimization/80153
15546 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15547 the first element of pointer type aff_tree. Build result expr in
15549 (add_elt_to_tree): Convert to type unconditionally. Remove other
15550 fold_convert calls.
15551 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15552 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15554 2017-04-10 Richard Biener <rguenther@suse.de>
15556 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15559 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15561 PR rtl-optimization/70478
15562 * lra-constraints.c (curr_small_class_check): New.
15563 (update_and_check_small_class_inputs): New.
15564 (process_alt_operands): Update curr_small_class_check. Disfavor
15565 alternative insn memory operands. Check available regs for small
15568 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15571 * config/mips/mips.opt (-mvirt): Update description.
15572 * doc/invoke.texi (-mvirt): Likewise.
15574 2017-04-10 Richard Biener <rguenther@suse.de>
15576 PR middle-end/80362
15577 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15578 looking for NEGATE_EXPR in -A / -B to A / B folding.
15580 2017-04-10 Martin Liska <mliska@suse.cz>
15582 PR gcov-profile/80224
15583 * gcov.c (print_usage): Fix usage string.
15584 (get_gcov_intermediate_filename): Remove.
15585 (output_gcov_file): Use both for normal and intermediate format.
15586 (generate_results): Do not initialize special file for
15587 intermediate format.
15589 2017-04-10 Richard Biener <rguenther@suse.de>
15591 PR tree-optimization/80304
15592 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15595 2017-04-10 Nathan Sidwell <nathan@acm.org>
15598 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15599 (rs6000_init_builtins): Use it.
15601 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15603 * config/arm/arm.md (<mrc>): Add mode to SET source.
15604 (<mrrc>): Likewise.
15606 2017-04-10 Richard Biener <rguenther@suse.de>
15608 PR middle-end/80344
15609 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
15611 2017-04-10 Jakub Jelinek <jakub@redhat.com>
15614 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
15615 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
15616 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
15617 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
15618 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
15619 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
15620 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
15621 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
15622 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
15623 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
15624 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
15625 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
15626 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
15627 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
15628 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
15629 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
15630 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
15631 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
15632 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
15633 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
15634 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
15635 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
15636 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
15638 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
15640 PR rtl-optimization/70478
15641 * lra-constraints.c: Reverse the last patch.
15643 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
15645 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
15646 Add comment for WCHAR_T.
15648 2017-04-08 Martin Liska <mliska@suse.cz>
15651 2017-04-07 Martin Liska <mliska@suse.cz>
15654 * ipa-split.c (split_function): Add function part to a same comdat
15657 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15660 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
15662 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
15664 * rs6000/rs6000.c (vec_load_pendulum): Rename...
15665 (vec_pairing): ...to this.
15666 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
15667 (rs6000_sched_init): Adjust for name change.
15668 (struct rs6000_sched_context): Likewise.
15669 (rs6000_init_sched_context): Likewise.
15670 (rs6000_set_sched_context): Likewise.
15672 2017-04-07 Jakub Jelinek <jakub@redhat.com>
15678 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
15680 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
15681 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
15682 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
15684 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
15686 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
15688 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15690 PR rtl-optimization/70703
15691 * ira-color.c (update_conflict_hard_regno_costs): Use
15692 int64_t instead of HOST_WIDE_INT.
15694 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15696 PR rtl-optimization/70478
15697 * lra-constraints.c (process_alt_operands): Disfavor alternative
15698 insn memory operands.
15700 2017-04-07 Jeff Law <law@redhat.com>
15702 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
15703 CALL and NOTE_INSN_CALL_ARG_LOCATION.
15705 2017-04-07 Martin Liska <mliska@suse.cz>
15708 * config/aarch64/aarch64.c (aarch64_process_target_attr):
15709 Show error message instead of an ICE.
15711 2017-04-07 Martin Liska <mliska@suse.cz>
15714 * ipa-split.c (split_function): Add function part to a same comdat
15717 2017-04-07 Richard Biener <rguenther@suse.de>
15719 PR middle-end/80341
15720 * tree.c (get_unwidened): Also handle ! for_type case for
15722 * convert.c (do_narrow): Split out from ...
15723 (convert_to_integer_1): ... here. Do not pass final truncation
15724 type to get_unwidened for TRUNC_DIV_EXPR.
15726 2017-04-07 Richard Biener <rguenther@suse.de>
15728 * tree-affine.c (wide_int_ext_for_comb): Take type rather
15730 (aff_combination_const): Adjust.
15731 (aff_combination_scale): Likewise.
15732 (aff_combination_add_elt): Likewise.
15733 (aff_combination_add_cst): Likewise.
15734 (aff_combination_convert): Likewise.
15735 (add_elt_to_tree): Likewise. Remove unused argument.
15736 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
15738 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
15740 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
15742 * config/arm/arm.c (arm_default_short_enums): Use
15743 ARM_DEFAULT_SHORT_ENUMS.
15744 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
15746 2017-04-06 Jakub Jelinek <jakub@redhat.com>
15749 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
15750 members with redundant out-of-class redeclaration.
15752 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15755 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
15756 * config/i386/i386.md (*zero_extendsidi2):
15757 Add (?*x,*x) and (?*v,*v) alternatives.
15759 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15762 * config/i386/i386.c (ix86_expand_builtin)
15763 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
15764 mode from insn data. Convert operands to insn operand mode.
15765 Copy operands that don't satisfy insn predicate to a register.
15767 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
15769 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
15772 2017-04-06 Richard Biener <rguenther@suse.de>
15774 PR tree-optimization/80334
15775 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
15776 preserve alignment of accesses.
15778 2017-04-06 Richard Biener <rguenther@suse.de>
15780 PR tree-optimization/80262
15781 * tree-sra.c (build_ref_for_offset): Preserve address-space
15783 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
15784 Drop useless address-space information on MEM_REF offsets.
15786 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
15788 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
15790 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15792 PR rtl-optimization/70703
15793 * ira-color.c (update_conflict_hard_regno_costs): Use
15794 HOST_WIDE_INT instead of long.
15796 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
15799 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
15800 not defined for x86_64 target. Add -mmmx target option when __SSE2__
15802 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
15803 for x86_64 target. Handle -m3dnowa option.
15805 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15807 PR rtl-optimization/70703
15808 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
15809 (update_conflict_hard_regno_costs): Use long instead of unsigned
15810 arithmetic for cost calculation.
15812 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15813 Bernd Edlinger <bernd.edlinger@hotmail.de>
15816 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
15819 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
15822 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
15823 ptr_mode with Pmode throughout.
15824 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
15825 into probe_stack_range and use DImode.
15827 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
15830 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
15831 call_eh_return is true.
15833 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15835 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
15836 Initialize last_match_fntype_index.
15838 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15841 * tree-nvr.c: Include internal-fn.h.
15842 (pass_return_slot::execute): Ignore internal calls without
15845 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15846 Richard Biener <rguenther@suse.de>
15849 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
15850 captures used multiple times, except for the last use.
15851 * generic-match-head.c: Include gimplify.h.
15853 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15855 PR tree-optimization/79390
15856 * target.h (struct noce_if_info): Declare.
15857 * targhooks.h (default_noce_conversion_profitable_p): Declare.
15858 * target.def (noce_conversion_profitable_p): New target hook.
15859 * ifcvt.h (struct noce_if_info): New type, moved from ...
15860 * ifcvt.c (struct noce_if_info): ... here.
15861 (noce_conversion_profitable_p): Renamed to ...
15862 (default_noce_conversion_profitable_p): ... this. No longer
15864 (noce_try_store_flag_constants, noce_try_addcc,
15865 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
15866 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
15867 instead of noce_conversion_profitable_p.
15868 * config/i386/i386.c: Include ifcvt.h.
15869 (ix86_option_override_internal): Don't override
15870 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
15871 (ix86_noce_conversion_profitable_p): New function.
15872 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
15873 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
15874 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
15875 * doc/tm.texi: Regenerated.
15877 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15879 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
15882 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15885 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
15886 instructions for small multiply cores.
15888 2017-04-04 Jeff Law <law@redhat.com>
15890 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
15892 (mips_expand_vec_perm_const): Initialize elements in orig_perm
15893 that are not set by the loop over the elements.
15895 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15898 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
15899 int mode, convert_modes it to mode as unsigned, otherwise use
15900 lowpart_subreg to mode rather than SImode.
15901 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
15902 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
15903 Use DImode instead of SImode for the shift count operand.
15904 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
15907 2017-04-04 Richard Biener <rguenther@suse.de>
15909 PR middle-end/80281
15910 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
15911 arithmetic done for the negate or the plus. Simplify.
15912 (A - (-B) -> A + B): Likewise.
15913 * fold-const.c (split_tree): Make sure to not negate pointers.
15915 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
15917 PR rtl-optimization/60818
15918 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
15919 a compare of comparisons with the thing compared if this results
15920 in a different machine mode.
15922 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
15924 * alias.c (base_alias_check): Fix typo in comment.
15925 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
15926 * cgraphunit.c (symbol_table::compile): Likewise.
15927 * collect2.c (maybe_run_lto_and_relink): Likewise.
15928 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
15929 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
15930 * config/avr/avr.c (avr_map_op_t): Likewise.
15931 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
15932 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
15933 * config/epiphany/epiphany.md (movcc): Likewise.
15934 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
15935 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
15937 * config/mips/mips.c (mips_save_restore_reg): Likewise.
15938 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
15939 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
15940 * config/sh/sh.c (sh_rtx_costs): Likewise.
15941 * fold-const.c (fold_truth_andor): Likewise.
15942 * genautomata.c (collapse_flag): Likewise.
15943 * gengtype.h (struct type::u::s): Likewise.
15944 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
15945 * input.c (FORMAT_AMOUNT): Likewise.
15946 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
15947 (known_aggs_to_agg_replacement_list): Likewise.
15948 * ipa-inline-analysis.c: Likewise.
15949 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
15950 * ipa-polymorphic-call.c
15951 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
15952 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
15953 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
15955 * modulo-sched.c (apply_reg_moves): Likewise.
15956 * omp-expand.c (build_omp_regions_1): Likewise.
15957 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
15958 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
15959 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
15960 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
15961 * value-prof.c: Likewise.
15962 * var-tracking.c (val_reset): Likewise.
15964 2017-04-03 Richard Biener <rguenther@suse.de>
15966 PR tree-optimization/80275
15967 * fold-const.c (split_address_to_core_and_offset): Handle
15970 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
15972 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
15973 descriptors is at least equal to that of functions.
15975 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
15977 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
15979 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
15982 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
15983 (mov<IMOD4:mode>): New expander.
15984 (*mov<IMOD4:mode>_internal): New insn and split pattern.
15986 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15988 PR rtl-optimization/79405
15989 * fwprop.c (propagations_left): New variable.
15990 (forward_propagate_into): Decrement it.
15991 (fwprop_init): Initialize it.
15992 (fw_prop): If the variable has reached zero, stop propagating.
15993 (fwprop_addr): Ditto.
15995 2017-03-31 Jakub Jelinek <jakub@redhat.com>
15998 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
15999 a FUNCTION_DECL, pass it as decl instead of origin to
16002 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16004 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16007 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16010 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16011 TARGET_VSX_SMALL_INTEGER.
16013 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16015 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16016 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16018 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16020 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16021 extraction from odd-numbered MSA register.
16023 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16025 PR middle-end/80173
16026 * expmed.c (store_bit_field_1): Don't attempt to create
16027 a word subreg out of hard registers wider than word if they
16028 have HARD_REGNO_NREGS of 1 for their mode.
16030 PR middle-end/80163
16031 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16032 conversions to integer types wider than word and pointer.
16035 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16036 (rtx_equal_for_cselib_p): Pass 0 to it.
16037 * cselib.c (cselib_hasher::equal): Likewise.
16038 (rtx_equal_for_cselib_1): Add depth argument. If depth
16039 is 128, don't look up VALUE locs and punt. Increment
16040 depth in recursive calls when walking VALUE locs.
16042 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16044 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16045 (make_gcov_file_name): Use the canonical path name for generating
16047 (read_line): Fix handling of files with ascii null bytes.
16049 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16051 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16052 to initialise a vector register instead
16053 of using a const_int.
16055 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16057 PR translation/80189
16058 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16059 diagnostic messages.
16061 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16064 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16065 (dfp_diex_<mode>): Update mode of operand 1.
16066 * doc/extend.texi (dxex, dxexq): Document change to return type.
16067 (diex, diexq): Document change to argument type.
16069 2017-03-30 Martin Jambor <mjambor@suse.cz>
16072 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16073 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16074 it reflects the signature changes performed at the callee side.
16075 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16076 to cgraph_build_function_type_skip_args.
16077 (build_function_decl_skip_args): Adjust call to the above function.
16079 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16082 * config/i386/sse.md
16083 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16084 register as dest whenever it is a MEM not rtx_equal_p to the
16085 corresponding dup operand, and when forcing into reg move the
16086 reg into the memory afterwards.
16087 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16088 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16089 for the force_reg mode.
16090 (avx512vl_vextractf128<mode>): Use register as dest either
16091 always when a MEM, or when it is a MEM not rtx_equal_p to the
16092 corresponding dup operand, or even not when it is a CONST_VECTOR
16093 depending on the mode and lo vs. hi.
16094 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16096 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16097 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16098 Likewise. Require that operands[2] is even.
16099 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16100 Remove extraneous parens. Require that operands[2] is a multiple
16102 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16103 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16104 disallow memory then.
16106 2017-03-30 Richard Biener <rguenther@suse.de>
16108 PR tree-optimization/77498
16109 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16110 to non-constants over backedges.
16112 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16114 PR rtl-optimization/80233
16115 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16116 as last_combined_insn. Do not test for BARRIER_P separately.
16118 2017-03-29 Andreas Schwab <schwab@suse.de>
16121 * calls.c (prepare_call_address): Convert funexp to Pmode before
16122 copying to temp reg.
16124 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16126 PR tree-optimization/80158
16127 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16128 Handle possible future case of more than one alternate
16130 (replace_rhs_if_not_dup): Likewise.
16131 (replace_one_candidate): Likewise.
16133 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16135 PR rtl-optimization/80193
16136 * ira.c (ira): Do not check allocation for LRA.
16138 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16140 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16141 (nvptx_output_simt_exit): Declare.
16142 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16143 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16144 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16145 (nvptx_declare_function_name): ...here. Emit declaration of local
16146 memory space buffer for omp_simt_enter insn.
16147 (nvptx_output_unisimt_switch): New.
16148 (nvptx_output_softstack_switch): New.
16149 (nvptx_output_simt_enter): New.
16150 (nvptx_output_simt_exit): New.
16151 * config/nvptx/nvptx.h (struct machine_function): New fields
16152 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16153 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16154 (UNSPECV_SIMT_EXIT): Ditto.
16155 (omp_simt_enter_insn): New insn.
16156 (omp_simt_enter): New expansion.
16157 (omp_simt_exit): New insn.
16158 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16160 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16161 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16162 (expand_GOMP_SIMT_EXIT): New.
16163 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16164 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16165 (GOMP_SIMT_EXIT): Ditto.
16166 * target-insns.def (omp_simt_enter): New insn.
16167 (omp_simt_exit): Ditto.
16168 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16170 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16171 (lower_rec_input_clauses): Likewise.
16172 (lower_lastprivate_clauses): Handle SIMT privatization.
16174 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16175 (ompdevlow_adjust_simt_enter): New.
16176 (find_simtpriv_var_op): New.
16177 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16178 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16180 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16181 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16182 (copy_decl_for_dup_finish): Ditto.
16184 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16186 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16189 * config/i386/i386.c (ix86_option_override_internal): Always
16190 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16192 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16194 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16196 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16198 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16199 mark new edge's irreducible flag accordign to it.
16200 (vect_do_peeling): Check loop preheader edge's irreducible flag
16201 and pass it to function slpeel_add_loop_guard.
16203 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16205 PR tree-optimization/80218
16206 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16207 Update block frequencies and counts.
16209 2017-03-28 Richard Biener <rguenther@suse.de>
16211 PR tree-optimization/78644
16212 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16213 of a simplification result we may not use it at all.
16215 2017-03-28 Richard Biener <rguenther@suse.de>
16218 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16219 without arguments, generate default definition of a SSA name.
16221 2017-03-28 Richard Biener <rguenther@suse.de>
16223 PR middle-end/80222
16224 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16225 TYPE_REF_CAN_ALIAS_ALL references.
16226 * fold-const.c (fold_indirect_ref_1): Likewise.
16228 2017-03-28 Martin Liska <mliska@suse.cz>
16231 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16232 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16234 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16235 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16237 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16238 (EXTRA_SPECS): Define.
16239 (SUBTARGET_EXTRA_SPECS): Likewise.
16240 (SUBTARGET_CPP_SPEC): Likewise.
16241 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16242 SUBTARGET_EXTRA_SPECS.
16243 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16245 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16247 * config/arc/simdext.md (vst64_insn): Update pattern.
16248 (vld32wh_insn): Likewise.
16249 (vld32wl_insn): Likewise.
16250 (vld64_insn): Likewise.
16251 (vld32_insn): Likewise.
16253 2017-03-28 Marek Polacek <polacek@redhat.com>
16256 * fold-const.c (fold_comparison): Use protected_set_expr_location
16257 instead of SET_EXPR_LOCATION.
16259 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16261 * tree.c (add_expr): Avoid name lookup warning.
16263 2017-03-27 Jeff Law <law@redhat.com>
16265 PR tree-optimization/80216
16266 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16267 function name. Limit recursion depth.
16268 (record_temporary_equivalences): Corresponding changes.
16270 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16272 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16275 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16278 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16280 * cfgcleanup.c (reg_note_cfa_p): New array.
16281 (insns_have_identical_cfa_notes): New function.
16282 (old_insns_match_p): Don't cross-jump in between /f
16283 and non-/f instructions. If both i1 and i2 are frame related,
16284 verify all CFA notes, their order and content.
16286 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16289 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16290 HImode and SImode with zero extend to DImode to one insn.
16291 (bswap<mode>2_extenddi): Likewise.
16292 (bswapsi2_extenddi): Likewise.
16293 (bswaphi2_extendsi): Likewise.
16294 (bswaphi2): Combine bswap HImode and SImode into one insn.
16295 Separate memory insns from swapping register.
16296 (bswapsi2): Likewise.
16297 (bswap<mode>2): Likewise.
16298 (bswaphi2_internal): Delete, no longer used.
16299 (bswapsi2_internal): Likewise.
16300 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16301 store, and gpr<-gpr swap insns.
16302 (bswap<mode>2_store): Likewise.
16303 (bswaphi2_reg): Register only splitter, combine with the splitter.
16304 (bswaphi2 splitter): Likewise.
16305 (bswapsi2_reg): Likewise.
16306 (bswapsi2 splitter): Likewise.
16307 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16308 the insns into load, store, and register/register insns.
16309 (bswapdi2_ldbrx): Likewise.
16310 (bswapdi2_load): Likewise.
16311 (bswapdi2_store): Likewise.
16312 (bswapdi2_reg): Likewise.
16314 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16316 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16317 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16319 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16322 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16324 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16325 special handling for target option conflicts between dform
16326 options (-mpower9-dform, -mpower9-dform-vector,
16327 -mpower9-dform-scalar) and -mno-direct-move.
16329 2017-03-27 Richard Biener <rguenther@suse.de>
16331 PR tree-optimization/80181
16332 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16334 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16336 * config/arc/predicates.md (move_double_src_operand): Replace the
16337 call to move_double_src_operand with a call to address_operand.
16339 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16341 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16342 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16343 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16345 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16347 * config/arc/predicates.md (long_immediate_loadstore_operand):
16348 Consider scaled addresses cases.
16350 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16352 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16353 restored when in interrupt.
16354 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16355 doesn't have delay slot.
16357 2017-03-27 Richard Biener <rguenther@suse.de>
16360 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16361 inlined thunk clones.
16363 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16366 * asan.c (instrument_derefs): Copy over last operand from
16367 original COMPONENT_REF to the new COMPONENT_REF with
16368 DECL_BIT_FIELD_REPRESENTATIVE.
16369 * ubsan.c (instrument_object_size): Likewise.
16371 2017-03-27 Richard Biener <rguenther@suse.de>
16373 PR tree-optimization/80170
16374 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16375 sure DR/SCEV didnt fold in constants we do not see when looking
16376 at the reference base alignment.
16378 2017-03-27 Richard Biener <rguenther@suse.de>
16380 PR middle-end/80171
16381 * gimple-fold.c (fold_ctor_reference): Properly guard against
16382 NULL return value from canonicalize_constructor_val.
16384 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16387 * config/i386/i386.c (ix86_expand_builtin)
16388 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16389 flags reg setting and flags reg using instructions.
16390 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16391 clobbering instructions to zero extend op2.
16393 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16395 * doc/install.texi (Configuration) <--with-aix-soname>:
16396 Update link to AIX ld.
16398 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16400 PR rtl-optimization/80160
16401 PR rtl-optimization/80159
16402 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16403 reg_alternate_class into account.
16405 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16408 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16409 to consider in curr_insn_transform.
16411 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16413 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16414 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16415 and emit_mode_inner.
16417 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16419 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16420 argument to the overloaded builtin variants. Use the new flag to
16421 deprecate certain builtin variants.
16422 * config/s390/s390-builtin-types.def: Add new builtin types.
16423 * config/s390/s390-builtins.h: Support new flags field for
16424 overloaded builtins.
16425 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16426 (s390_macro_to_expand): Enable vector float data type.
16427 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16428 builtins by incrementing the __VEC__ version number.
16429 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16431 (s390_resolve_overloaded_builtin): Emit error messages depending
16432 on the builtin flags.
16433 * config/s390/s390.c (s390_expand_builtin): Support additional
16434 flags argument. Change error message to match the messages
16435 emitted in s390-c.c.
16436 * config/s390/s390.md: New UNSPEC_* constants.
16437 (op_type): Add new instruction types.
16438 * config/s390/vecintrin.h: Add new builtins and test data class
16440 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16441 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16442 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16443 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16444 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16445 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16447 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16448 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16449 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16450 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16452 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16453 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16454 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16455 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16457 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16458 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16459 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16460 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16461 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16462 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16463 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16465 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16466 ("vec_scatter_element<V_HW_4:mode>_DI")
16467 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16468 ("vec_fpint<mode>", "vflls")
16469 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16470 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16471 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16472 ("*vec_cmphe<mode>_cc"): ... these.
16474 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16475 mode constant instead of magic value.
16477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16479 * config/s390/s390.c (s390_expand_vec_compare): Support other
16480 vector floating point modes than just V2DF.
16481 (s390_expand_vcond): Likewise.
16482 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16483 (s390_cannot_change_mode_class): Prevent mode changes between TF
16484 and V1TF in vector registers.
16485 * config/s390/s390.md (DF, SF): New mode attributes.
16486 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16487 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16488 SFmode support for VRs.
16489 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16491 (VFT, VF_HW): New mode iterators.
16492 (vw, sdx): New mode attributes.
16493 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16494 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16495 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16496 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16497 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16498 also the new vector floating point modes. Renaming to ...
16500 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16501 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16502 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16503 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16504 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16505 ("vec_unordered<mode>"): ... these.
16507 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16508 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16509 ("*vec_extendv2df"): New insn definitions.
16511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16513 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16514 ("mulditi3_2", "*muldi3_sign"): New patterns.
16515 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16516 rename the pattern definition.
16518 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16520 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16522 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16524 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16526 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16527 instruction if possible.
16528 * config/s390/vector.md (vec_halfnumelts): New mode
16530 ("*vec_vllezlf<mode>"): New pattern.
16532 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16534 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16535 ("popcountv4si2", "popcountv2di2"): Rename to ...
16536 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16537 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16539 ("popcount<mode>2_vxe"): New pattern.
16541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16543 * common/config/s390/s390-common.c (processor_flags_table): Add
16545 * config.gcc: Add arch12.
16546 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16547 Default to arch12 for unknown CPU model numbers.
16548 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16549 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16550 PROCESSOR_max sanity check.
16551 * config/s390/s390-opts.h (enum processor_type): Add
16553 * config/s390/s390.c (processor_table): Add arch12.
16554 (s390_expand_builtin): Add check for B_VXE flag.
16555 (s390_issue_rate): Add PROCESSOR_ARCH12.
16556 (s390_get_sched_attrmask): Likewise.
16557 (s390_get_unit_mask): Likewise.
16558 (s390_sched_score): Enable z13 scheduling for arch12.
16559 (s390_sched_reorder): Likewise.
16560 (s390_sched_variable_issue): Likewise.
16561 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16563 (s390_tune_attr): Use z13 scheduling also for arch12.
16564 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16565 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16566 (TARGET_VXE_P): New macros.
16567 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16568 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16569 * config/s390/s390.opt: Add arch12 as processor_type.
16571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16573 * config/s390/s390.md
16574 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16575 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16576 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16578 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16579 Rename expanders to ...
16581 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16582 ("fixuns_truncdddi2_emu"): ... these.
16584 ("fixuns_trunc<mode>si2_emu"): New expander.
16586 ("*fixuns_truncdfdi2_z13"): Rename to ...
16587 ("*fixuns_truncdfdi2_vx"): ... this.
16589 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16591 * config/s390/2964.md: Remove the single element vector compare
16592 instructions which are no longer used.
16593 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16595 (s390_canonicalize_comparison): Remove handling of DFmode
16597 (s390_expand_vec_compare_scalar): Remove function.
16598 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16599 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16601 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16605 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16607 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
16608 will anyway by matched by mov<mode>_64dfp.
16610 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16612 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
16613 vlef/vstef. Add missing operand to vleif.
16615 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16617 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
16618 pair for all vector types with 64 bit elements.
16619 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
16620 * config/s390/vector.md (V_HW_64): ... here.
16621 (V_128_NOSINGLE): New mode iterator.
16622 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
16623 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
16624 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
16625 ("*vec_load_pairv2di"): Change to ...
16626 ("*vec_load_pair<mode>"): ... this one.
16628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16630 * config/s390/constraints.md: Add comments.
16631 (jKK): Reject element sizes > 8 bytes.
16632 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
16634 * config/s390/s390.md: Add the s_operand checks formerly in
16635 s390_split_ok_p to various splitters where they are still
16637 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
16638 for 128 bit vectors. Plus two splitters.
16640 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16642 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
16645 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16648 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
16649 error if the boundary argument is not constant.
16651 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16653 PR rtl-optimization/80112
16654 * loop-doloop.c (doloop_condition_get): Don't check condition
16655 if cmp isn't SET with IF_THEN_ELSE src.
16657 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16659 PR tree-optimization/80158
16660 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
16661 replacing a candidate statement, also replace it for the
16662 candidate's alternate interpretation.
16663 (replace_rhs_if_not_dup): Likewise.
16664 (replace_one_candidate): Likewise.
16666 2017-03-24 Richard Biener <rguenther@suse.de>
16668 PR tree-optimization/80167
16669 * graphite-isl-ast-to-gimple.c
16670 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
16672 (translate_isl_ast_to_gimple::get_rename): Likewise.
16674 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
16677 handling of certain combinations of target options, including the
16678 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
16679 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
16681 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16684 * config/arm/arm.md (*load_multiple): Add reload_completed to
16685 matching condition.
16687 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16688 Richard Biener <rguenth@suse.de>
16690 PR tree-optimization/79908
16691 PR tree-optimization/80136
16692 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16693 been cast away, gimplify_and_add suffices.
16695 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
16697 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
16699 2017-03-23 Richard Biener <rguenther@suse.de>
16701 PR tree-optimization/80032
16702 * gimplify.c (gimple_push_cleanup): Forced unconditional
16703 cleanups still have to go to the conditional_cleanups
16706 2017-03-22 Jakub Jelinek <jakub@redhat.com>
16708 PR tree-optimization/80072
16709 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
16711 (next_operand_entry_id): Change type to unsigned int.
16712 (sort_by_operand_rank): Make sure to return the right return value
16713 even if unsigned fields are bigger than INT_MAX.
16714 (struct oecount): Change cnt and id type to unsigned int.
16715 (oecount_hasher::equal): Formatting fix.
16716 (oecount_cmp): Make sure to return the right return value
16717 even if unsigned fields are bigger than INT_MAX.
16718 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
16721 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
16722 TREE_READONLY on result if writing it more than once.
16725 * doc/invoke.texi (-fsanitize=thread): Document that with
16726 -fnon-call-exceptions atomics are not able to throw
16730 * tsan.c: Include tree-eh.h.
16731 (instrument_builtin_call): Call maybe_clean_eh_stmt or
16732 maybe_clean_or_replace_eh_stmt where needed.
16733 (instrument_memory_accesses): Add cfg_changed argument.
16734 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
16735 if it returned true.
16736 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
16738 PR rtl-optimization/63191
16739 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
16740 wrapper function, moved the whole old content into ...
16741 (ix86_delegitimize_address_1): ... this. New inline function.
16742 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
16743 true as last argument instead of ix86_delegitimize_address.
16745 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16747 * config/aarch64/aarch64.c (generic_branch_cost): Copy
16748 cortexa57_branch_cost.
16750 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16752 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
16754 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16757 * doc/md.texi (Constraints): Document wA constraint.
16758 * config/rs6000/constraints.md (wA): New.
16759 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
16760 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
16761 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
16762 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
16764 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
16767 * gimplify.c (is_oacc_declared): New function.
16768 (oacc_default_clause): Use it to set default flags for acc declared
16769 variables inside parallel regions.
16770 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
16771 declared variables.
16772 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
16773 declare attribute to any decl as necessary.
16775 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16778 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
16779 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
16780 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
16781 (arm_arch_lpae): This.
16782 * config/arm/arm.c (arm_arch7ve): Rename into ...
16783 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
16784 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
16787 2017-03-22 Martin Liska <mliska@suse.cz>
16790 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
16791 error message instead of an ICE.
16793 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16795 * doc/extend.texi (6.11 Additional Floating Types): Revise.
16797 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
16799 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
16801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16804 2017-03-21 Martin Sebor <msebor@redhat.com>
16806 * doc/extend.texi: Use "cannot" instead of "can't."
16807 * doc/hostconfig.texi: Same.
16808 * doc/install.texi: Same.
16809 * doc/invoke.texi: Same.
16810 * doc/loop.texi: Same.
16811 * doc/md.texi: Same.
16812 * doc/objc.texi: Same.
16813 * doc/rtl.texi: Same.
16814 * doc/tm.texi: Same.
16815 * doc/tm.texi.in: Same.
16816 * doc/trouble.texi: Same.
16818 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
16821 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
16822 (collect_checksum_attributes): Set it.
16823 (die_checksum_ordered): Use it.
16825 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16827 PR tree-optimization/79908
16828 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
16829 change: For a VA_ARG whose LHS has been cast away, use
16830 force_gimple_operand to construct the side effects.
16832 2017-03-21 David Malcolm <dmalcolm@redhat.com>
16834 PR translation/80001
16835 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
16836 more amenable to translation.
16837 (oacc_loop_auto_partitions): Likewise.
16839 2017-03-21 Marek Polacek <polacek@redhat.com>
16840 Martin Sebor <msebor@redhat.com>
16842 PR tree-optimization/80109
16843 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
16844 on INTEGRAL_TYPE_P.
16846 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16847 Segher Boessenkool <segher@kernel.crashing.org>
16850 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
16851 check reg_used_between_p between insn and one of succ or succ2
16852 depending on if succ is artificial insn not inserted into insn
16855 2017-03-21 Martin Liska <mliska@suse.cz>
16857 PR gcov-profile/80081
16858 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
16859 * doc/gcc.texi: Include gcov-dump stuff.
16860 * doc/gcov-dump.texi: New file.
16862 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
16864 PR rtl-optimization/79150
16865 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
16866 conditional jump, if the jump is the last insn of the loop.
16868 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16869 Richard Biener <rguenth@suse.de>
16871 PR tree-optimization/79908
16872 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16873 been cast away, use force_gimple_operand to construct the side
16876 2017-03-21 Martin Liska <mliska@suse.cz>
16878 PR libfortran/79956
16879 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
16882 2017-03-21 Brad Spengler <spender@grsecurity.net>
16885 * plugin.c (htab_hash_plugin): New function.
16886 (add_new_plugin): Use it and adjust.
16887 (parse_plugin_arg_opt): Adjust.
16888 (init_one_plugin): Likewise.
16890 2017-03-21 Richard Biener <rguenther@suse.de>
16892 PR tree-optimization/80032
16893 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
16894 if set force the cleanup to happen unconditionally.
16895 (gimplify_target_expr): Push inserted clobbers with force_uncond
16896 to avoid them being removed by control-dependent DCE.
16898 2017-03-21 Richard Biener <rguenther@suse.de>
16900 PR tree-optimization/80122
16901 * tree-inline.c (copy_bb): Do not expans va-arg packs or
16902 va_arg_pack_len when the inlined call stmt requires pack
16904 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
16906 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16909 * tsan.c (instrument_builtin_call): If the memory model argument
16910 is not a constant, assume it is valid.
16913 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
16916 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
16918 PR rtl-optimization/79910
16919 * combine.c (can_combine_p): Do not allow combining an I0 or I1
16920 if its dest is used by an insn before I2 (other than the combined
16921 insns themselves, which are properly handled already).
16923 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
16926 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
16928 * combine.c (record_used_regs): New static function.
16929 (try_combine): Handle situations where there is an additional
16930 instruction between I2 and I3 which needs to have a LOG_LINK
16934 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
16936 * combine.c (try_combine): Delete redundant i1 test. Call
16937 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
16939 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16942 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
16943 alternatives 13/14.
16945 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16947 PR tree-optimization/80054
16948 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
16949 the optimization if a PHI or any of its arguments is not dominated
16950 by the candidate's basis. Use gphi* rather than gimple* as
16952 (replace_profitable_candidates): Clean up a gimple* variable that
16953 should be a gphi* variable.
16955 2017-03-20 Martin Sebor <msebor@redhat.com>
16958 * doc/extend.texi (attribute constructor): Document present limitation.
16960 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
16963 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
16964 __POWER9_VECTOR__ #ifdef control, change template definition to
16965 use Power9-specific built-in function.
16966 (vec_any_eq): Likewise.
16967 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
16968 to control outcomes from this test.
16969 (vector_ae_<mode>p): For VEC_F modes, likewise.
16971 2017-03-20 Ian Lance Taylor <iant@google.com>
16973 * config/i386/i386.c (ix86_function_regparm): Save an extra
16974 register for -fsplit-stack with DECL_STATIC_CHAIN.
16976 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
16979 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
16980 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
16982 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
16984 * config/riscv/riscv.c (riscv_print_operand): Use "fence
16986 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
16989 2017-03-20 Marek Polacek <polacek@redhat.com>
16992 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
16994 2017-03-20 Richard Biener <rguenther@suse.de>
16996 PR tree-optimization/80113
16997 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
16998 allocate extra SSA name for PHI def.
16999 (add_close_phis_to_outer_loops): Likewise.
17000 (add_close_phis_to_merge_points): Likewise.
17001 (copy_loop_close_phi_args): Likewise.
17002 (copy_cond_phi_nodes): Likewise.
17004 2017-03-20 Martin Liska <mliska@suse.cz>
17006 PR middle-end/79753
17007 * tree-chkp.c (chkp_build_returned_bound): Do not build
17008 returned bounds for a LHS that's not a BOUNDED_P type.
17010 2017-03-20 Martin Liska <mliska@suse.cz>
17014 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17015 COMPLEX_CST and VECTOR_CST.
17017 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17020 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17021 target operand. A new splitter adds the clobber statement in case
17022 the target operand is dead anyway.
17024 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17026 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17027 to age-old versions of binutils and glibc.
17029 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17031 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17033 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17035 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17037 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17039 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17040 requirement for binutils 2.13.
17042 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17044 * combine.c (try_combine): Delete redundant i1 test. Call
17045 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17047 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17049 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17050 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17052 <riscv64-*-elf>: Re-arrange section
17053 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17054 <riscv32-*-linux>: Likewise.
17055 <riscv64-*-elf>: Likewise
17056 <riscv64-*-linux>: Likewise.
17058 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17061 * aarch64.opt(verbose-cost-dump): Fix typo.
17063 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17066 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17067 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17069 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17071 * reload.c (find_reloads): When reloading a nonoffsettable address,
17072 use RELOAD_OTHER for it and its address reloads.
17074 PR rtl-optimization/79910
17075 * combine.c (record_used_regs): New static function.
17076 (try_combine): Handle situations where there is an additional
17077 instruction between I2 and I3 which needs to have a LOG_LINK
17080 2017-03-17 Jeff Law <law@redhat.com>
17082 PR tree-optimization/71437
17083 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17084 conditional in the hash table first.
17085 (vrp_dom_walker::before_dom_children): Extract condition from
17086 ASSERT_EXPR. Record condition, its inverion and any implied
17087 conditions as well.
17089 2017-03-17 Marek Polacek <polacek@redhat.com>
17090 Markus Trippelsdorf <markus@trippelsdorf.de>
17092 PR tree-optimization/80079
17093 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17096 2017-03-17 Richard Biener <rguenther@suse.de>
17098 PR middle-end/80075
17099 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17100 Properly verify the LHS before the RHS possibly claims to be
17102 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17105 2017-03-17 Martin Jambor <mjambor@suse.cz>
17107 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17108 (List of -O2 options): Likewise.
17109 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17112 2017-03-17 Tom de Vries <tom@codesourcery.com>
17114 * gcov-dump.c (print_usage): Print bug_report_url.
17116 2017-03-17 Richard Biener <rguenther@suse.de>
17118 PR middle-end/80050
17119 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17120 (parser::peek): Likewise.
17122 2017-03-17 Richard Biener <rguenther@suse.de>
17124 PR tree-optimization/80048
17125 * sese.c (free_sese_info): Properly release rename_map and
17126 copied_bb_map elements.
17128 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17130 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17131 Add linked-list forward and backlinks. Insert on
17132 construction, remove on destruction.
17133 (class pass_store_merging): Add m_stores_head field.
17134 (pass_store_merging::terminate_and_process_all_chains):
17135 Iterate over m_stores_head list.
17136 (pass_store_merging::terminate_all_aliasing_chains):
17138 (pass_store_merging::execute): Check for debug stmts first.
17139 Push new chains onto the m_stores_head stack.
17141 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17144 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17145 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17146 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17148 2017-03-16 Jeff Law <law@redhat.com>
17150 PR tree-optimization/71437
17151 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17152 member function. Implementation moved into after_dom_children
17153 member function and into the threader's thread_outgoing_edges
17155 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17156 some code into new thread_outgoing_edges.
17157 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17158 definition. Simplify marker handling (do it here). Assume we always
17159 have the available expression and the const/copies tables.
17160 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17162 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17163 * tree-vrp.c (equiv_stack): No longer file scoped.
17164 (vrp_dom_walker): New class.
17165 (vrp_dom_walker::before_dom_children): New member function.
17166 (vrp_dom_walker::after_dom_children): Likewise.
17167 (identify_jump_threads): Setup domwalker. Use it rather than
17168 walking edges in a random order by hand. Simplify setup/finalization.
17169 (finalize_jump_threads): Remove.
17170 (vrp_finalize): Do not call identify_jump_threads here.
17171 (execute_vrp): Do it here instead and call thread_through_all_blocks
17174 PR tree-optimization/71437
17175 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17177 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17179 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17180 (dom_opt_dom_walker::thread_across_edge): Remove
17181 handle_dominating_asserts argument. All callers changed.
17182 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17183 changes. Remove calls to lhs_of_dominating_assert. Other
17184 uses of handle_dominating_asserts turn into unconditional code
17185 (simplify_control_stmt_condition_1): Likewise.
17186 (simplify_control_stmt_condition): Likewise.
17187 (thread_through_normal_block, thread_across_edge): Likewise.
17188 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17189 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17190 object if it is not an SSA_NAME.
17191 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17192 before calling into the VRP specific simplifiers.
17193 (identify_jump_threads): Remove handle_dominating_asserts
17196 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17199 * tree-diagnostic.c (default_tree_printer): No longer static.
17200 * tree-diagnostic.h (default_tree_printer): New prototype.
17202 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17204 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17205 Change ins into fmov.
17207 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17209 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17210 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17211 Use h_con constraint for operand 1.
17212 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17213 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17215 2017-03-15 Jeff Law <law@redhat.com>
17217 PR tree-optimization/71437
17218 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17219 (record_temporary_equivalences): Use it.
17221 PR tree-optimization/71437
17222 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17223 tree-ssa-scopedtables.
17224 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17225 (record_conditions, record_cond, vuse_eq): Likewise.
17226 (record_edge_info): Adjust to API tweak of record_conditions.
17227 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17228 (record_temporary_equivalences, optimize_stmt): Likewise.
17229 (eliminate_redundant_computations): Likewise.
17230 (record_equivalences_from_stmt): Likewise.
17231 * tree-ssa-scopedtables.c: Include options.h and params.h.
17232 (vuse_eq): New function, moved from tree-ssa-dom.c
17233 (build_and_record_new_cond): Likewise.
17234 (record_conditions): Likewise. Accept vector of conditions rather
17235 than edge_equivalence structure for first argument.
17236 for the first argument.
17237 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17238 from tree-ssa-dom.c.
17239 (avail_exprs_stack::record_cond): Likewise.
17240 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17241 from tree-ssa-dom.c.
17242 (avail_exprs_stack): Add new member functions lookup_avail_expr
17244 (record_conditions): Declare.
17246 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17249 * lra-constraints.c (process_alt_operands): Increase reject for
17250 reloading an input/output operand.
17252 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17255 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17256 insns to convert from signed/unsigned char/short to IEEE 128-bit
17258 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17260 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17263 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17264 subreg of inner mode for values already in registers.
17266 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17268 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17269 iteration reg is used after the loop.
17271 2017-03-14 Martin Sebor <msebor@redhat.com>
17273 PR tree-optimization/79800
17274 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17275 precision in negative-positive range.
17276 (format_floating): Call non-const overload with adjusted precision.
17278 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17281 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17284 2017-03-14 Martin Sebor <msebor@redhat.com>
17286 PR middle-end/80020
17287 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17288 * builtins.def (aligned_alloc): Use it.
17291 * Makefile.in (GTFILES): Add calls.c.
17292 * calls.c: Include "gt-calls.h".
17294 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17296 PR rtl-optimization/79728
17297 * regs.h (struct target_regs): New field
17298 x_contains_allocatable_regs_of_mode.
17299 (contains_allocatable_regs_of_mode): New macro.
17300 * reginfo.c (init_reg_sets_1): Initialize it, and change
17301 contains_reg_of_mode so it includes global regs as well.
17302 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17303 rather than contains_regs_of_mode.
17305 2017-03-14 Martin Liska <mliska@suse.cz>
17307 * doc/invoke.texi: Document options that can't be combined with
17308 -fcheck-pointer-bounds.
17310 2017-03-14 Martin Liska <mliska@suse.cz>
17312 PR middle-end/79831
17313 * doc/invoke.texi (-Wchkp): Document the option.
17315 2017-03-14 Martin Liska <mliska@suse.cz>
17317 * Makefile.in: Install gcov-dump.
17319 2017-03-14 Martin Liska <mliska@suse.cz>
17321 * multiple_target.c (expand_target_clones): Bail out for
17322 an invalid attribute.
17324 2017-03-14 Richard Biener <rguenther@suse.de>
17326 * alias.c (struct alias_set_entry): Pack properly.
17327 * cfgloop.h (struct loop): Likewise.
17328 * cse.c (struct set): Likewise.
17329 * ipa-utils.c (struct searchc_env): Likewise.
17330 * loop-invariant.c (struct invariant): Likewise.
17331 * lra-remat.c (struct cand): Likewise.
17332 * recog.c (struct change_t): Likewise.
17333 * rtl.h (struct address_info): Likewise.
17334 * symbol-summary.h (function_summary): Likewise.
17335 * tree-loop-distribution.c (struct partition): Likewise.
17336 * tree-object-size.c (struct object_size_info): Likewise.
17337 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17338 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17339 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17340 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17341 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17342 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17343 (struct _stmt_vec_info): Likewise.
17345 2017-03-14 Martin Liska <mliska@suse.cz>
17348 * multiple_target.c (create_dispatcher_calls): Check that
17349 a target can create a function dispatcher.
17351 2017-03-14 Martin Liska <mliska@suse.cz>
17354 * multiple_target.c (expand_target_clones): Drop local.local
17355 flag for default implementation.
17357 2017-03-14 Richard Biener <rguenther@suse.de>
17359 PR tree-optimization/80030
17360 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17362 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17364 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17365 gcc_fallthrough() instead of __attribute__((fallthrough));
17367 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17369 * doc/gcc.texi: Remove "up" link to (DIR).
17370 * doc/gccint.texi: Ditto.
17372 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17374 * doc/install.texi (Specific) <avr>: Remove reference to
17377 2017-03-13 Jeff Law <law@redhat.com>
17379 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17380 attribute rather than comments.
17382 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17383 match_scratch operand is highest.
17385 2017-03-13 Martin Liska <mliska@suse.cz>
17387 PR middle-end/78339
17388 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17389 is a CHKP clone, use original declaration.
17391 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17393 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17394 (arc_conditional_register_usage): Use a different allocation order
17395 when optimizing for size.
17396 * common/config/arc/arc-common.c (arc_option_optimization_table):
17397 Section anchors default on when optimizing for size.
17399 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17401 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17403 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17405 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17406 * config/arc/arc.md (cpu_facility): Add cd variant.
17407 (*movqi_insn): Add code density variant.
17408 (*movhi_insn): Likewise.
17409 (*movqi_insn): Likewise.
17410 (*addsi3_mixed): Likewise.
17411 (subsi3_insn): Likewise.
17413 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17415 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17417 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17419 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17420 expressions with MINUS and UNARY ops.
17422 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17425 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17427 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17428 between vec_select and vector argument.
17429 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17430 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17431 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17432 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17433 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17434 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17436 2017-03-13 Richard Biener <rguenther@suse.de>
17439 * params.def (vect-max-peeling-for-alignment): Fix typo.
17441 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17443 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17444 issue that only occurred with binutils below 2.18.
17446 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17448 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17449 refer to binutils 2.11/2.12 minimum.
17451 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17453 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17454 ftp.kernel.org and simplify binutils requirement.
17456 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17458 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17460 (Optimize Options): Ditto. Also remove redundancy.
17462 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17464 PR translation/79848
17465 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17467 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17468 to G_ to avoid double translation.
17470 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17472 PR translation/79923
17473 * auto-profile.c (get_combined_location): Convert leading
17474 character of diagnostics to lower case and remove trailing period.
17475 (read_profile): Likewise for various diagnostics.
17476 * config/arm/arm.c (arm_option_override): Remove trailing period
17477 from various diagnostics.
17478 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17479 (msp430_expand_delay_cycles): Likewise.
17481 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17484 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17485 full command-line argument, rather than just "str".
17486 (aarch64_validate_march): Likewise.
17487 (aarch64_validate_mtune): Likewise.
17489 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17491 PR rtl-optimization/78911
17492 * lra-assigns.c (must_not_spill_p): New function.
17493 (spill_for): Use it.
17495 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17497 PR tree-optimization/79981
17498 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17499 ATOMIC_COMPARE_EXCHANGE ifn result.
17500 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17501 IFN_ATOMIC_COMPARE_EXCHANGE.
17503 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17506 * opts.c (parse_sanitizer_options): Add missing question mark to
17507 "did you mean" message.
17509 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17511 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17513 (VMULEUH_UNS): Likewise.
17514 (VMULOUB_UNS): Likewise.
17515 (VMULOUH_UNS): Likewise.
17516 * config/rs6000/rs6000.c (builtin_function_type): Remove
17517 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17519 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17522 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17523 x with result of extra_parsing_for_operand_code_0.
17524 (function_reader::extra_parsing_for_operand_code_0): Convert
17525 return type from void to rtx, returning x. When reading
17526 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17527 larger size containing struct block_symbol.
17529 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17531 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17532 -mfloat128-hardware without -m64.
17534 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17537 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17538 entries to the case statement that marks unsigned arguments to
17539 overloaded functions.
17541 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17543 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17544 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17546 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17549 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17550 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17552 2017-03-10 Martin Liska <mliska@suse.cz>
17556 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17557 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17560 2017-03-10 Olivier Hainque <hainque@adacore.com>
17562 * tree-switch-conversion (array_value_type): Start by resetting
17563 candidate type to it's main variant.
17565 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17567 PR rtl-optimization/79909
17568 * combine.c (try_combine): Use simplify_replace_rtx on individual
17569 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17570 of the whole CALL_INSN_FUNCTION_USAGE.
17572 PR tree-optimization/79972
17573 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17574 get_range_info on SSA_NAMEs. Formatting fixes.
17576 2017-03-10 Richard Biener <rguenther@suse.de>
17577 Jakub Jelinek <jakub@redhat.com>
17579 PR tree-optimization/77975
17580 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17581 edge to be constant.
17582 (get_val_for): For constant x return it. Formatting fix.
17583 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17584 would use the same bases as the current one.
17586 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17588 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17589 instead of vec_select for V1TImode.
17590 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17592 (VSX_LE_128): Add V1TI to this mode iterator.
17593 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17594 (*vsx_le_perm_store_<mode>): Likewise.
17595 (pre-reload splitter for VSX stores): Likewise.
17596 (post-reload splitter for VSX stores): Likewise.
17597 (*vsx_xxpermdi2_le_<mode>): Likewise.
17598 (*vsx_lxvd2x2_le_<mode>): Likewise.
17599 (*vsx_stxvd2x2_le_<mode>): Likewise.
17601 2017-03-09 Michael Eager <eager@eagercon.com>
17603 Correct failures with --enable-checking=yes,rtl.
17605 * config/microblaze/microblaze.c (microblaze_expand_shift):
17606 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17607 test for const0_rtx.
17608 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
17609 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
17611 2017-03-09 Richard Biener <rguenther@suse.de>
17613 PR tree-optimization/79977
17614 * graphite-scop-detection.c (scop_detection::merge_sese):
17615 Handle the case of extra exits to blocks dominating the entry.
17617 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
17619 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17622 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
17624 PR rtl-optimization/79949
17625 * lra-constraints.c (process_alt_operands): Check memory when
17626 trying to predict a cycle. Print about the overall increase.
17628 2017-03-09 Richard Biener <rguenther@suse.de>
17630 PR middle-end/79971
17631 * gimple-expr.c (useless_type_conversion_p): Preserve
17632 TYPE_SATURATING for fixed-point types.
17634 2017-03-09 Richard Biener <rguenther@suse.de>
17637 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
17638 alignment of BLKmode params.
17640 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17643 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
17644 (VALL_NO_V2Q): Likewise.
17646 * config/aarch64/aarch64-simd.md
17647 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
17649 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
17650 VALL_NO_V2Q mode iterator.
17651 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
17653 2017-03-09 Martin Liska <mliska@suse.cz>
17655 PR tree-optimization/79631
17656 * tree-chkp-opt.c (chkp_is_constant_addr): Call
17657 tree_int_cst_sign_bit just for INTEGER constants.
17659 2017-03-09 Martin Liska <mliska@suse.cz>
17663 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
17666 2017-03-09 Marek Polacek <polacek@redhat.com>
17669 * tree.c (inchash::add_expr): Handle TREE_VEC.
17671 2017-03-09 Martin Liska <mliska@suse.cz>
17674 (chkp_narrow_size_and_offset): New function.
17675 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
17676 (void chkp_parse_bit_field_ref): New function.
17677 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
17678 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
17680 2017-03-09 Martin Liska <mliska@suse.cz>
17683 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
17684 (chkp_find_bounds_1): Remove gcc_unreachable.
17686 2017-03-09 Jakub Jelinek <jakub@redhat.com>
17689 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
17690 BUILT_IN_SYNC*, determine the access type from the size suffix and
17691 always build a MEM_REF with that type. Handle forgotten
17692 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
17695 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
17696 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
17697 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
17698 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
17699 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
17700 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
17701 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
17702 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
17703 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
17704 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
17705 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
17706 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
17707 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
17708 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
17709 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
17710 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
17711 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
17712 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
17713 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
17714 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
17715 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
17716 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
17717 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
17718 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
17719 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
17720 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
17721 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
17722 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
17723 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
17724 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
17725 definitions outside of __OPTIMIZE__ guarded section.
17728 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
17729 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
17730 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
17731 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
17734 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17736 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
17737 ("vfenez<mode>"): Add missing constraints.
17739 2017-03-08 Martin Sebor <msebor@redhat.com>
17742 * config/nds32/nds32.c (nds32_option_override):
17743 Fix misspelled diagnostic.
17745 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17748 * gimplify.c (gimplify_omp_for): Replace index var in outer
17749 taskloop statement with an artificial variable and add
17750 OMP_CLAUSE_PRIVATE clause for it.
17752 2017-03-08 Richard Biener <rguenther@suse.de>
17754 PR tree-optimization/79955
17755 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
17756 for accesses that are completely outside of the variable.
17758 2017-03-08 Andrew Haley <aph@redhat.com>
17760 PR tree-optimization/79943
17761 * tree-ssa-loop-split.c (compute_new_first_bound): When
17762 calculating the new upper bound, (END-BEG) should be added, not
17765 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17767 * config/avr/avr.md (setmemhi): Make sure match_dup
17768 operand number comes before match_scratch.
17770 2017-03-08 Richard Biener <rguenther@suse.de>
17772 PR tree-optimization/79920
17773 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
17774 with ncopies == 1 to ...
17775 (vect_transform_slp_perm_load): ... here. Properly compute
17776 all element loads by iterating VF times over the group. Do
17777 not handle ncopies (computed in a broken way) in
17778 vect_create_mask_and_perm.
17780 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17783 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
17784 is a uniform vector, use uniform_vector_p return value instead of
17785 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
17787 2017-03-07 Marek Polacek <polacek@redhat.com>
17789 PR middle-end/79809
17790 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
17791 (alloca_call_type): Likewise.
17793 2017-03-07 Martin Liska <mliska@suse.cz>
17795 * gcov.c (process_args): Put comment to correct location.
17797 2017-03-07 Martin Liska <mliska@suse.cz>
17799 PR middle-end/68270
17800 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
17801 Use array_at_struct_end_p instead of DECL_CHAIN (field).
17802 (chkp_narrow_bounds_for_field): Likewise.
17803 (chkp_parse_array_and_component_ref): Pass one more argument to
17806 2017-03-07 Richard Biener <rguenther@suse.de>
17808 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
17811 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17813 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
17814 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
17816 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17819 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
17820 to end of description.
17821 (PARAM_MAX_STORES_TO_MERGE): Likewise.
17823 2017-03-07 Jakub Jelinek <jakub@redhat.com>
17825 PR rtl-optimization/79901
17826 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
17828 (*avx512f_<code><mode>3<mask_name>): ... this.
17829 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
17830 iterator instead of VI8_AVX2_AVX512BW.
17832 PR rtl-optimization/79901
17833 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
17834 min/max expander, expand it using expand_vec_cond_expr.
17837 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
17840 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17843 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
17844 to void * for PCH reasons.
17845 * dwarf2out.c (output_loc_operands, output_die): Cast
17846 v.val_vec.array to unsigned char *.
17848 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
17851 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
17854 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
17856 PR rtl-optimization/79571
17857 * lra-constraints.c (process_alt_operands): Calculate static
17858 reject and subtract it from overall when only addresses will be
17861 2017-03-06 Julia Koval <julia.koval@intel.com>
17864 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
17865 incoming stack boundary to 128 for 64-bit targets.
17867 2017-03-06 Richard Biener <rguenther@suse.de>
17869 PR tree-optimization/79894
17870 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
17871 to NULL after folding it.
17873 2017-03-06 Richard Biener <rguenther@suse.de>
17875 PR tree-optimization/79824
17876 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
17877 check disabling peeling for gaps.
17879 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
17881 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17882 attributes): Document gettimeofday.
17884 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17886 * config/s390/s390.c (s390_option_override_internal): Set
17887 PARAM_MIN_VECT_LOOP_BOUND
17889 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17891 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
17892 * config/s390/s390.md: Likewise.
17894 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17897 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
17898 (<avx2_avx512>_perm<mode>): Rename to ...
17899 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
17901 (<avx512>_perm<mode>_mask): Rename to ...
17902 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
17904 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
17905 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
17906 instead of VI8F_256_512.
17907 (avx512f_perm<mode>): New define_expand.
17908 (avx512f_perm<mode>_mask): Likewise.
17909 (avx512f_perm<mode>_1<mask_name>): New define_insn.
17910 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
17912 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17914 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
17915 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
17917 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
17919 2017-03-06 Martin Liska <mliska@suse.cz>
17922 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
17923 when having a SSA NAME w/o VAR_DECL assigned to it.
17925 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17927 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
17928 msa_dpsub_<su>_d): Fix MODE for vec_select.
17930 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17932 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
17934 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
17936 2017-03-06 Richard Biener <rguenther@suse.de>
17938 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
17939 * plugin.c (register_plugin_info): Likewise.
17940 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
17942 2017-03-05 Jakub Jelinek <jakub@redhat.com>
17944 * config/i386/sse.md (sse_storehps, sse_storelps,
17945 avx_<castmode><avxsizesuffix>_<castmode>,
17946 avx512f_<castmode><avxsizesuffix>_<castmode>,
17947 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
17948 in condition that at least one operand is not a MEM.
17950 2017-03-03 Jakub Jelinek <jakub@redhat.com>
17952 PR middle-end/79805
17953 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
17954 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
17956 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
17957 gimple_call_nothrow_p flag based on whether original builtin can throw.
17958 If it can, emit following stmts on the fallthrough edge.
17959 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
17960 don't create new bb if inserting just debug stmts on the edge, try to
17961 insert them on the fallthru bb or just reset debug stmts.
17963 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
17966 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
17967 restore recog_data (including the operand rtxes inside it) around
17968 the call to get_insn_template.
17970 2017-03-03 Martin Sebor <msebor@redhat.com>
17972 PR tree-optimization/79699
17973 * context.c (context::~context): Free MPFR caches to avoid
17974 a memory leak on program exit.
17976 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17978 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
17979 Use wide_int::ulow () instead of .elt (0).
17981 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
17983 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
17984 (*pushxf): Limit oF constraint to 32bit targets and add oC
17985 constraint for 64bit targets.
17986 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
17987 (*pushdf): Change rmF constraint to rmC.
17989 2017-03-03 Martin Liska <mliska@suse.cz>
17991 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
17992 Remove unused variable.
17994 2017-03-03 Jakub Jelinek <jakub@redhat.com>
17997 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
17998 is a memory operand, increase num_memory.
17999 (ix86_expand_args_builtin): Likewise.
18001 2017-03-03 Jan Hubicka <jh@suse.cz>
18004 * ipa-devirt.c (maybe_record_node): Properly handle
18005 __cxa_pure_virtual visibility.
18007 2017-03-03 Martin Liska <mliska@suse.cz>
18009 PR tree-optimization/79803
18010 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18012 (pass_loop_prefetch::execute): Disabled optimization if an
18013 assumption about L1 cache size is not met.
18015 2017-03-03 Martin Liska <mliska@suse.cz>
18017 PR rtl-optimization/79574
18018 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18019 (hash_scan_set): Likewise.
18020 (dump_hash_table): Likewise.
18021 (hoist_code): Likewise.
18023 2017-03-03 Richard Biener <rguenther@suse.de>
18025 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18026 in place of uhigh ().
18027 (fixed_convert_from_real): Likewise.
18029 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18032 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18034 2017-03-03 Richard Biener <rguenther@suse.de>
18036 PR middle-end/79818
18037 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18038 TYPE_OVERFLOW_UNDEFINED check.
18040 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18042 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18044 (vector_ae_<mode>_p): Likewise.
18045 (vector_nez_<mode>_p): Likewise.
18046 (vector_ne_v2di_p): Likewise.
18047 (vector_ae_v2di_p): Likewise.
18048 (vector_ne_<mode>_p): Likewise.
18049 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18051 (vsx_tsqrt<mode>2_fe): Likewise.
18053 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18056 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18058 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18060 PR rtl-optimization/79780
18061 * cprop.c (one_cprop_pass): When second and further conditional trap
18062 in a single basic block is turned into an unconditional trap, turn it
18063 into a deleted note to avoid RTL verification failures.
18065 2017-03-02 Richard Biener <rguenther@suse.de>
18067 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18069 2017-03-02 Richard Biener <rguenther@suse.de>
18071 PR tree-optimization/79345
18073 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18074 param and abort the walk, returning -1 if it is hit.
18075 (walk_aliased_vdefs): Take a limit param and pass it on.
18076 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18077 defaulting to 0 and return a signed int.
18078 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18079 (check_defs): New helper.
18080 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18081 about uninitialized memory.
18082 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18083 bogus uninitialized warning.
18084 (fixed_convert_from_real): Likewise.
18086 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18088 PR tree-optimization/66768
18089 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18090 iv_use if base object can't be determined.
18092 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18094 PR tree-optimization/79345
18095 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18096 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18097 (get_pattern_stats): Initialize it.
18098 * genemit.c (gen_expand): Verify match_scratch numbers come after
18099 match_operand/match_dup numbers.
18100 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18101 match_scratch numbers.
18102 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18104 * config/s390/s390.md (trunctdsd2): Likewise.
18106 2017-03-02 Richard Biener <rguenther@suse.de>
18108 * wide-int.h (wide_int_storage::operator=): Implement in terms
18111 2017-03-02 Richard Biener <rguenther@suse.de>
18113 PR tree-optimization/79777
18114 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18115 the to insert expression to sth existing.
18117 2017-03-01 Martin Sebor <msebor@redhat.com>
18119 PR middle-end/79692
18120 * gimple-ssa-sprintf.c
18121 (directive::known_width_and_precision): New function.
18122 (format_integer): Use it.
18123 (get_mpfr_format_length): Consider the full range of precision
18124 when computing %g output with the # flag. Set the likely byte
18125 count to 3 rather than 1 when precision is indeterminate.
18126 (format_floating): Correct the lower bound of precision.
18128 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18130 * doc/invoke.texi: Document default code model for 64-bit Linux.
18132 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18135 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18136 udiv rather than div since input pattern is unsigned.
18138 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18140 * config/i386/i386.c (print_reg): Warn for values of
18141 unsupported size in integer register.
18143 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18146 * config/rs6000/predicates.md (current_file_function_operand): Do
18147 not allow self calls to be local if the function is replaceable.
18149 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18152 * config/rs6000/altivec.h (vec_ctz and others): Change the
18153 preprocessor macro that controls conditional compilation from
18154 _ARCH_PWR9 to __POWER9_VECTOR__.
18155 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18156 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18157 control (instead of _ARCH_PWR9 control) so that template
18158 definition uses power9-specific function.
18159 (vec_any_eq): Likewise.
18160 (vec_all_ne): Change macro definition to use a power9-specific
18161 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18162 _ARCH_PWR9 control).
18163 (vec_any_eq) Likewise.
18164 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18165 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18166 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18167 support for xvcmpnedp instruction.
18168 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18169 macro expansion so that Power9 implementation of vec_all_ne does
18170 not use the AltiVec predicate framework.
18171 (VCMPNEH_P): Likewise.
18172 (VCMPNEW_P): Likewise.
18173 (VCMPNED_P): Likewise.
18174 (VCMPNEFP_P): Likewise.
18175 (VCMPNEDP_P): Likewise.
18176 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18177 implementation of vec_any_eq to not use AltiVec predicate
18179 (VCMPAEH_P): Likewise.
18180 (VCMPAEW_P): Likewise.
18181 (VCMPAED_P): Likewise.
18182 (VCMPAEFP_P): Likewise.
18183 (VCMPAEDP_P): Likewise.
18184 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18185 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18186 not use the AltiVec predicate framework.
18187 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18188 of vec_any_eq to not use AltiVec predicate framework.
18189 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18190 support for predefined __POWER9_VECTOR__ macro to indicate that
18191 Power9 instruction selection is enabled.
18192 (altivec_overloaded_builtins): Remove extraneous
18193 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18194 function argument types RS6000_BTI_bool_V16QI and
18195 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18196 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18197 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18198 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18199 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18200 Power9 for implementations of vec_cmpne. Change the signature for
18201 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18202 (representing vec_all_ne) to remove the previously described first
18203 argument of type RS6000_BTI_INTSI, as this was an artifact of
18204 reliance on the AltiVec predicate framework, which is no longer
18205 used in the implementation of these functions. Add
18206 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18207 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18208 since, unlike the AltiVec predicate framework implementation, we
18209 do not share function descriptors between vec_alle and vec_anyeq.
18210 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18211 set of modes that receive special treatment even when
18212 TARGET_P9_VECTOR is true. The special treatment emits code that
18213 does not depend on Power9 instructions.
18214 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18215 define_expand to not rely on AltiVec predicate framework.
18216 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18218 (vector_ne_v2di_p): Change this define_expand to not rely on
18219 AltiVec predicate framework.
18220 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18222 (vector_ne_<mode>_p): Change this define_expand to not rely on
18223 AltiVec predicate framework.
18224 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18226 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18227 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18228 define_insn pattern.
18229 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18230 define_insn pattern because the xvcmpne<VSs>. instruction is not
18232 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18233 instruction is not supported.
18235 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18237 * config/nvptx/nvptx.c: Include intl.h.
18239 2017-03-01 Martin Jambor <mjambor@suse.cz>
18242 * ipa-prop.h (ipa_bits): Removed field known.
18243 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18244 to pointers. Adjusted their comments to warn about their sharing.
18245 (ipcp_transformation_summary): Change bits to a vector of pointers.
18246 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18247 (ipa_get_ipa_bits_for_value): Declare.
18248 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18249 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18250 (ipa_bits_hash_table): Likewise.
18251 (ipa_vr_ggc_hash_traits): Likewise.
18252 (ipa_vr_hash_table): Likewise.
18253 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18254 being pointers and vr_known being removed.
18255 (ipa_set_jf_unknown): Likewise.
18256 (ipa_get_ipa_bits_for_value): New function.
18257 (ipa_set_jfunc_bits): Likewise.
18258 (ipa_get_value_range): New overloaded functions.
18259 (ipa_set_jfunc_vr): Likewise.
18260 (ipa_compute_jump_functions_for_edge): Use the above functions to
18261 construct bits and vr parts of jump functions.
18262 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18263 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18265 (ipcp_grow_transformations_if_necessary): Also allocate
18266 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18268 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18269 them. Fix too long lines.
18270 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18271 vr_known being removed.
18272 (ipa_read_jump_function): Use new setter functions to construct bits
18273 and vr parts of jump functions or set them to NULL.
18274 (write_ipcp_transformation_info): Adjust for bits being pointers.
18275 (read_ipcp_transformation_info): Likewise.
18276 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18278 Include gt-ipa-prop.h.
18279 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18281 (ipcp_store_bits_results): Likewise.
18282 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18283 Do not write to existing jump functions but use a temporary instead.
18285 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18288 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18289 attempt to use its first operand as BIT_FIELD_REF base.
18291 2017-03-01 Richard Biener <rguenther@suse.de>
18293 PR middle-end/79721
18294 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18295 interpolating formula in wrapping arithmetic.
18296 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18298 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18300 PR tree-optimization/79734
18301 * tree-vect-generic.c (expand_vector_condition): Optimize
18302 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18303 Handle VEC_COND_EXPR where comparison has different inner width from
18304 type's inner width.
18306 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18308 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18309 markup, and similar issues. Remove @opindex entries for things
18310 that aren't options. Add missing -mmpy-option entries.
18312 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18314 PR tree-optimization/79737
18315 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18316 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18317 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18318 instead of byte_size. Formatting fix.
18319 (shift_bytes_in_array_right): Formatting fix.
18321 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18324 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18325 condition on optimize for the leaf function test.
18327 2017-02-28 Martin Liska <mliska@suse.cz>
18330 * read-rtl-function.c (function_reader::handle_unknown_directive):
18331 Bail out when one uses -flto.
18333 2017-02-28 Martin Liska <mliska@suse.cz>
18335 * common.opt: Replace space with tabular for options of <number>
18337 * config/i386/i386.opt: Show <number> value for
18338 -mlarge-data-threshold.
18339 * opts.c (print_filtered_help): Do not display number in hexadecimal
18342 2017-02-28 Martin Liska <mliska@suse.cz>
18344 * common.opt: Fix --help=option -Q for options which are of
18347 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18349 * config/i386/i386.c (print_reg): Error out for values
18350 of 8-bit size in invalid integer register.
18352 2017-02-28 Martin Sebor <msebor@redhat.com>
18354 PR tree-optimization/79691
18355 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18357 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18360 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18361 gcc_unreachable with output_operand_lossage.
18363 2017-02-28 Richard Biener <rguenther@suse.de>
18365 PR tree-optimization/79740
18366 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18368 (visit_nary_op): Insert the nary into the hashtable if we
18369 pattern-matched sth.
18370 * tree-ssa-pre.c (eliminate_insert): Robustify.
18372 2017-02-28 Richard Biener <rguenther@suse.de>
18374 PR middle-end/79731
18375 * fold-const.c (decode_field_reference): Reject out-of-bound
18378 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18380 * config/i386/i386.c: Include intl.h.
18381 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18382 instead of just cond ? "..." : "...".
18383 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18384 * coverage.c (read_counts_file): Likewise.
18385 * omp-offload.c: Include intl.h.
18386 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18387 of just cond ? "..." : "...".
18388 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18389 of just cond ? "..." : "...".
18391 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18394 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18396 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18397 'tune for' CPU name.
18398 * config/arm/arm-cpu-data.h: Regenerated.
18400 2017-02-28 Richard Biener <rguenther@suse.de>
18402 PR tree-optimization/79732
18403 * tree-inline.c (expand_call_inline): Do not shadow var.
18405 2017-02-28 Richard Biener <rguenther@suse.de>
18407 PR tree-optimization/79723
18408 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18409 address-space properly.
18411 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18413 * doc/optinfo.texi (Optimization groups): Fix option used for
18415 * doc/invoke.texi (-fopt-info): Document "omp".
18416 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18417 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18418 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18419 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18420 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18422 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18424 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18425 OPTGROUP_OMP with "omp".
18427 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18430 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18431 for arithmetic shift of unsigned V2DI.
18433 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18435 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18436 arc/linux.h headers.
18437 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18438 (LINK_SPEC): Likewise.
18439 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18440 (EXTRA_SPECS): Likewise.
18441 (STARTFILE_SPEC): Likewise.
18442 (ENDFILE_SPEC): Likewise.
18443 (LIB_SPEC): Likewise.
18444 (TARGET_SDATA_DEFAULT): Likewise.
18445 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18446 (MULTILIB_DEFAULTS): Likewise.
18447 (DWARF2_UNWIND_INFO): Likewise.
18448 * config/arc/big.h: New file.
18449 * config/arc/elf.h: Likewise.
18450 * config/arc/linux.h: Likewise.
18451 * config/arc/t-uClibc: Remove.
18453 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18455 PR tree-optimization/77536
18456 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18457 (tree_transform_and_unroll_loop): Use above function to compute the
18458 estimated niter of unrolled loop and use it when scaling profile.
18459 Also use count info rather than frequency if it's non-zero.
18460 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18461 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18462 (vect_transform_loop): Call above function.
18464 2017-02-27 Richard Biener <rguenther@suse.de>
18466 PR tree-optimization/45397
18467 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18468 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18469 (visit_nary_op): Add pattern matching for CSEing sign-changed
18470 or truncated operations with wider ones.
18472 2017-02-27 Richard Biener <rguenther@suse.de>
18474 PR tree-optimization/79690
18475 * tree-vect-stmts.c (vectorizable_store): Use vector type
18476 built from the DR with address-space.
18478 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18480 * doc/invoke.texi (Optimize Options): Refine the description
18481 of asan-use-after-return.
18483 2017-02-25 Alan Modra <amodra@gmail.com>
18485 PR rtl-optimization/79584
18486 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18487 base, not ad->base_term, the reg within base. Remove assertion
18488 that ad->base == ad->base_term. Replace gen_int_mode using
18489 bogus mode with const0_rtx.
18491 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18493 PR middle-end/79396
18494 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18495 FMA_EXPR like tcc_binary or tcc_unary.
18497 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18500 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18502 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18503 (output_loc_operands): Handle DW_OP_call_ref and
18504 DW_OP_GNU_variable_value.
18505 (struct variable_value_struct): New type.
18506 (struct variable_value_hasher): Likewise.
18507 (variable_value_hash): New variable.
18508 (string_types): Remove.
18509 (copy_loc_descr): New function.
18510 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18511 (prepend_loc_descr_to_each): New function.
18512 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18513 instead of add_loc_descr_to_each if the first argument is single
18514 location list and the second has multiple.
18515 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18516 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18517 when looking for variable value which doesn't have other location info.
18518 (loc_list_from_tree): Formatting fix.
18519 (gen_array_type_die): Simplify DW_AT_string_length handling.
18520 (adjust_string_types): Remove.
18521 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18522 string_types. Call resolve_variable_values.
18523 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18524 (resolve_addr_in_expr): Likewise. Add A argument.
18525 (copy_deref_exprloc): Remove deref argument. Adjust for the
18526 original expression being DW_OP_GNU_variable_value with optionally
18527 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18528 optionally after it.
18529 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18530 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18531 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18532 (variable_value_hasher::hash, variable_value_hasher::equal): New
18534 (resolve_variable_value_in_expr, resolve_variable_value,
18535 resolve_variable_values, note_variable_value_in_expr,
18536 note_variable_value): New functions.
18537 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18540 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18543 * opts.h (handle_generated_option): Add GENERATED_P argument.
18544 * opts-common.c (handle_option): Adjust function comment.
18545 (handle_generated_option): Add GENERATED_P argument, pass it to
18547 (control_warning_option): Pass false to handle_generated_option
18549 * opts.c (maybe_default_option): Pass true to handle_generated_option
18551 * optc-gen.awk: Likewise.
18553 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18555 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18556 a REG, look at the REG it is a SUBREG of.
18557 (splitter for cmpeqsi_t): Ditto.
18559 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18561 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18562 the special USEs with the pattern of the insn, not the insn itself.
18564 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18567 * doc/invoke.texi: Document -mload-store-pairs.
18569 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18570 Sandra Loosemore <sandra@codesourcery.com>
18572 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18573 argument isn't a CONST_INT.
18574 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18575 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18576 (nios2_validate_compare): Bypass alternate compare logic if *op2
18577 is not a CONST_INT.
18578 (ldstwm_operation_p): Return false if first_base is not a REG or
18579 if first_offset is not a CONST_INT.
18581 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18583 * config/cris/cris.md: Use correct operand in a define_peephole2.
18585 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18587 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18589 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18591 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18592 this_insn if it is an INSN or JUMP_INSN.
18593 (force_offsettable): Look at base, not at addr.
18594 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18595 on things that aren't necessarily CONST_INTs.
18597 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18599 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18600 -mfpmath=sse is the default also for x86-32 targets with SSE2
18601 instruction set when @option{-ffast-math} is enabled
18603 2017-02-24 Jeff Law <law@redhat.com>
18605 PR rtl-optimizatoin/79286
18606 * ira.c (update_equiv_regs): Drop may_trap_p exception to
18609 2017-02-24 Richard Biener <rguenther@suse.de>
18611 PR tree-optimization/79389
18612 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
18615 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
18617 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
18618 function comment to reflect reality.
18619 (loop_exits_before_overflow): Fix typo in function description.
18621 2017-02-24 Richard Biener <rguenther@suse.de>
18623 PR tree-optimization/79389
18624 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
18625 properly that a threading opportunity exists. Detect conditional
18626 copy/constant propagation opportunities.
18628 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
18630 * config/visium/visium.md (type): Add trap.
18631 (b): New mode attribute.
18632 (*btst): Rename into...
18633 (*btst<mode>): ...this and adjust.
18634 (*cbranchsi4_btst_insn): Rename into...
18635 (*cbranch<mode>4_btst_insn): ...this and adjust.
18636 (trap): New define_insn.
18638 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18640 PR tree-optimization/79389
18641 * ifcvt.c (struct noce_if_info): Add rev_cond field.
18642 (noce_reversed_cond_code): New function.
18643 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
18644 reversed_comparison_code. Formatting fix.
18645 (noce_try_store_flag): Test rev_cond != NULL in addition to
18646 reversed_comparison_code.
18647 (noce_try_store_flag_constants): Likewise.
18648 (noce_try_store_flag_mask): Likewise.
18649 (noce_try_addcc): Use rev_cond if non-NULL instead of
18650 reversed_comparison_code.
18651 (noce_try_cmove_arith): Likewise. Formatting fixes.
18652 (noce_try_minmax, noce_try_abs): Clear rev_cond.
18653 (noce_find_if_block): Initialize rev_cond.
18654 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
18655 instead of false as last argument never attempt to reverse it
18658 2017-02-23 Bin Cheng <bin.cheng@arm.com>
18660 PR tree-optimization/79663
18661 * tree-predcom.c (combine_chains): Process refs in reverse order
18662 only for ZERO length chains, and add explaining comment.
18664 2017-02-23 Jeff Law <law@redhat.com>
18666 PR tree-optimization/79578
18667 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
18668 in call to operand_equal_p.
18670 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
18673 * config/i386/cpuid.h: Fix another undefined behavior.
18675 2017-02-23 Richard Biener <rguenther@suse.de>
18677 PR tree-optimization/79683
18678 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
18679 vector types for data-refs.
18681 2017-02-23 Martin Liska <mliska@suse.cz>
18683 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
18685 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18687 PR middle-end/79665
18688 * internal-fn.c (get_range_pos_neg): Moved to ...
18689 * tree.c (get_range_pos_neg): ... here. No longer static.
18690 * tree.h (get_range_pos_neg): New prototype.
18691 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
18692 are known to be in between 0 and signed maximum inclusive, try to
18693 expand both unsigned and signed divmod and use the cheaper one from
18696 2017-02-22 Jeff Law <law@redhat.com>
18698 PR tree-optimization/79578
18699 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
18700 to compare base operands.
18702 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
18705 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
18706 gpc_reg_operand instead of fpr_reg_operand.
18708 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
18710 * config/mips/mips.c (mips_return_in_memory): Force FP
18711 vector types to be returned in memory for o32 ABI.
18713 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18715 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
18716 instead of DW_TAG_member for static data member declarations and don't
18717 set no_linkage_name for static inline data members.
18718 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
18721 2017-02-22 Martin Liska <mliska@suse.cz>
18723 * doc/invoke.texi: Replace inequality signs with square brackets
18726 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18728 PR tree-optimization/68644
18729 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
18731 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
18734 * lra-constraints.c (simplify_operand_subreg): Handle
18735 WORD_REGISTER_OPERATIONS targets.
18737 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18740 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
18741 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
18742 elimination by swapping fld*.
18744 2017-02-22 Richard Biener <rguenther@suse.de>
18746 PR tree-optimization/79673
18747 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
18748 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
18749 irrelevant address-space qualifiers and avoiding a
18750 ADDR_SPACE_CONVERT_EXPR from fold_convert.
18752 2017-02-22 Richard Biener <rguenther@suse.de>
18754 PR tree-optimization/79666
18755 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
18756 to not symbolically negate if that may introduce undefined
18759 2017-02-22 Martin Liska <mliska@suse.cz>
18762 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
18763 * data-streamer-out.c (streamer_write_gcov_count_stream):
18765 * value-prof.c (stream_out_histogram_value): Make assert more
18766 precise based on type of counter.
18768 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
18771 * config/i386/i386.md (standard_x87sse_constant_load splitter):
18772 Use nonimmediate_operand instead of memory_operand for operand 1.
18773 (float-extend standard_x87sse_constant_load splitter): Ditto.
18775 2017-02-21 Jeff Law <law@redhat.com>
18777 PR tree-optimization/79621
18778 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
18779 blocks with edges to themselves.
18781 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18784 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
18785 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
18786 Use gimple_call_builtin_p.
18789 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
18790 on temporarily removed DEBUG_INSNs.
18792 PR tree-optimization/79649
18793 * tree-loop-distribution.c (classify_partition): Give up on
18794 non-generic address space loads/stores.
18796 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
18798 * doc/loop.texi (Loop manipulation): Remove nonexistent
18799 tree_ssa_loop_version from the documentation.
18800 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
18802 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18805 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
18806 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
18807 * config/rs6000/rs6000.c: Include except.h.
18808 (rs6000_expand_split_stack_prologue): Call
18809 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
18811 2017-02-21 Martin Jambor <mjambor@suse.cz>
18814 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
18815 have been analyzed.
18817 2017-02-21 Martin Jambor <mjambor@suse.cz>
18819 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
18820 for backward compatibility only.
18821 * doc/invoke.texi (Option Summary): Remove all references to
18822 -fipa-cp-alignment.
18824 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
18828 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18830 * lra-constraints.c (curr_insn_transform): Handle
18831 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18833 2017-02-21 Martin Liska <mliska@suse.cz>
18835 * config/i386/i386.opt: Replace -masm-dialect with -masm.
18837 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
18839 PR translation/79638
18840 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
18842 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
18845 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
18846 (arm_function_ok_for_sibcall): Return false for an indirect call by
18847 descriptor if all the argument registers are used.
18848 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
18849 alignment of the function.
18851 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18853 PR tree-optimization/61441
18854 * simplify-rtx.c (simplify_const_unary_operation): For
18855 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
18856 the sNaN unmodified.
18858 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18860 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18861 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18862 instead of SYSTEM_HEADER_DIR.
18864 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
18865 Martin Liška <mliska@suse.cz>
18867 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
18868 Fix typos and grammar, use active voice, and clarify.
18870 2017-02-20 Marek Polacek <polacek@redhat.com>
18872 PR middle-end/79537
18873 * gimplify.c (gimplify_expr): Handle unused *&&L;.
18876 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
18878 2017-02-20 Jakub Jelinek <jakub@redhat.com>
18881 * config/i386/i386.c (ix86_expand_builtin): Handle
18882 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
18883 ix86_builtins_isa[fcode].isa as a requirement of those
18884 flags and any other flag in the bitmask.
18885 (ix86_init_mmx_sse_builtins): Use 0 instead of
18886 ~OPTION_MASK_ISA_64BIT as mask.
18887 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
18888 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
18889 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
18890 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
18892 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18895 * lra-constraints.c (split_reg): Check requested split mode
18896 is supported by the register.
18898 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18900 * lra-constraints.c (simplify_operand_subreg): Remove early
18903 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18906 * lra-constraints.c (curr_insn_transform): Tighten condition
18907 for converting SUBREG reloads from OP_OUT to OP_INOUT.
18909 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18912 * lra-constraints.c (curr_insn_transform): Handle
18913 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18915 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
18918 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
18920 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
18922 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
18925 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
18928 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18930 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
18933 2017-02-18 Jakub Jelinek <jakub@redhat.com>
18935 * final.c (last_columnnum, override_columnnum): New variables.
18936 (final_start_function): Set last_columnnum, pass it to begin_prologue
18937 hook and pass 0 to dwarf2out_begin_prologue.
18938 (final_scan_insn): Update override_columnnum. Pass last_columnnum
18939 to source_line debug hook.
18940 (notice_source_line): Compute last_columnnum and for debug_column_info
18941 return true on column changes.
18942 * debug.h (struct gcc_debug_hooks): Add column argument to
18943 source_line and begin_prologue hooks.
18944 (debug_nothing_int_charstar_int_bool): Remove prototype.
18945 (debug_nothing_int_int_charstar,
18946 debug_nothing_int_int_charstar_int_bool): New prototypes.
18947 (dwarf2out_begin_prologue): Add column argument.
18948 * debug.c (do_nothing_debug_hooks): Adjust source_line and
18949 begin_prologue hooks.
18950 (debug_nothing_int_charstar_int_bool): Remove.
18951 (debug_nothing_int_int_charstar,
18952 debug_nothing_int_int_charstar_int_bool): New functions.
18953 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
18954 through to dwarf2out_source_line.
18955 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
18956 (dwarf2out_source_line): Add column argument, emit it if requested.
18957 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
18959 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
18960 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
18961 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
18962 through to dwarf2out_begin_prologue.
18963 (vmsdbgout_source_line): Add column argument, pass it through to
18964 dwarf2out_source_line.
18965 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
18966 dbxout_source_line caller.
18967 (dbxout_source_line): Add column argument.
18969 * common.opt (gno-column-info, gcolumn-info): New options.
18970 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
18971 (check_die): Also test for multiple DW_AT_decl_column attributes.
18972 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
18973 DW_AT_decl_column if requested.
18974 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
18976 (gen_variable_die): Likewise.
18977 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
18978 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
18981 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
18982 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
18983 (ix86_handle_option): Handle OPT_m3dnowa.
18984 * doc/invoke.texi (-m3dnowa): Document.
18985 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
18986 -m3dnowa instead of -m3dnow -march=athlon.
18989 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
18990 instead of gcc_assert for K, r and R code checks. Formatting fixes.
18992 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18995 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
18996 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
18997 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
18998 generator for vsx_xxpermdi_<mode>_be.
18999 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19000 force big-endian semantics.
19001 (vsx_xxpermdi_<mode>_be): New define_expand with same
19002 implementation as previous version of vsx_xxpermdi_<mode>.
19004 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19006 PR tree-optimization/79327
19007 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19008 variable, its initialization and use.
19010 2017-02-17 Julia Koval <julia.koval@intel.com>
19012 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19013 (OPTION_MASK_ISA_PKU_UNSET): New.
19014 (ix86_handle_option): Handle -mrdpid.
19015 * config/i386/cpuid.h (bit_RDPID): New.
19016 * config/i386/driver-i386.c (host_detect_local_cpu):
19017 Detect RDPID feature.
19018 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19019 * config/i386/i386-c.c (ix86_target_macros_internal):
19021 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19022 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19023 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19024 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19025 * config/i386/i386.md (define_insn "rdpid"): New.
19026 * config/i386/i386.opt Add -mrdpid.
19027 * config/i386/immintrin.h (_rdpid_u32): New.
19029 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19031 PR rtl-optimization/79541
19032 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19033 instead of transforming it into USE.
19035 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19037 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19038 If HONOR_SNANS (SFmode) force the input to a register.
19039 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19040 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19041 an frsp or similar insn.
19043 2017-02-17 Martin Liska <mliska@suse.cz>
19045 PR rtl-optimization/79577
19046 * params.def (selsched-max-sched-times): Increase minimum to 1.
19048 2017-02-17 Martin Liska <mliska@suse.cz>
19050 PR rtl-optimization/79574
19051 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19053 2017-02-17 Martin Liska <mliska@suse.cz>
19055 PR tree-optimization/79529
19056 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19057 ssa_defined_default_def_p to handle cases which are implicitly
19059 * tree-ssa.c (ssa_defined_default_def_p): New function.
19060 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19061 which are implicitly defined.
19062 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19064 2017-02-17 Richard Biener <rguenther@suse.de>
19066 PR middle-end/79576
19067 * params.def (max-ssa-name-query-depth): Limit to 10.
19069 2017-02-17 Richard Biener <rguenther@suse.de>
19071 PR tree-optimization/79552
19072 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19075 2017-02-17 Richard Biener <rguenther@suse.de>
19078 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19080 2017-02-17 Marek Polacek <polacek@redhat.com>
19082 PR middle-end/79536
19083 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19084 (fold_negate_expr): New wrapper.
19086 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19088 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19089 Correct terminology and de-emphasize pre-standard behavior.
19091 2017-02-16 Alan Modra <amodra@gmail.com>
19093 PR rtl-optimization/79286
19094 * ira.c (def_dominates_uses): New function.
19095 (update_equiv_regs): Don't create an equivalence for insns that
19096 may trap where the register def does not dominate the use.
19098 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19100 PR rtl-optimization/78127
19101 * lra.c (lra): Call lra_eliminate before finish the loop after
19104 2017-02-16 Richard Biener <rguenther@suse.de>
19106 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19108 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19109 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19110 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19111 (isl_val_int_from_wi): New function.
19112 (extract_affine_gmp): Rename to ...
19113 (extract_affine_wi): ... this, take a widest_int.
19114 (extract_affine_int): Just wrap extract_affine_wi.
19115 (add_param_constraints): Use isl_val_int_from_wi.
19116 (add_loop_constraints): Likewise, and extract_affine_wi.
19118 2017-02-15 Jeff Law <law@redhat.com>
19120 PR middle-end/79521
19121 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19122 ira_init_register_move_cost_if_necessary.
19124 2017-02-15 Martin Sebor <msebor@redhat.com>
19126 PR middle-end/32003
19127 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19128 removed in a prior commit.
19130 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19132 PR tree-optimization/79347
19133 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19134 counters during peeling.
19136 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19138 * Makefile.in (site.exp): Remove "set ISLVER".
19140 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19143 * real.c (real_from_integer): Call real_convert even for decimal.
19145 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19148 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19150 2017-02-14 Andrew Pinski <apinski@cavium.com>
19152 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19153 cores and change the partno/implementer to be correct.
19154 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19155 the 'B" as the implementer.
19156 * config/aarch64/aarch64-tune.md: Regenerate.
19158 2017-02-14 Carl Love <cel@us.ibm.com>
19160 * config/rs6000/rs6000.c: Add case statement entry to make the
19161 xvcvuxdsp built-in argument unsigned.
19162 * config/rs6000/vsx.md: Fix the source and return operand types so they
19163 match the instruction definitions from the ISA document. Fix typo
19164 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19167 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19170 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19171 member early_clobber_alts.
19172 * lra-lives.c (reg_early_clobber_p): New.
19173 (process_bb_lives): Use it.
19174 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19175 (debug_operand_data): Initialize early_clobber_alts.
19176 (setup_operand_alternative): Set up early_clobber_alts.
19177 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19178 alternatives to new_insn_reg.
19179 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19181 (lra_update_insn_regno_info): Pass the new arg.
19183 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19185 PR middle-end/79505
19186 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19187 (new_oacc_loop_raw): Don't clear already cleared fields.
19190 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19191 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19192 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19194 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19197 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19199 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19202 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19203 the extra instruction to the right place to store 128-bit constant
19206 2017-02-14 Martin Sebor <msebor@redhat.com>
19208 PR middle-end/79448
19209 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19210 warning for strings of unknown length.
19212 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19214 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19216 2017-02-14 Jeff Law <law@redhat.com>
19219 * ira-costs.c (scan_one_insn): Initialize register move costs
19220 for pseudos seen in USE/CLOBBER insns.
19222 PR tree-optimization/79095
19223 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19224 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19225 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19226 if the operands are known to be not equal, then the resulting range
19228 (intersect_ranges): If the new range is ~[0,0] and the old range is
19229 wide, then prefer ~[0,0].
19230 * tree-vrp.c (overflow_comparison_p_1): New function.
19231 (overflow_comparison_p): New function.
19232 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19233 if NAME is used in an overflow test.
19234 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19235 overflow check that can be expressed as an equality test, then adjust
19236 ops to be that equality test.
19238 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19240 * config/s390/s390-builtin-types.def: Remove flags argument.
19241 * config/s390/s390.c (s390_init_builtins): Likewise.
19243 2017-02-14 Martin Liska <mliska@suse.cz>
19245 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19246 vector. Fix trailing white spaces.
19248 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19250 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19253 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19255 PR rtl-optimization/68664
19256 * config/arm/arm.c (arm_sched_can_speculate_insn):
19257 New function. Declare prototype.
19258 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19260 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19262 PR rtl-optimization/68664
19263 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19265 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19267 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19269 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19270 max skip bytes for function, loop and jump.
19272 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19274 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19275 ABS_EXPR for gimple dump.
19277 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19280 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19282 PR tree-optimization/79408
19283 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19284 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19285 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19286 also if rhs1 is INTEGER_CST.
19288 2017-02-14 Richard Biener <rguenther@suse.de>
19290 PR middle-end/79432
19291 * tree-into-ssa.c (insert_phi_nodes): When the function can
19292 have abnormal edges rewrite SSA names with broken use-def
19293 dominance out of SSA and register them for PHI insertion.
19295 2017-02-13 Martin Sebor <msebor@redhat.com>
19297 PR middle-end/79496
19298 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19299 clearing info.nowrite flag when snprintf size argument is a range.
19301 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19303 * cprop.c (cprop_jump): Add missing space in string literal.
19304 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19305 (get_constraint_for_component_ref): Likewise.
19306 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19307 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19308 * lra-constraints.c (process_alt_operands): Likewise.
19309 * ipa-inline.c (inline_small_functions): Likewise.
19310 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19311 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19312 * trans-mem.c (diagnose_tm_1_op): Likewise.
19313 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19314 (grid_parallel_clauses_gridifiable): Likewise.
19316 * config/nvptx/mkoffload.c (process): Add space in between
19319 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19320 "MOD4_SSE_REGS" and "ALL_REGS".
19322 * spellcheck.c (test_data): Add , in between "foo" and "food".
19324 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19327 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19328 boundary crossing check and subsequent code generation agree.
19330 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19332 * config/aarch64/aarch64.c (has_memory_op): Delete.
19333 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19336 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19338 PR rtl-optimization/79388
19339 PR rtl-optimization/79450
19340 * combine.c (distribute_notes): When removing TEM_INSN for which
19341 corresponding dest has last value recorded, invalidate that last
19344 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19346 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19347 of explicit '@'. Add missing assembly comment marker on branch costs
19350 2017-02-13 Nathan Sidwell <nathan@acm.org>
19352 * gengtype-lex.l (<in_struct>): Add '/'.
19354 2017-02-13 Martin Liska <mliska@suse.cz>
19357 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19359 2017-02-13 Richard Biener <rguenther@suse.de>
19361 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19363 * configure: Re-generate.
19364 * config.in: Likewise.
19365 * graphite-dependences.c: Simplify as if
19366 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19367 * graphite-isl-ast-to-gimple.c: Likewise.
19368 * graphite-optimize-isl.c: Likewise.
19369 * graphite-poly.c: Likewise.
19370 * graphite-sese-to-poly.c: Likewise.
19371 * graphite.h: Likewise.
19372 * toplev.c: Include isl/version.h and use isl_version () for
19373 printing the ISL version.
19374 * doc/install.texi: Update ISL requirement.
19376 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19378 * doc/standards.texi (Standards): Update reference to
19381 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19383 * doc/extend.texi (Named Address Spaces): sourceware.org now
19385 * doc/install.texi (Binaries): Ditto.
19388 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19390 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19391 standard terminology "stringize"/"stringizing" throughout.
19392 * doc/cppinternals.texi: Likewise.
19394 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19396 * doc/extend.texi: Fix some spelling mistakes and typos.
19397 * doc/invoke.texi: Likewise.
19399 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19402 * params.def (inline-min-speedup) Change from 10 to 8.
19404 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19406 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19409 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19412 * ipa-inline-analysis.c (get_minimal_bb): New function.
19413 (record_modified): Use it.
19414 (remap_edge_change_prob): Handle also ancestor functions.
19416 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19418 * doc/contrib.texi (Contributors): Remove broken link into
19419 the Mauve CVS repository.
19421 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19423 PR middle-end/79454
19424 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19425 result computation whenever lhs doesn't have vector mode, not
19426 just when it has BLKmode.
19428 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19430 * doc/makefile.texi (profiledbootstrap): Refer to the
19431 installation instructions only in textual form.
19433 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19436 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19438 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19440 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19441 (Specific): Update mingw-w64 reference.
19443 (Specific): Remove broken link to Renesas RX processor.
19445 2017-02-10 Richard Biener <rguenther@suse.de>
19447 * toplev.c (process_options): Do not mention obsolete graphite
19448 options when printing sorry message about missing graphite support.
19449 Mention -floop-nest-optimize.
19451 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19453 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19454 (vtst_p16): Likewise.
19455 (vtstq_p8): Likewise.
19456 (vtstq_p16): Likewise.
19458 (vtstq_p64): Likewise.
19459 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19460 (vset_lane_p64): New.
19461 (vsetq_lane_p64): New.
19463 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19465 PR tree-optimization/79411
19466 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19467 stmt operands are SSA_NAMEs used in abnormal phis.
19468 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19471 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19474 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19477 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19479 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19481 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19483 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19487 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19488 not arbitrary TREE_CONSTANT.
19491 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19492 "omp declare target link" attribute unless is_global_var.
19493 * omp-offload.c (find_link_var_op): Likewise.
19495 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19496 Chung-Lin Tang <cltang@codesourcery.com>
19498 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19500 (gimplify_adjust_omp_clauses): Don't delete TILE.
19501 (gimplify_omp_for): Deal with TILE.
19502 * internal-fn.c (expand_GOACC_TILE): New function.
19503 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19505 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19506 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19508 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19509 avoid DIV for outermost collapse var.
19510 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19511 Remove out of date comments, fix whitespace.
19512 * omp-general.c (omp_extract_for_data): Deal with tiling.
19513 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19514 adjust OLF_DIM_BASE value.
19515 (struct omp_for_data): Add tiling field.
19516 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19517 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19518 for auto loops. Remove default auto determining, moved to
19519 oacc_loop_fixed_partitions.
19520 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19521 stmts, add e_mask field.
19522 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19523 (oacc_thread_numbers): Use oacc_dim_call.
19524 (oacc_xform_tile): New.
19525 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19526 (finish_oacc_loop): Adjust for ifns vector.
19527 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19528 add case for GOACC_TILE fns.
19529 (oacc_loop_xform_loop): Delete.
19530 (oacc_loop_process): Iterate over call list directly, and add
19531 handling for GOACC_TILE fns.
19532 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19534 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19535 vector partitioning to outer loops. Assign 2 partitions to loops
19536 when available. Add TILE handling.
19537 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19538 (execite_oacc_device_lower): Process GOACC_TILE fns,
19539 ignore unknown specs.
19540 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19541 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19542 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19544 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19546 * configure.ac (ACX_BUGURL): Update.
19547 * configure: Regenerate.
19549 2017-02-09 Richard Biener <rguenther@suse.de>
19551 PR tree-optimization/69823
19552 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19553 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19555 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19557 * config/arc/arc-c.def: Add __NPS400__ definition.
19558 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19559 (TARGET_NPS400): Define.
19561 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19563 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19565 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19566 pointer, arch_info.
19567 (arc_cpu_types): Fill the arch_info field with a pointer into the
19568 arc_arch_types table.
19569 (arc_selected_cpu): Declare.
19570 * config/arc/arc.c (arc_selected_cpu): Make global.
19571 (arc_selected_arch): Delete.
19572 (arc_base_cpu): Delete.
19573 (arc_override_options): Remove references to deleted variables,
19574 update access to arch information.
19575 (ARC_OPT): Update access to arch information.
19576 (ARC_OPTX): Likewise.
19577 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19578 (TARGET_ARC600): Update access to arch information.
19579 (TARGET_ARC601): Likewise.
19580 (TARGET_ARC700): Likewise.
19581 (TARGET_EM): Likewise.
19582 (TARGET_HS): Likewise.
19583 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19586 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19589 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19590 condition/operands for integer GE/LE/GEU/LEU operations.
19592 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19594 PR translation/79397
19595 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19598 2017-02-08 Martin Jambor <mjambor@suse.cz>
19601 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19602 whether allocation happened.
19603 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19604 nothing was allocated.
19606 2017-02-08 Jakub Jelinek <jakub@redhat.com>
19608 PR tree-optimization/79408
19609 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
19610 constant, but SSA_NAME with a known integer range, use the minimum
19611 of that range instead of op1 to determine if modulo can be replaced
19612 with its first operand.
19614 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19616 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
19618 2017-02-08 Richard Biener <rguenther@suse.de>
19620 PR tree-optimization/71824
19621 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19622 Check all loops contained in the merged region.
19624 2017-02-07 Andrew Pinski <apinski@cavium.com>
19626 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
19628 2017-02-07 Andrew Pinski <apinski@cavium.com>
19630 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
19631 (thunderxt88): Likewise.
19632 (thunderxt81): Disable LSE and change v8.1 to v8.
19633 (thunderxt83): Likewise.
19635 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19636 Richard Biener <rguenther@suse.de>
19638 PR middle-end/79399
19639 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
19640 type from int to size_t.
19641 * ira-costs.c (struct_costs_size): Change type from int to size_t.
19643 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19645 PR rtl-optimization/79386
19646 * cprop.c (bypass_conditional_jumps): Initialize
19647 bypass_last_basic_block already before splitting bbs after
19648 unconditional traps...
19649 (bypass_conditional_jumps): ... rather than here.
19652 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
19653 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
19654 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
19655 fix -masm=intel patterns.
19657 2017-02-07 Richard Biener <rguenther@suse.de>
19659 PR tree-optimization/79256
19660 PR middle-end/79278
19661 * builtins.c (get_object_alignment_2): Use min_align_of_type
19662 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
19663 and ADJUST_FIELD_ALIGN.
19665 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
19667 * doc/tm.texi: Regenerate.
19668 * stor-layout.c (layout_decl): Adjust.
19669 (update_alignment_for_field): Likewise.
19670 (place_field): Likewise.
19671 (min_align_of_type): Likewise.
19672 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
19673 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
19674 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
19675 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
19676 * config/frv/frv.c (frv_adjust_field_align): Likewise.
19677 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
19678 * config/i386/i386.c (x86_field_alignment): Likewise.
19679 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
19680 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
19681 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
19682 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
19683 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
19684 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
19688 2017-01-30 Richard Biener <rguenther@suse.de>
19690 PR tree-optimization/79256
19691 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
19692 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
19695 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
19697 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
19698 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
19699 builtins to SImode and emit a zero-extend, if necessary.
19701 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19703 * docs/invoke.texi (RISC-V Options): Alphabetize.
19705 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19707 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
19710 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19712 * config/riscv/riscv.c: New file.
19713 * common/config/riscv/riscv-common.c: Likewise.
19714 * config.gcc: Likewise.
19715 * config/riscv/constraints.md: Likewise.
19716 * config/riscv/elf.h: Likewise.
19717 * config/riscv/generic.md: Likewise.
19718 * config/riscv/linux.h: Likewise.
19719 * config/riscv/multilib-generator: Likewise.
19720 * config/riscv/peephole.md: Likewise.
19721 * config/riscv/pic.md: Likewise.
19722 * config/riscv/predicates.md: Likewise.
19723 * config/riscv/riscv-builtins.c: Likewise.
19724 * config/riscv/riscv-c.c: Likewise.
19725 * config/riscv/riscv-ftypes.def: Likewise.
19726 * config/riscv/riscv-modes.def: Likewise.
19727 * config/riscv/riscv-opts.h: Likewise.
19728 * config/riscv/riscv-protos.h: Likewise.
19729 * config/riscv/riscv.h: Likewise.
19730 * config/riscv/riscv.md: Likewise.
19731 * config/riscv/riscv.opt: Likewise.
19732 * config/riscv/sync.md: Likewise.
19733 * config/riscv/t-elf-multilib: Likewise.
19734 * config/riscv/t-linux: Likewise.
19735 * config/riscv/t-linux-multilib: Likewise.
19736 * config/riscv/t-riscv: Likewise.
19737 * configure.ac: Likewise.
19738 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
19739 Waterman as RISC-V maintainers.
19740 * doc/install.texi: Add RISC-V entries.
19741 * doc/invoke.texi: Add RISC-V options section.
19742 * doc/md.texi: Add RISC-V constraints section.
19743 * configure: Regenerated.
19745 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19748 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
19749 false values to be constant vectors with all 0 or all 1 bits set.
19750 (vcondu<mode><mode>): Likewise.
19751 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
19753 (fpmask_comparison_operator): Update comment.
19754 (vecint_comparison_operator): New predicate.
19755 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
19756 vector conditionals when the true and false values are constant
19757 vectors with all 0 bits or all 1 bits set.
19759 2017-02-06 Martin Sebor <msebor@redhat.com>
19761 PR tree-optimization/79376
19762 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
19764 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
19766 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
19767 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
19768 to simplify split condition.
19770 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19772 * omp-expand.c (oxpand_omp_atomic_fetch_op,
19773 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
19776 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
19778 PR rtl-optimization/68664
19779 * target.def (can_speculate_insn): New hook.
19780 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
19781 * doc/tm.texi: Regenerate.
19782 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
19783 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
19784 (rs6000_sched_can_speculate_insn): New function.
19786 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19788 PR tree-optimization/79284
19789 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
19790 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
19791 vectorizable_mask_load_store, vectorizable_operation,
19792 vect_is_simple_cond, get_same_sized_vectype): Use it instead
19793 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
19794 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
19795 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
19796 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
19797 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19798 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
19799 is_gimple_assign (stmt). Replace another such test with
19800 is_gimple_assign (stmt).
19802 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19805 * config/avr/avr.c (rtl-iter.h): Include it.
19806 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
19807 (avr_legitimate_combined_insn): ...and implementation.
19809 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
19811 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
19812 * config/s390/s390.c (s390_const_operand_ok)
19813 (s390_canonicalize_comparison, s390_extract_part)
19814 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
19815 (s390_contiguous_bitmask_p, s390_rtx_costs)
19816 (legitimize_pic_address): Likewise.
19817 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
19818 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
19819 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
19820 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
19821 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
19823 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19825 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
19826 REGNO($0) == REGNO($1).
19828 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19830 * config/s390/linux.h(SIZE_TYPE): Add comment.
19832 2017-02-06 Julian Brown <julian@codesourcery.com>
19833 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19834 Virendra Pathak <virendra.pathak@broadcom.com>
19836 * config/aarch64/aarch64-cores.def: Change the scheduler
19838 * config/aarch64/aarch64.md: Include thunderx2t99.md.
19839 * config/aarch64/thunderx2t99.md: New file.
19841 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19843 * doc/standards.texi (Go Language): Update link to language
19846 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19848 * tree-eh.c (lower_resx): Sanitize profile.
19849 (cleanup_empty_eh_move_lp): Likewise.
19851 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19854 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
19856 * cfgloopmanip.h (loop_version): Update prototype.
19857 * modulo-sched.c (sms_schedule): Update call of loop_version.
19858 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
19859 * tree-parloops.c (gen_parallel_loop): Likewise.
19860 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
19861 * tree-ssa-loop-split.c (split_loop): Likewise.
19862 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
19863 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19865 2017-02-05 Martin Liska <mliska@suse.cz>
19868 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
19870 (print_operand_address): Initialize a struct to zero.
19872 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19874 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
19875 garbage collector only in textual form.
19877 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19879 * doc/extend.texi (x86 specific memory model extensions for
19880 transactional memory): Simplify a phrase.
19882 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
19885 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
19886 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
19887 (atomic_storedi_1): Likewise.
19889 2017-02-04 Jakub Jelinek <jakub@redhat.com>
19891 PR tree-optimization/79338
19892 * tree-parloops.c (gather_scalar_reductions): Don't call
19893 vect_analyze_loop_form for loop->inner before destroying loop's
19896 2017-02-03 Martin Sebor <msebor@redhat.com>
19898 PR tree-optimization/79327
19899 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
19900 when precision has resulted in leading zeros.
19901 (format_integer): Adjust the likely counter to assume an unknown
19902 argument that may be zero is non-zero.
19904 2017-02-03 Jason Merrill <jason@redhat.com>
19907 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
19908 avoid copying non-taken branch.
19910 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19912 PR tree-optimization/79340
19913 * tree-vect-loop.c (vectorizable_reduction): Release
19914 vec_defs elements after safe_splicing them into other vectors.
19917 PR tree-optimization/79327
19918 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
19919 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
19921 (format_integer): Use wide_int_to_tree instead of build_int_cst
19922 + to_?hwi. If argmin is NULL, just set argmin and argmax to
19923 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
19924 of shortest and longest sequence.
19926 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
19928 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
19929 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
19931 2017-02-03 Walter Lee <walt@tilera.com>
19934 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
19935 after initial stackframe link reg save.
19936 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
19938 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19941 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
19942 wu for stxssp alternative.
19944 2017-02-03 Martin Sebor <msebor@redhat.com>
19946 PR tree-optimization/79352
19947 * gimple-fold.c (get_range_strlen): Add argument.
19948 (get_range_strlen): Change return type to bool.
19949 (get_maxval_strlen): Pass in a dummy argument.
19950 * gimple-fold.h (get_range_strlen): Change return type to bool.
19951 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
19952 * tree.h (array_at_struct_end_p): Add argument.
19953 * tree.c (array_at_struct_end_p): Handle it.
19955 2017-02-03 Martin Liska <mliska@suse.cz>
19958 * multiple_target.c (create_dispatcher_calls): Redirect edge
19959 from a caller of a dispatcher.
19960 (expand_target_clones): Make the clones local.
19961 (ipa_target_clone): Do both target clones and resolvers.
19962 (ipa_dispatcher_calls): Remove the pass.
19963 (pass_dispatcher_calls::gate): Likewise.
19964 (make_pass_dispatcher_calls): Likewise.
19965 * passes.def (pass_target_clone): Put as very first IPA early
19968 2017-02-03 Martin Liska <mliska@suse.cz>
19970 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
19971 in case of a function with ifunc attribute.
19973 2017-02-03 Martin Liska <mliska@suse.cz>
19975 * cgraph.c (cgraph_node::dump): Dump function version info.
19976 * symtab.c (symtab_node::dump_base): Add missing new line.
19978 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
19980 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
19981 (ifcombine_ifandif): Use it.
19983 2017-02-03 Martin Liska <mliska@suse.cz>
19985 * doc/invoke.texi: Document default value for
19986 use-after-scope-direct-emission-threshold.
19988 2017-02-03 Martin Liska <mliska@suse.cz>
19990 PR tree-optimization/79339
19991 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
19992 (format_floating): Likewise.
19994 2017-02-03 Martin Liska <mliska@suse.cz>
19997 * ipa-prop.c (ipa_node_params_t::insert): Remove current
19999 (ipa_node_params_t::remove): Likewise.
20000 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20001 initialization from removed ipa_node_params_t::insert.
20002 (ipa_node_params::~ipa_node_params): Move from removed
20003 ipa_node_params_t::release.
20004 * symbol-summary.h (symbol_summary::m_released): New member.
20005 Do not release a summary twice. Do not allow to call finalizer
20006 for types of a summary that live in GGC memory.
20008 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20010 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20013 2017-02-02 Martin Sebor <msebor@redhat.com>
20015 PR middle-end/79275
20016 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20017 (format_string): Tighten up the range of output for non-constant
20018 strings and correct the expected range for wide non-constant strings.
20020 2017-02-02 Martin Sebor <msebor@redhat.com>
20022 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20024 PR middle-end/32003
20025 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20027 (-fdump-tree-@var): Add to index and document how to come up
20028 with pass-specific option and dump file names.
20029 (-fdump-passes): Clarify where to look for output.
20031 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20033 PR middle-end/77445
20034 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20035 statistics of the analyzed path; allow threading for speed when
20036 any of BBs along the path are optimized for speed.
20038 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20040 PR middle-end/78468
20041 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20042 settings of the virtual registers.
20045 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20047 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20048 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20051 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20053 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20054 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20056 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20058 * config/s390/s390.md: Add missing comments with the expanded
20060 * config/s390/vector.md: Likewise.
20061 * config/s390/vx-builtins.md: Likewise.
20063 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20066 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20067 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20068 conditions on a single line.
20070 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20072 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20073 __S390_VX__ to __VX__.
20075 2017-02-01 Andrew Pinski <apinski@cavium.com>
20077 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20078 stmt_info to record_stmt_cost.
20079 (vect_get_known_peeling_cost): Pass stmt_info if known to
20081 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20082 cpu_vector_cost field into
20083 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20084 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20085 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20086 splitting of scalar_stmt_cost and vec_stmt_cost.
20087 (thunderx_vector_cost): Likewise.
20088 (cortexa57_vector_cost): LIkewise.
20089 (exynosm1_vector_cost): Likewise.
20090 (xgene1_vector_cost): Likewise.
20091 (thunderx2t99_vector_cost): Improve after the splitting of the two
20093 (aarch64_builtin_vectorization_cost): Update for the splitting of
20094 scalar_stmt_cost and vec_stmt_cost.
20096 2017-02-01 Torvald Riegel <triegel@redhat.com>
20097 Richard Henderson <rth@redhat.com>
20099 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20100 conditional on existance of a fast atomic load.
20101 * optabs-query.c (can_atomic_load_p): New function.
20102 * optabs-query.h (can_atomic_load_p): Declare it.
20103 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20104 no fast atomic load is available for the particular size of access.
20105 (expand_atomic_compare_and_swap): Likewise.
20106 (expand_atomic_load): Likewise.
20107 (expand_atomic_store): Likewise.
20108 (expand_atomic_fetch_op): Likewise.
20109 * testsuite/lib/target-supports.exp
20110 (check_effective_target_sync_int_128): Remove x86 because it provides
20111 no fast atomic load.
20112 (check_effective_target_sync_int_128_runtime): Likewise.
20114 2017-02-01 Richard Biener <rguenther@suse.de>
20116 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20117 (graphite_transform_loops): Provide opt-info for optimized nests.
20118 * tree-parloop.c (parallelize_loops): Provide opt-info for
20119 parallelized loops.
20121 2017-02-01 Richard Biener <rguenther@suse.de>
20123 PR middle-end/79315
20124 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20125 was not set before.
20127 2017-02-01 Richard Biener <rguenther@suse.de>
20129 PR tree-optimization/71824
20130 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20131 Verify the loops are valid in the merged SESE region.
20132 (scop_detection::can_represent_loop_1): Check analyzing the
20133 evolution of the number of iterations in the region succeeds.
20135 2017-01-31 Ian Lance Taylor <iant@golang.org>
20137 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20138 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20140 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20142 PR preprocessor/79210
20143 * input.c (get_substring_ranges_for_loc): Replace line_width
20144 assertion with error-handling.
20146 2017-01-31 Richard Biener <rguenther@suse.de>
20148 PR tree-optimization/77318
20149 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20150 (create_pw_aff_from_tree): Take loop parameter.
20151 (add_condition_to_pbb): Pass loop of the condition to
20152 create_pw_aff_from_tree.
20154 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20156 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20157 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20159 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20163 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20165 (convert_int_to_float128): Likewise.
20166 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20167 (convert_int_to_float128): Likewise.
20168 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20169 (UNSPEC_IEEE128_CONVERT): Likewise.
20170 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20171 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20172 Use local variables for IBM extended format.
20173 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20174 (fix_trunc<mode>si2_fprs): Likewise.
20175 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20176 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20177 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20178 to know that we can now have integers of all sizes in vector
20180 (fix<uns>_<mode>di2_hw): Likewise.
20181 (float<uns>_<mode>si2_hw): Likewise.
20182 (fix_<mode>si2_hw): Likewise.
20183 (fixuns_<mode>si2_hw): Likewise.
20184 (float<uns>_<mode>di2_hw): Likewise.
20185 (float_<mode>di2_hw): Likewise.
20186 (float_<mode>si2_hw): Likewise.
20187 (floatuns_<mode>di2_hw): Likewise.
20188 (floatuns_<mode>si2_hw): Likewise.
20189 (xscvqp<su>wz_<mode>): Delete, no longer used.
20190 (xscvqp<su>dz_<mode>): Likewise.
20191 (xscv<su>dqp_<mode>): Likewise.
20192 (ieee128_mfvsrd_64bit): Likewise.
20193 (ieee128_mfvsrd_32bit): Likewise.
20194 (ieee128_mfvsrwz): Likewise.
20195 (ieee128_mtvsrw): Likewise.
20196 (ieee128_mtvsrd_64bit): Likewise.
20197 (ieee128_mtvsrd_32bit): Likewise.
20199 2017-01-31 Martin Liska <mliska@suse.cz>
20202 * ipa-prop.c (ipa_free_all_node_params): Call release method
20203 instead of ~sumbol_summary to not to trigger double times
20206 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20208 PR tree-optimization/71691
20209 * bitmap.h (class auto_bitmap): New.
20210 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20211 is_maybe_undefined instead of ssa_undefined_value_p.
20213 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20215 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20216 __S390_ARCH_LEVEL__ to __ARCH__.
20218 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20220 PR tree-optimization/79267
20221 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20222 if should_remove_lhs_p is true.
20224 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20227 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20228 (add_alignment_attribute): New.
20229 (base_type_die): Add alignment attribute.
20230 (subrange_type_die): Likewise.
20231 (modified_type_die): Likewise.
20232 (gen_array_type_die): Likewise.
20233 (gen_descr_array_type_die: Likewise.
20234 (gen_enumeration_type_die): Likewise.
20235 (gen_subprogram_die): Likewise.
20236 (gen_variable_die): Likewise.
20237 (gen_field_die): Likewise.
20238 (gen_ptr_to_mbr_type_die): Likewise.
20239 (gen_struct_or_union_type_die): Likewise.
20240 (gen_subroutine_type_die): Likewise.
20241 (gen_typedef_die): Likewise.
20242 (base_type_cmp): Compare alignment attribute.
20244 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20247 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20248 (setb_unsigned) New pattern for setb with CCUNS.
20249 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20250 subfc./subfe sequence to avoid overflow problems. Generate a
20251 shorter sequence with cmpld/setb for power9.
20252 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20253 for generating subfc. instruction.
20254 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20255 now uses this instruction.
20257 2017-01-30 Ian Lance Taylor <iant@google.com>
20260 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20261 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20263 2017-01-30 Martin Sebor <msebor@redhat.com>
20265 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20266 Move constant to the right of a relational operator.
20267 (get_mpfr_format_length, format_character, format_string): Ditto.
20268 (should_warn_p, maybe_warn): Same.
20270 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20272 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20275 * asan.c (get_translation_unit_decl): Remove function.
20276 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20278 2017-01-30 Martin Liska <mliska@suse.cz>
20280 PR gcov-profile/79259
20281 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20282 -fprofile-generate.
20284 2017-01-30 Martin Liska <mliska@suse.cz>
20287 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20288 Initialize variables with NULL value.
20290 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20293 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20295 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20297 2017-01-30 Richard Biener <rguenther@suse.de>
20299 PR tree-optimization/79276
20300 * tree-vrp.c (process_assert_insertions): Properly adjust common
20301 when removing a duplicate.
20303 * gcc.dg/torture/pr79276.c: New testcase.
20305 2017-01-30 Richard Biener <rguenther@suse.de>
20307 PR tree-optimization/79256
20308 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20309 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20311 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20313 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20316 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20317 ("*r<noxa>sbg_<mode>_sll_bitmask")
20318 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20319 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20320 Use contiguous_bitmask_nowrap_operand.
20322 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20325 * config/rs6000/altivec.h (vec_xl): Revise #define.
20326 (vec_xst): Likewise.
20328 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20330 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20332 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20334 PR rtl-optimization/79194
20335 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20336 traps before call to bypass_conditional_jumps.
20338 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20340 PR tree-optimization/71374
20341 * lra-constraints.c (check_conflict_input_operands): New.
20342 (match_reload): Use it.
20344 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20347 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20348 account to calculate conflict_set.
20350 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20352 PR rtl-optimization/78559
20353 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20354 other_insn in combine.
20356 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20358 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20359 uint16_type_node for BT_UINT16.
20361 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20363 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20364 "RTL Tests" to menu.
20365 (GIMPLE Tests): New node.
20366 (RTL Tests): New node.
20368 2017-01-27 Richard Biener <rguenther@suse.de>
20370 PR tree-optimization/79245
20371 * tree-loop-distribution.c (distribute_loop): Apply cost
20372 modeling also to detected patterns.
20374 2017-01-27 Richard Biener <rguenther@suse.de>
20376 PR tree-optimization/71433
20377 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20378 (compare_assert_loc): New function.
20379 (process_assert_insertions): Sort and optimize assert locations
20380 to remove duplicates and push down identical assertions on
20381 edges to their destination block.
20383 2017-01-27 Richard Biener <rguenther@suse.de>
20385 PR tree-optimization/79244
20386 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20387 out SSA names even if abnormal.
20389 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20391 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20392 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20393 instead of MPFR_RNDN.
20395 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20398 * arm.c (arm_option_override): Don't call build_target_option_node
20399 until after doing all option overrides.
20400 (arm_valid_target_attribute_tree): Likewise.
20402 2017-01-27 Martin Liska <mliska@suse.cz>
20404 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20405 for {cd}tors and C++ {cd}tors.
20407 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20409 * config/s390/s390.md ("*setmem_long_and")
20410 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20412 2017-01-26 Martin Sebor <msebor@redhat.com>
20414 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20417 2017-01-26 Martin Sebor <msebor@redhat.com>
20419 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20420 HAVE_DFmode before using XFmode or DFmode.
20421 (parse_directive): Avoid using the z length modifier to avoid
20422 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20424 PR middle-end/78703
20425 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20426 to accept adjustment as an array.
20427 (get_int_range): New function.
20428 (struct directive): Make width and prec arrays.
20429 (directive::set_width, directive::set_precision): Call get_int_range.
20430 (format_integer, format_floating): Handle width and precision ranges.
20431 (format_string, parse_directive): Same.
20433 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20436 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20437 comdat_type_p set, just clone them, but keep the children in the
20441 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20442 which have direct callers with -fvar-tracking-assignments enabled
20444 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20445 inside of type units.
20447 2017-01-26 Martin Sebor <msebor@redhat.com>
20449 PR middle-end/78703
20450 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20452 (struct format_result): Replace number_chars, number_chars_min,
20453 and number_chars_max with a single member of struct result_range.
20455 (format_result::operator+=): Adjust.
20456 (struct fmtresult): Remove bounded. Handle likely and unlikely
20458 (fmtresult::adjust_for_width_or_precision): New function.
20459 (fmtresult:type_max_digits): New function.
20460 (bytes_remaining): Handle likely and unlikely counters.
20461 (min_bytes_remaining): Remove.
20462 (format_percent): Simplify.
20463 (format_integer, format_floating): Set likely and unlikely counters.
20464 (get_string_length, format_character, format_string): Same.
20465 (format_plain, should_warn_p): New function.
20466 (maybe_warn): Call should_warn_p. Update diagnostic messages
20467 and handle those for all directives, including plain strings.
20468 (format_directive): Handle likely and unlikely counters.
20469 Remove unnecessary quoting from diagnostics. Add an informational
20471 (add_bytes): Remove.
20472 (pass_sprintf_length::compute_format_length): Simplify.
20473 (try_substitute_return_value): Handle likely and unlikely counters.
20475 2017-01-26 Carl Love <cel@us.ibm.com>
20477 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20478 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20480 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20483 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20484 endianess for subregs into account.
20485 * lra-constraints.c (lra_constraints): Do risky transformations
20486 always on the first iteration.
20487 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20488 last_call_used_reg_set.
20489 (process_bb_lives): Define and use last_call_used_reg_set.
20490 * lra.c (lra): Always continue after lra_constraints on the first
20493 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20495 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20497 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20499 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20501 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20502 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20503 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20504 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20505 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20506 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20507 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20508 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20509 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20511 2017-01-26 Marek Polacek <polacek@redhat.com>
20514 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20515 for the third operand.
20517 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20519 PR middle-end/79236
20520 * omp-low.c (struct omp_context): Add simt_stmt field.
20521 (scan_omp_for): Return omp_context *.
20522 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20523 context to the _simt_ SIMD stmt.
20524 (lower_omp_for): For combined SIMD with sibling _simt_
20525 SIMD, make sure to use the same decls in _looptemp_
20526 clauses as in the sibling.
20528 2017-01-26 David Sherwood <david.sherwood@arm.com>
20530 PR middle-end/79212
20531 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20534 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20537 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20538 emit fld b; fld a; if possible.
20540 * brig-builtins.def: Update copyright years.
20541 * config/arm/arm_acle_builtins.def: Update copyright years.
20543 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20546 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20547 constraint instead of o for the stxsd instruction.
20549 2017-01-25 Carl Love <cel@us.ibm.com>
20551 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20552 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20554 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20556 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20558 2017-01-25 Richard Biener <rguenther@suse.de>
20560 PR tree-optimization/69264
20561 * target.def (vector_alignment_reachable): Improve documentation.
20562 * doc/tm.texi: Regenerate.
20563 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20565 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20566 earlier changes with respect to TYPE_USER_ALIGN.
20567 (vector_alignment_reachable_p): Likewise. Improve dumping.
20569 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20572 * config/arm/arm.md (xordi3): Force constant operand into a register
20575 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20577 * doc/invoke.texi (-fstore-merging): Correct default optimization
20578 levels at which it is enabled.
20579 (-O): Move -fstore-merging from list to...
20582 2017-01-25 Richard Biener <rguenther@suse.de>
20585 * omp-expand.c: Include debug.h.
20586 (expand_omp_taskreg): Make sure to generate early debug before
20587 outlining anything from a function.
20588 (expand_omp_target): Likewise.
20589 (grid_expand_target_grid_body): Likewise.
20591 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20594 * asan.c (get_translation_unit_decl): New function.
20595 (asan_add_global): Extract modules file name from globals
20596 TRANSLATION_UNIT_DECL name.
20598 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20601 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20602 for long calls with APCS frame and VFP.
20604 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20606 * cfg.c (original_copy_tables_initialized_p): New function.
20607 * cfg.h (original_copy_tables_initialized_p): New decl.
20608 * cfgrtl.c (relink_block_chain): Guard the call to
20609 free_original_copy_tables with a call to
20610 original_copy_tables_initialized_p.
20611 * cgraph.h (symtab_node::native_rtl_p): New decl.
20612 * cgraphunit.c (symtab_node::native_rtl_p): New function.
20613 (symtab_node::needed_p): Don't assert for early assembly output
20614 for __RTL functions.
20615 (cgraph_node::finalize_function): Set "force_output" for __RTL
20617 (cgraph_node::analyze): Bail out early for __RTL functions.
20618 (analyze_functions): Update assertion to support __RTL functions.
20619 (cgraph_node::expand): Bail out early for __RTL functions.
20620 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
20622 * function.h (struct function): Update comment for field
20624 * gimple-expr.c: Include "tree-pass.h".
20625 (gimple_has_body_p): Return false for __RTL functions.
20626 * Makefile.in (OBJS): Add run-rtl-passes.o.
20627 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
20629 (gcc::pass_manager::get_clean_slate): New accessor.
20630 * passes.c: Include "insn-addr.h".
20631 (should_skip_pass_p): Add logging. Update logic for running
20632 "expand" to be compatible with both __GIMPLE and __RTL. Guard
20633 property-provider override so it is only done for gimple passes.
20635 (skip_pass): New function.
20636 (execute_one_pass): Call skip_pass when skipping passes.
20637 * read-md.c (md_reader::read_char): Support filtering
20638 the input to a subset of line numbers.
20639 (md_reader::md_reader): Initialize fields
20640 m_first_line and m_last_line.
20641 (md_reader::read_file_fragment): New function.
20642 * read-md.h (md_reader::read_file_fragment): New decl.
20643 (md_reader::m_first_line): New field.
20644 (md_reader::m_last_line): New field.
20645 * read-rtl-function.c (function_reader::create_function): Only
20646 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
20647 curr_properties. Set DECL_INITIAL to a dummy block.
20648 (read_rtl_function_body_from_file_range): New function.
20649 * read-rtl-function.h (read_rtl_function_body_from_file_range):
20651 * run-rtl-passes.c: New file.
20652 * run-rtl-passes.h: New file.
20654 2017-01-24 Jeff Law <law@redhat.com>
20656 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
20659 2017-01-24 Bin Cheng <bin.cheng@arm.com>
20661 PR tree-optimization/79159
20662 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
20663 (record_nonwrapping_iv): Improve boundary using above function if no
20664 value range information.
20666 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
20667 Martin Jambor <mjambor@suse.cz>
20669 * brig-builtins.def: New file.
20670 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
20671 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
20672 (DEF_HSAIL_SAT_BUILTIN): Likewise.
20673 (DEF_HSAIL_INTR_BUILTIN): Likewise.
20674 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
20675 * builtin-types.def (BT_INT8): New.
20676 (BT_INT16): Likewise.
20677 (BT_UINT8): Likewise.
20678 (BT_UINT16): Likewise.
20679 (BT_FN_ULONG): Likewise.
20680 (BT_FN_UINT_INT): Likewise.
20681 (BT_FN_UINT_ULONG): Likewise.
20682 (BT_FN_UINT_LONG): Likewise.
20683 (BT_FN_UINT_PTR): Likewise.
20684 (BT_FN_ULONG_PTR): Likewise.
20685 (BT_FN_INT8_FLOAT): Likewise.
20686 (BT_FN_INT16_FLOAT): Likewise.
20687 (BT_FN_UINT32_FLOAT): Likewise.
20688 (BT_FN_UINT16_FLOAT): Likewise.
20689 (BT_FN_UINT8_FLOAT): Likewise.
20690 (BT_FN_UINT64_FLOAT): Likewise.
20691 (BT_FN_UINT16_UINT32): Likewise.
20692 (BT_FN_UINT32_UINT16): Likewise.
20693 (BT_FN_UINT16_UINT16_UINT16): Likewise.
20694 (BT_FN_INT_PTR_INT): Likewise.
20695 (BT_FN_UINT_PTR_UINT): Likewise.
20696 (BT_FN_LONG_PTR_LONG): Likewise.
20697 (BT_FN_ULONG_PTR_ULONG): Likewise.
20698 (BT_FN_VOID_UINT64_UINT64): Likewise.
20699 (BT_FN_UINT8_UINT8_UINT8): Likewise.
20700 (BT_FN_INT8_INT8_INT8): Likewise.
20701 (BT_FN_INT16_INT16_INT16): Likewise.
20702 (BT_FN_INT_INT_INT): Likewise.
20703 (BT_FN_UINT_FLOAT_UINT): Likewise.
20704 (BT_FN_FLOAT_UINT_UINT): Likewise.
20705 (BT_FN_ULONG_UINT_UINT): Likewise.
20706 (BT_FN_ULONG_UINT_PTR): Likewise.
20707 (BT_FN_ULONG_ULONG_ULONG): Likewise.
20708 (BT_FN_UINT_UINT_UINT): Likewise.
20709 (BT_FN_VOID_UINT_PTR): Likewise.
20710 (BT_FN_UINT_UINT_PTR: Likewise.
20711 (BT_FN_UINT32_UINT64_PTR): Likewise.
20712 (BT_FN_INT_INT_UINT_UINT): Likewise.
20713 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
20714 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
20715 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
20716 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
20717 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
20718 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
20719 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
20720 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
20721 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
20722 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
20723 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
20724 * doc/frontends.texi: List BRIG FE.
20725 * doc/install.texi (Testing): Add BRIG tesring requirements.
20726 * doc/invoke.texi (Overall Options): Mention BRIG.
20727 * doc/standards.texi (Standards): Doucment BRIG HSA version.
20729 2017-01-24 Richard Biener <rguenther@suse.de>
20731 PR translation/79208
20732 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
20734 2017-01-24 Martin Jambor <mjambor@suse.cz>
20737 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
20738 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
20739 and known_contexts.
20741 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
20743 PR middle-end/79123
20744 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
20745 casts from signed to unsigned really don't have a range.
20747 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
20749 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
20750 GMP_RNDx for compatiblity.
20752 2017-01-24 Martin Liska <mliska@suse.cz>
20755 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
20756 that would prevent us to call alloca with -1 as argument.
20758 2017-01-24 Jakub Jelinek <jakub@redhat.com>
20760 * dwarf2out.c (output_compilation_unit_header, output_file_names):
20761 Avoid -Wformat-security warning.
20763 2017-01-23 Andrew Pinski <apinski@cavium.com>
20765 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
20768 2017-01-23 Martin Sebor <msebor@redhat.com>
20770 PR middle-end/78703
20771 * gimple-ssa-sprintf.c (warn_level): New global.
20772 (format_integer): Use it here and throughout the rest of the file.
20773 Use the same switch to compute sign as base.
20774 (maybe_warn): New function.
20775 (format_directive): Factor out warnings into maybe_warn.
20776 Add debugging output. Use warn_level.
20777 (add_bytes): Use warn_level.
20778 (pass_sprintf_length::compute_format_length): Add debugging output.
20779 (try_substitute_return_value): Same.
20780 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
20782 PR middle-end/78703
20783 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
20784 (struct fmtresult, format_integer, format_floating): Adjust.
20785 (fmtresult::fmtresult): Set max correctly in two argument ctor.
20786 (get_string_length, format_string,format_directive): Same.
20787 (pass_sprintf_length::compute_format_length): Same.
20788 (try_substitute_return_value): Simplify slightly.
20790 PR middle-end/78703
20791 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
20792 (fmtresult::operator+=): Outlined.
20793 (struct fmtresult): Add ctors.
20794 (struct conversion_spec): Rename...
20795 (struct directive): ...to this. Add and remove data members.
20796 (directive::set_width, directive::set_precision): New functions.
20797 (format_percent): Use fmtresult ctor.
20798 (get_width_and_precision): Remove.
20799 (format_integer): Make naming changes. Avoid computing width and
20801 (format_floating): Same. Adjust indentation.
20802 (format_character, format_none): New functions.
20803 (format_string): Moved character handling to format_character.
20804 (format_directive): Remove arguments, change return type.
20805 (parse_directive): New function.
20806 (pass_sprintf_length::compute_format_length): Move directive
20807 parsing to parse_directive.
20809 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20811 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
20812 (assign_assembler_name_if_needed): ... this.
20813 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
20814 (assign_assembler_name_if_needed): ... this.
20815 (free_lang_data_in_cgraph): Adjust callers.
20816 * cgraphunit.c (cgraph_node::analyze): Likewise.
20817 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
20820 2017-01-23 Richard Biener <rguenther@suse.de>
20822 PR tree-optimization/79088
20823 PR tree-optimization/79188
20824 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
20825 resetting loop bounds after last path deletion. Reset loop
20826 bounds of the target loop, make code match the comments.
20827 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
20828 Make sure loops need no fixups.
20830 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20832 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
20833 exponent support with double type for first argument.
20834 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
20835 type returned by __builtin_vec_extract_sig,
20836 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
20837 functions from "vector int" to "vector unsigned int" or from
20838 "vector long long int" to "vector unsigned long long int".
20839 Changed type returned by __builtin_vec_extract_exp,
20840 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
20841 functions from "vector int" to "vector unsigned int" or from
20842 "vector long long int" to "vector unsigned long long int".
20843 Changed return type of __builtin_vec_test_data_class,
20844 __builtin_vec_test_data_class_sp, and
20845 __builtin_vec_test_data_class_dp from "vector int" to
20846 "vector bool int" or from "vector long long int" to "vector bool
20847 long long int" and changed second argument type from "unsigned
20848 int" to "int". Added new overloaded function forms "vector float
20849 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
20850 "vector float __builtin_vec_insert_exp_sp (vector float, vector
20851 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
20852 double, vector unsigned long long int)" and "vector double
20853 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
20854 long int)". Changed return type of
20855 __builtin_scalar_test_data_class and
20856 __builtin_scalar_test_data_class_sp and
20857 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
20858 int" and changed second argument from "unsigned int" to "int".
20859 Changed type returned by __builtin_scalar_test_neg,
20860 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
20861 from "int" to "bool int". Added new overloaded function form
20862 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
20863 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
20864 exponent double-precision with floating point first argument.
20865 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
20866 documentation of scalar_test_data_class, scalar_test_neg,
20867 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
20868 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
20869 vec_test_data_class built-in functions to reflect refinements in
20870 their type signatures.
20872 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
20874 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
20876 (aarch64_elf_asm_destructor): Likewise.
20878 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
20880 PR rtl-optimization/78634
20881 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
20882 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
20883 * ifcvt.c (noce_try_cmove): Add missing cost check.
20885 PR rtl-optimization/71724
20886 * combine.c (if_then_else_cond): Look for situations where it is
20887 beneficial to undo the work of one of the recursive calls.
20889 2017-01-23 Bin Cheng <bin.cheng@arm.com>
20891 PR tree-optimization/70754
20892 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
20893 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
20894 combined stmt before it if not NULL.
20895 (combine_chains): Process refs reversely and compute dominance point
20898 2017-01-23 Martin Liska <mliska@suse.cz>
20900 PR tree-optimization/79196
20901 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
20902 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
20904 (strlen_optimize_stmt): Call the renamed function.
20906 2017-01-23 Michael Matz <matz@suse.de>
20908 PR tree-optimization/78384
20909 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
20911 2017-01-23 Richard Biener <rguenther@suse.de>
20913 PR tree-optimization/79186
20914 * tree-vrp.c (register_new_assert_for): Make sure we've seen
20915 both incoming edges before moving an assert.
20917 2017-01-23 Martin Jambor <mjambor@suse.cz>
20919 * ipa-prop.c (load_from_param_1): Removed.
20920 (load_from_unmodified_param): Bits from load_from_param_1 put back
20922 (load_from_param): Removed.
20923 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
20924 with stmt. Reverted back to use of load_from_unmodified_param.
20926 2017-01-23 Martin Jambor <mjambor@suse.cz>
20929 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
20930 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
20931 field a pointer to garbage collected vector, mark lattices and
20932 ipcp_orig_node with GTY((skip)).
20933 (ipa_get_param_count): Adjust to descriptors being a pointer.
20934 (ipa_get_param): Likewise.
20935 (ipa_get_type): Likewise.
20936 (ipa_get_param_move_cost): Likewise.
20937 (ipa_set_param_used): Likewise.
20938 (ipa_get_controlled_uses): Likewise.
20939 (ipa_set_controlled_uses): Likewise.
20940 (ipa_is_param_used): Likewise.
20941 (ipa_node_params_t): Move into garbage collector. New methods insert
20943 (ipa_node_params_sum): Annotate wth GTY(()).
20944 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
20946 (ipa_load_from_parm_agg): Adjust declaration.
20947 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
20948 * ipa-profile.c (ipa_profile): Likewise.
20949 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
20950 (ipa_populate_param_decls): Make descriptors parameter garbage
20952 (ipa_dump_param): Adjust to descriptors being a pointer.
20953 (ipa_alloc_node_params): Likewise.
20954 (ipa_initialize_node_params): Likewise.
20955 (load_from_param_1): Make descriptors parameter garbage collected.
20956 (load_from_unmodified_param): Likewise.
20957 (load_from_param): Likewise.
20958 (ipa_load_from_parm_agg): Likewise.
20959 (ipa_node_params::~ipa_node_params): Removed.
20960 (ipa_free_all_node_params): Remove call to delete operator.
20961 (ipa_node_params_t::insert): New.
20962 (ipa_node_params_t::remove): Likewise.
20963 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
20964 copy known_csts and known_contexts vectors.
20965 (ipa_read_node_info): Adjust to descriptors being a pointer.
20966 (ipcp_modif_dom_walker): Make m_descriptors field garbage
20968 (ipcp_transform_function): Make descriptors variable garbage
20971 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
20973 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
20974 * config/i386/avx512dqintrin.h: Ditto.
20975 * config/i386/avx512fintrin.h: Ditto.
20976 * config/i386/i386.c: Handle new builtins.
20977 * config/i386/i386-builtin.def: Add new builtins.
20978 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
20979 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
20981 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20982 Martin Liska <mliska@suse.cz>
20984 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
20985 * asan.c (asan_expand_poison_ifn): Support stores and use
20986 appropriate ASAN report function.
20987 * internal-fn.c (expand_ASAN_POISON_USE): New function.
20988 * internal-fn.def (ASAN_POISON_USE): Declare.
20989 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
20990 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
20991 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
20992 ASAN_POISON calls w/o LHS.
20993 * tree-ssa.c (execute_update_addresses_taken): Create clobber
20994 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
20995 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
20996 * gimplify.c (asan_poison_variables): Add attribute
20997 use_after_scope_memory to variables that really needs to live
20999 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21000 having the attribute.
21002 2017-01-23 Martin Liska <mliska@suse.cz>
21004 * asan.c (create_asan_shadow_var): New function.
21005 (asan_expand_poison_ifn): Likewise.
21006 * asan.h (asan_expand_poison_ifn): New declaration.
21007 * internal-fn.c (expand_ASAN_POISON): Likewise.
21008 * internal-fn.def (ASAN_POISON): New builtin.
21009 * sanopt.c (pass_sanopt::execute): Expand
21010 asan_expand_poison_ifn.
21011 * tree-inline.c (copy_decl_for_dup_finish): Make function
21013 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21014 * tree-ssa.c (is_asan_mark_p): New function.
21015 (execute_update_addresses_taken): Rewrite local variables
21016 (identified just by use-after-scope as addressable) into SSA.
21018 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21020 * doc/install.texi (Specific): opensource.apple.com uses https
21021 now. Remove trailing slash.
21023 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21025 * README.Portability: Remove note on an Irix compatibility issue.
21027 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21029 * gcov.c (INCLUDE_ALGORITHM): Define.
21030 (INCLUDE_VECTOR): Define.
21031 No longer include <vector> and <algorithm> directly.
21033 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21035 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21037 * doc/invoke.texi (Code Gen Options): Ditto.
21039 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21042 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21044 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21046 rtl-optimization/79125
21047 * cprop.c (local_cprop_pass): Handle cases where we make an
21048 unconditional trap.
21050 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21054 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21055 read from, for big endian.
21057 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21059 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21060 register pauth builtins for LP64 only.
21062 2017-01-20 Marek Polacek <polacek@redhat.com>
21065 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21068 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21070 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21072 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21073 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21074 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21076 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21079 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21080 in big-endian mode when they are not a single duplicated value.
21082 2017-01-20 Richard Biener <rguenther@suse.de>
21084 * BASE-VER: Bump to 7.0.1.
21086 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21088 * omp-low.c (omplow_simd_context): New struct. Use it...
21089 (lower_rec_simd_input_clauses): ...here and...
21090 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21091 references to idx, lane, max_vf, is_simt.
21093 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21095 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21098 2017-01-20 Martin Jambor <mjambor@suse.cz>
21100 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21101 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21103 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21104 (GTFILES): Rename hsa.c to hsa-common.c.
21105 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21106 * hsa-dump.c: Likewise.
21107 * hsa-gen.c: Likewise.
21108 * hsa-regalloc.c: Likewise.
21109 * ipa-hsa.c: Likewise.
21110 * omp-expand.c: Likewise.
21111 * omp-low.c: Likewise.
21112 * toplev.c: Likewise.
21114 2017-01-20 Marek Polacek <polacek@redhat.com>
21117 * doc/invoke.texi: Document -Wduplicated-branches.
21118 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21119 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21120 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21121 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21122 return 0 only when not OEP_LEXICOGRAPHIC.
21123 (fold_build_cleanup_point_expr): Use the expression
21124 location when building CLEANUP_POINT_EXPR.
21125 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21126 * tree.c (add_expr): Handle error_mark_node.
21128 2017-01-20 Martin Liska <mliska@suse.cz>
21131 * tree-profile.c (init_ic_make_global_vars): Do not call
21133 (gimple_init_gcov_profiler): Likewise.
21135 2017-01-20 Martin Liska <mliska@suse.cz>
21138 * cgraph.h (maybe_create_reference): Remove argument and
21140 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21142 * ipa-cp.c (create_specialized_node): Likewise.
21143 * symtab.c (symtab_node::maybe_create_reference): Handle
21144 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21146 2017-01-20 Martin Liska <mliska@suse.cz>
21148 * read-rtl-function.c (function_reader::create_function): Use
21149 build_decl instread of build_decl_stat.
21151 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21153 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21154 * config/i386/avx512dqintrin.h: Ditto.
21155 * config/i386/avx512fintrin.h: Ditto.
21156 * config/i386/i386-builtin-types.def: Add new types.
21157 * config/i386/i386.c: Handle new types.
21158 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21159 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21160 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21161 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21162 (__builtin_ia32_kshiftridi): New.
21163 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21165 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21169 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21170 define to rs6000_init_stack_protect_guard.
21171 (rs6000_init_stack_protect_guard): New function.
21173 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21174 Yunqiang Su <yunqiang.su@imgtec.com>
21176 * config.gcc (supported_defaults): Add madd4.
21177 (with_madd4): Add validation.
21178 (all_defaults): Add madd4.
21179 * config/mips/mips.opt (mmadd4): New option.
21180 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21182 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21184 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21185 (ISA_HAS_FUSED_MADD4): Likewise.
21186 * doc/invoke.texi (-mmadd4): Document the new option.
21187 * doc/install.texi (--with-madd4): Document the new option.
21189 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21191 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21192 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21193 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21194 (aarch64_init_pauth_hint_builtins): New.
21195 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21196 (aarch64_expand_builtin): Expand new builtins.
21198 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21200 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21201 * combine-stack-adj.c (no_unhandled_cfa): Handle
21202 REG_CFA_TOGGLE_RA_MANGLE.
21203 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21204 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21205 info for return address signing.
21206 (aarch64_expand_epilogue): Likewise.
21208 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21210 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21211 * config/aarch64/aarch64-protos.h
21212 (aarch64_return_address_signing_enabled): New declaration.
21213 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21215 (aarch64_expand_prologue): Sign return address before it's pushed onto
21217 (aarch64_expand_epilogue): Authenticate return address fetched from
21219 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21220 (aarch64_attributes): New function attributes for "sign-return-address".
21221 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21222 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21223 ("*do_return"): Generate combined instructions according to key index.
21224 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21225 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21227 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21228 * config/aarch64/aarch64.opt (msign-return-address=): New.
21229 * doc/extend.texi (AArch64 Function Attributes): Documents
21230 "sign-return-address=".
21231 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21233 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21235 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21236 overall option summary.
21238 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21240 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21241 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21242 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21243 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21245 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21247 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21248 -mpower9-minmax by default for -mcpu=power9.
21249 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21250 128-bit floating point.
21252 2017-01-20 Alan Modra <amodra@gmail.com>
21254 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21255 optimizing for size.
21257 2017-01-20 Alan Modra <amodra@gmail.com>
21260 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21261 for strcmp and strncmp from corresponding builtin decl.
21263 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21265 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21266 instead of i386/rtems-64.h.
21267 * config/i386/rtems-64.h: Remove.
21269 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21273 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21275 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21277 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21279 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21280 Change int to HOST_WIDE_INT.
21281 * config/aarch64/aarch64-protos.h
21282 (aarch64_simd_gen_const_vector_dup): Likewise.
21283 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21285 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21287 * langhooks-def.h (lhd_type_for_size): New decl.
21288 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21289 * langhooks.c (lhd_type_for_size): New function, taken from
21292 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21294 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21295 define_bypass for CR latency.
21296 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21297 (power9-alu2): Add define_bypass for CR latency.
21299 (power9-mul): Update insn latency.
21300 (power9-mul-compare): Update insn latency, bypass latency and remove
21303 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21305 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21307 * config/aarch64/aarch64.md
21308 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21309 aarch64_nopcrelative_literal_loads.
21310 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21312 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21314 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21315 TARGET_LOONGSON_3A.
21316 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21318 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21321 * config.gcc (supported_defaults): Add lxc1-sxc1.
21322 (with_lxc1_sxc1): Add validation.
21323 (all_defaults): Add lxc1-sxc1.
21324 * config/mips/mips.opt (mlxc1-sxc1): New option.
21325 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21327 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21328 __mips_no_lxc1_sxc1.
21329 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21330 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21331 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21333 2017-01-19 Richard Biener <rguenther@suse.de>
21335 PR tree-optimization/72488
21336 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21337 sure to restore SSA info.
21338 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21340 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21342 PR rtl-optimization/79121
21343 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21344 of the inner type when shifting an extended value.
21346 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21349 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21350 interposable aliases.
21352 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21355 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21356 Use the evmergelohi instruction.
21357 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21358 (mov_sitf_e500_subreg8_2_be): Likewise.
21359 (mov_sitf_e500_subreg12_2_le): Likewise.
21360 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21361 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21362 (mov_sitf_e500_subreg8_2_le): Likewise.
21363 (mov_sitf_e500_subreg12_2_be): Likewise.
21365 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21367 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21368 attribute from vecsimple to vecperm.
21369 (altivec_vbpermq2): Likewise.
21371 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21374 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21376 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21377 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21378 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21379 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21380 case where N arg is SIZE_MAX.
21381 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21382 (cmpstrsi): Add pattern.
21384 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21386 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21387 __builtin_vec_revb builtins.
21388 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21389 built-in functions to support generation of the ISA 3.0 XXBR<x>
21390 vector byte reverse instructions.
21391 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21392 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21393 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21394 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21395 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21396 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21397 (P9V_BUILTIN_VEC_REVB): Likewise.
21398 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21399 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21400 (p9_xxbrq_v16qi): Likewise.
21401 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21402 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21403 (p9_xxbrh_v8hi): Likewise.
21404 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21405 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21406 vec_revb built-in functions.
21408 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21410 PR rtl-optimization/78952
21411 * config/i386/i386.md (any_extract): New code iterator.
21412 (*insvqi_2): Use any_extract for source operand.
21413 (*insvqi_3): Use any_shiftrt for source operand.
21415 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21417 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21419 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21421 2017-01-18 Matthias Klose <doko@ubuntu.com>
21423 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21425 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21427 * config/rs6000/altivec.h (vec_bperm): Change #define.
21428 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21429 (altivec_vbpermq2): New define_insn.
21430 (altivec_vbpermd): Likewise.
21431 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21432 function interface.
21433 (VBPERMD): Likewise.
21434 (VBPERM): New polymorphic function interface.
21435 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21436 Add entries for P9V_BUILTIN_VEC_VBPERM.
21437 * doc/extend.texi: Add interfaces for vec_bperm.
21439 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21441 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21442 first letter of error messages.
21443 (s390_resolve_overloaded_builtin): Likewise.
21444 * config/s390/s390.c (s390_expand_builtin): Likewise.
21445 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21446 (s390_valid_target_attribute_inner_p): Likewise.
21447 * config/s390/s390.md ("tabort"): Likewise.
21449 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21451 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21452 (ISA_AVOID_DIV_HILO): New macro.
21453 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21454 (ISA_HAS_DDIV): Likewise.
21456 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21458 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21460 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21462 * doc/invoke.texi (fabi-version): Spelling fix.
21464 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21467 * doc/invoke.texi (fabi-version): Mention mangling fix for
21470 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21473 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21475 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21478 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21481 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21483 (stack_protect_test): Ditto.
21484 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21485 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21487 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21488 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21489 -mstack-protector-guard-offset=.
21490 (RS/6000 and PowerPC Options): Ditto.
21492 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21494 * config/i386/i386.h (MASK_CLASS_P): New define.
21495 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21496 there are no registers from different register sets also when
21497 mask registers are used. Update function comment.
21498 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21499 to (*k/*r) and (*k/*km) alternatives.
21501 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21503 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21504 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21505 (EH_RETURN_HANDLER_RTX): New define.
21506 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21507 Force frame pointer in EH return functions.
21508 (aarch64_expand_epilogue): Add barrier for eh_return.
21509 (aarch64_final_eh_return_addr): Remove.
21510 (aarch64_eh_return_handler_rtx): New function.
21511 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21513 (aarch64_eh_return_handler_rtx): New prototype.
21515 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21517 * config/rs6000/altivec.h (vec_rlmi): New #define.
21518 (vec_vrlnm): Likewise.
21519 (vec_rlnm): Likewise.
21520 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21521 (UNSPEC_VRLNM): Likewise.
21522 (VIlong): New mode iterator.
21523 (altivec_vrl<VI_char>mi): New define_insn.
21524 (altivec_vrl<VI_char>nm): Likewise.
21525 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21527 (VRLDNM): Likewise.
21528 (RLNM): New polymorphic function entry.
21529 (VRLWMI): New monomorphic function entry.
21530 (VRLDMI): Likewise.
21531 (RLMI): New polymorphic function entry.
21532 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21533 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21534 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21537 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21540 * dwarf2out.c (field_byte_offset): Restore the
21541 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21542 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21543 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21546 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21549 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21551 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21554 * dwarf2out.c (add_data_member_location_attribute): For constant
21555 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21556 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21557 DW_AT_byte_size attributes.
21559 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21561 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21562 after forcing to constant memory when the code model is medium.
21564 2017-01-17 Julia Koval <julia.koval@intel.com>
21567 * config/i386/avx512fintrin.h
21568 (_mm512_i32gather_ps): Change __addr type to void const*.
21569 (_mm512_mask_i32gather_ps): Ditto.
21570 (_mm512_i32gather_pd): Ditto.
21571 (_mm512_mask_i32gather_pd): Ditto.
21572 (_mm512_i64gather_ps): Ditto.
21573 (_mm512_mask_i64gather_ps): Ditto.
21574 (_mm512_i64gather_pd): Ditto.
21575 (_mm512_mask_i64gather_pd): Ditto.
21576 (_mm512_i32gather_epi32): Ditto.
21577 (_mm512_mask_i32gather_epi32): Ditto.
21578 (_mm512_i32gather_epi64): Ditto.
21579 (_mm512_mask_i32gather_epi64): Ditto.
21580 (_mm512_i64gather_epi32): Ditto.
21581 (_mm512_mask_i64gather_epi32): Ditto.
21582 (_mm512_i64gather_epi64): Ditto.
21583 (_mm512_mask_i64gather_epi64): Ditto.
21584 (_mm512_i32scatter_ps): Change __addr type to void*.
21585 (_mm512_mask_i32scatter_ps): Ditto.
21586 (_mm512_i32scatter_pd): Ditto.
21587 (_mm512_mask_i32scatter_pd): Ditto.
21588 (_mm512_i64scatter_ps): Ditto.
21589 (_mm512_mask_i64scatter_ps): Ditto.
21590 (_mm512_i64scatter_pd): Ditto.
21591 (_mm512_mask_i64scatter_pd): Ditto.
21592 (_mm512_i32scatter_epi32): Ditto.
21593 (_mm512_mask_i32scatter_epi32): Ditto.
21594 (_mm512_i32scatter_epi64): Ditto.
21595 (_mm512_mask_i32scatter_epi64): Ditto.
21596 (_mm512_i64scatter_epi32): Ditto.
21597 (_mm512_mask_i64scatter_epi32): Ditto.
21598 (_mm512_i64scatter_epi64): Ditto.
21599 (_mm512_mask_i64scatter_epi64): Ditto.
21600 * config/i386/avx512pfintrin.h
21601 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21602 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21603 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21604 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21605 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21606 (_mm512_prefetch_i32scatter_ps): Ditto.
21607 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
21608 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
21609 (_mm512_prefetch_i64scatter_pd): Ditto.
21610 (_mm512_prefetch_i64scatter_ps): Ditto.
21611 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
21612 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
21613 * config/i386/avx512vlintrin.h
21614 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
21615 (_mm_mmask_i32gather_ps): Ditto.
21616 (_mm256_mmask_i32gather_pd): Ditto.
21617 (_mm_mmask_i32gather_pd): Ditto.
21618 (_mm256_mmask_i64gather_ps): Ditto.
21619 (_mm_mmask_i64gather_ps): Ditto.
21620 (_mm256_mmask_i64gather_pd): Ditto.
21621 (_mm_mmask_i64gather_pd): Ditto.
21622 (_mm256_mmask_i32gather_epi32): Ditto.
21623 (_mm_mmask_i32gather_epi32): Ditto.
21624 (_mm256_mmask_i32gather_epi64): Ditto.
21625 (_mm_mmask_i32gather_epi64): Ditto.
21626 (_mm256_mmask_i64gather_epi32): Ditto.
21627 (_mm_mmask_i64gather_epi32): Ditto.
21628 (_mm256_mmask_i64gather_epi64): Ditto.
21629 (_mm_mmask_i64gather_epi64): Ditto.
21630 (_mm256_i32scatter_ps): Change __addr type to void*.
21631 (_mm256_mask_i32scatter_ps): Ditto.
21632 (_mm_i32scatter_ps): Ditto.
21633 (_mm_mask_i32scatter_ps): Ditto.
21634 (_mm256_i32scatter_pd): Ditto.
21635 (_mm256_mask_i32scatter_pd): Ditto.
21636 (_mm_i32scatter_pd): Ditto.
21637 (_mm_mask_i32scatter_pd): Ditto.
21638 (_mm256_i64scatter_ps): Ditto.
21639 (_mm256_mask_i64scatter_ps): Ditto.
21640 (_mm_i64scatter_ps): Ditto.
21641 (_mm_mask_i64scatter_ps): Ditto.
21642 (_mm256_i64scatter_pd): Ditto.
21643 (_mm256_mask_i64scatter_pd): Ditto.
21644 (_mm_i64scatter_pd): Ditto.
21645 (_mm_mask_i64scatter_pd): Ditto.
21646 (_mm256_i32scatter_epi32): Ditto.
21647 (_mm256_mask_i32scatter_epi32): Ditto.
21648 (_mm_i32scatter_epi32): Ditto.
21649 (_mm_mask_i32scatter_epi32): Ditto.
21650 (_mm256_i32scatter_epi64): Ditto.
21651 (_mm256_mask_i32scatter_epi64): Ditto.
21652 (_mm_i32scatter_epi64): Ditto.
21653 (_mm_mask_i32scatter_epi64): Ditto.
21654 (_mm256_i64scatter_epi32): Ditto.
21655 (_mm256_mask_i64scatter_epi32): Ditto.
21656 (_mm_i64scatter_epi32): Ditto.
21657 (_mm_mask_i64scatter_epi32): Ditto.
21658 (_mm256_i64scatter_epi64): Ditto.
21659 (_mm256_mask_i64scatter_epi64): Ditto.
21660 (_mm_i64scatter_epi64): Ditto.
21661 (_mm_mask_i64scatter_epi64): Ditto.
21662 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
21663 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
21664 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
21665 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
21666 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
21667 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
21668 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
21669 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
21670 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
21671 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
21672 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
21673 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
21674 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
21675 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
21676 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
21677 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
21678 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
21679 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
21680 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
21681 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
21682 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
21683 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
21684 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
21685 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
21686 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
21687 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
21688 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
21689 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
21690 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
21691 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
21692 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
21693 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
21694 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
21695 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
21696 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
21697 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
21698 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
21699 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
21700 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
21701 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
21702 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
21703 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
21704 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
21705 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
21706 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
21707 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
21708 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
21709 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
21710 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
21711 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
21712 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
21713 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
21714 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
21715 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
21716 definitions accordingly.
21718 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
21719 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21722 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
21725 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
21728 * ira-conflicts.c (ira_build_conflicts): Update total conflict
21729 hard regs for inner regno.
21731 2017-01-17 Martin Liska <mliska@suse.cz>
21734 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
21735 assumption and add comment.
21737 2017-01-17 Nathan Sidwell <nathan@acm.org>
21739 * ipa-visibility.c (localize_node): New function, broken out of ...
21740 (function_and_variable_visibility): ... here. Call it.
21742 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21744 PR middle-end/77445
21745 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
21746 correctly set frequency of oudgoing edge.
21747 (duplicate_thread_path): Fix profile updating.
21749 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21752 * configure.ac: Add GCC_BASE_VER.
21753 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
21754 version from BASE-VER file.
21755 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
21756 (gcc.o): Depend on $(BASEVER).
21757 * common.opt (dumpfullversion): New option.
21758 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
21759 * doc/invoke.texi: Document -dumpfullversion.
21760 * doc/install.texi: Document --with-gcc-major-version-only.
21761 * configure: Regenerated.
21763 2017-01-17 Richard Biener <rguenther@suse.de>
21765 PR tree-optimization/71433
21766 * tree-vrp.c (register_new_assert_for): Merge same asserts
21767 on all incoming edges.
21768 (process_assert_insertions_for): Handle insertions at the
21771 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
21773 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
21774 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
21776 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
21779 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
21782 2017-01-17 Alan Modra <amodra@gmail.com>
21785 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
21786 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
21787 symbolic stack limit when pic.
21789 2017-01-16 Martin Sebor <msebor@redhat.com>
21791 PR tree-optimization/78608
21792 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
21794 2017-01-16 Jeff Law <law@redhat.com>
21797 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
21798 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
21799 for several include directories that may be relative to sysroot.
21800 * config/i386/x-mingw32 (gplus_includedir): Define.
21801 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
21802 (native_system_includedir): Likewise.
21803 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
21804 override if TARGET_SYSTEM_ROOT is defined.
21805 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
21807 PR tree-optimization/79090
21808 PR tree-optimization/33562
21809 PR tree-optimization/61912
21810 PR tree-optimization/77485
21811 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
21812 and computed trims into the dump file.
21814 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21816 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
21818 2017-01-16 Jakub Jelinek <jakub@redhat.com>
21821 * gimplify.c (gimplify_init_constructor): If want_value and
21822 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
21826 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
21827 sequence. Formatting fixes.
21828 (doloop_optimize): Formatting fixes.
21831 * gcc.c (debug_level_greater_than_spec_func): New function.
21832 (static_spec_functions): Add debug-level-gt spec function.
21833 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
21835 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21836 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21837 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
21838 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
21839 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
21840 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
21842 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
21844 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
21845 QImode fixups to general and mask registers only.
21847 2017-01-16 Carl Love <cel@us.ibm.com>
21849 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
21850 for built-in functions
21851 vector signed char vec_nabs (vector signed char)
21852 vector signed short vec_nabs (vector signed short)
21853 vector signed int vec_nabs (vector signed int)
21854 vector signed long long vec_nabs (vector signed long long)
21855 vector float vec_nabs (vector float)
21856 vector double vec_nabs (vector double)
21857 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
21859 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
21860 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
21861 * doc/extend.texi: Update the documentation file for the new built-in
21864 2017-01-16 Martin Sebor <msebor@redhat.com>
21866 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
21869 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21871 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
21872 UNSPEC_VSX__XXSPLTD to require special splat handling.
21874 2017-01-16 David Malcolm <dmalcolm@redhat.com>
21877 * system.h: Poison strndup.
21879 2017-01-16 Alan Modra <amodra@gmail.com>
21882 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
21885 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
21887 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
21889 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
21891 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
21892 call recog here. Assert that INSN_CODE (insn) is non-negative.
21894 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21897 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
21899 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
21900 in the currently scheduled RTL fragment.
21902 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21904 PR rtl-optimization/78751
21905 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
21908 2017-01-14 Jeff Law <law@redhat.com>
21910 PR tree-optimization/79090
21911 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
21912 variable length stores.
21913 (compute_trims): Delete dead assignment to *trim_tail.
21914 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
21917 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
21919 PR rtl-optimization/78626
21920 PR rtl-optimization/78727
21921 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
21922 of a block, and split such blocks after everything else is finished.
21924 2017-01-14 Alan Modra <amodra@gmail.com>
21927 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
21928 target legitimate_combined_insn.
21929 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
21930 (rs6000_legitimate_combined_insn): New function.
21931 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
21933 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
21934 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
21935 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
21937 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
21939 * doc/frontends.texi (G++ and GCC): Remove references to Java.
21941 2017-01-13 Jeff Law <law@redhat.com>
21943 PR tree-optimization/33562
21944 PR tree-optimization/61912
21945 PR tree-optimization/77485
21946 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
21948 (delete_dead_assignment): Likewise.
21949 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
21950 statement to delete_dead_call and delete_dead_assignment.
21952 2017-01-13 David Malcolm <dmalcolm@redhat.com>
21955 * substring-locations.c (format_warning_va): Strengthen case 1 so
21956 that both endpoints of the substring must be within the format
21957 range for just the substring to be printed.
21959 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
21961 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
21962 * config/i386/i386.c (ix86_target_string): Add missing options
21963 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
21964 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
21965 flags_other and ix86_target_other to flags2_other. Display unknown
21967 (ix86_valid_target_attribute_inner_p): Add missing options and
21968 reorder options by implied ISAs, as in ix86_target_string.
21970 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
21972 * hash-table.h (hash_table::too_empty_p): New function.
21973 (hash_table::expand): Use it.
21974 (hash_table::traverse): Likewise.
21975 (hash_table::empty_slot): Use sizeof (value_type) instead of
21976 sizeof (PTR) to convert bytes to elements. Shrink the table
21977 if the current size is excessive for the current number of
21980 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
21982 * ira-costs.c (record_reg_classes): Break from the inner loop
21983 early once alt_fail is known to be true. Update outer loop
21984 handling accordingly.
21986 2017-01-13 Jeff Law <law@redhat.com>
21988 * tree-ssa-dse.c (decrement_count): New function.
21989 (increment_start_addr, maybe_trim_memstar_call): Likewise.
21990 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
21991 when we know the partially dead statement is a mem* function.
21993 PR tree-optimization/61912
21994 PR tree-optimization/77485
21995 * tree-ssa-dse.c: Include expr.h.
21996 (maybe_trim_constructor_store): New function.
21997 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
21999 PR tree-optimization/33562
22000 PR tree-optimization/61912
22001 PR tree-optimization/77485
22002 * doc/invoke.texi: Document new dse-max-object-size param.
22003 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22004 * tree-ssa-dse.c: Include params.h.
22005 (dse_store_status): New enum.
22006 (initialize_ao_ref_for_dse): New, partially extracted from
22008 (valid_ao_ref_for_dse, normalize_ref): New.
22009 (setup_live_bytes_from_ref, compute_trims): Likewise.
22010 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22011 (maybe_trim_partially_dead_store): Likewise.
22012 (maybe_trim_complex_store): Likewise.
22013 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22014 Track what bytes live from the original store. Return tri-state
22015 for dead, partially dead or live.
22016 (dse_dom_walker): Add constructor, destructor and new private members.
22017 (delete_dead_call, delete_dead_assignment): New extracted from
22019 (dse_optimize_stmt): Make a member of dse_dom_walker.
22020 Use initialize_ao_ref_for_dse.
22022 PR tree-optimization/33562
22023 PR tree-optimization/61912
22024 PR tree-optimization/77485
22025 * sbitmap.h (bitmap_count_bits): Prototype.
22026 (bitmap_clear_range, bitmap_set_range): Likewise.
22027 * sbitmap.c (bitmap_clear_range): New function.
22028 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22030 2017-01-13 Martin Liska <mliska@suse.cz>
22033 * function.c (set_cfun): Add new argument force.
22034 * function.h (set_cfun): Likewise.
22035 * ipa-inline-transform.c (inline_call): Use the function when
22036 strict alising from is dropped for function we inline to.
22038 2017-01-13 Richard Biener <rguenther@suse.de>
22040 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22041 for dumping GIMPLE INTEGER_CSTs.
22043 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22045 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22046 to 201112L since C++17.
22048 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22051 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22052 if -fsanitize=kernel-address is present.
22054 2017-01-13 Richard Biener <rguenther@suse.de>
22056 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22057 as _Literal ( type ) number in case usual suffixes do not
22058 preserve all information.
22060 2017-01-13 Richard Biener <rguenther@suse.de>
22062 PR tree-optimization/77283
22063 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22064 and ssa-iterators.h.
22065 (is_feasible_trace): Implement a cost model based on joiner
22068 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22071 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22072 char or short to __float128/_Float128 directly.
22074 2017-01-12 Martin Sebor <msebor@redhat.com>
22076 to -Wformat-overflow.
22077 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22078 (min_bytes_remaining): Same.
22079 (get_string_length): Same.
22080 (format_string): Same.
22081 (format_directive): Same.
22083 (pass_sprintf_length::handle_gimple_call): Same.
22085 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22087 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22088 info.nowrite calls with no lhs that can't throw. Return bool
22089 whether gsi_remove has been called or not.
22090 (pass_sprintf_length::handle_gimple_call): Return bool whether
22091 try_substitute_return_value called gsi_remove. Formatting fix.
22092 (pass_sprintf_length::execute): Don't use gsi_remove if
22093 handle_gimple_call returned true.
22096 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22097 be removed due to side-effects, don't remove following barrier nor
22098 turn the successor edge into fallthru edge.
22100 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22103 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22104 element-reversing loads and stores as not swappable.
22106 2017-01-12 Nathan Sidwell <nathan@acm.org>
22107 Nicolai Stange <nicstange@gmail.com>
22109 * combine.c (try_combine): Don't ignore result of overlap checking
22110 loop. Combine overlap & asm check into single loop.
22112 2017-01-12 Richard Biener <rguenther@suse.de>
22114 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22115 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22117 2017-01-12 Richard Biener <rguenther@suse.de>
22119 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22120 and TS_TARGET_OPTION directly derive from TS_BASE.
22121 * tree-core.h (tree_optimization_option): Derive from tree_base.
22122 (tree_target_option): Likewise.
22124 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22126 * config/i386/i386.c (memory_address_length): Increase len
22127 only when rip_relative_addr_p returns false.
22129 2017-01-11 Julia Koval <julia.koval@intel.com>
22131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22132 (OPTION_MASK_ISA_SGX_SET): New.
22133 (ix86_handle_option): Handle OPT_msgx.
22134 * config.gcc: Added sgxintrin.h.
22135 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22136 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22137 * config/i386/i386.c (ix86_target_string): Add -msgx.
22139 (ix86_option_override_internal): Handle new options.
22140 (ix86_valid_target_attribute_inner_p): Add sgx.
22141 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22142 * config/i386/i386.opt: Add msgx.
22143 * config/i386/sgxintrin.h: New file.
22144 * config/i386/x86intrin.h: Add sgxintrin.h.
22146 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22149 * fold-const.c (maybe_nonzero_address): Return 1 for function
22151 (tree_single_nonzero_warnv_p): Don't handle function local objects
22155 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22158 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22161 * opts.c: Include "spellcheck.h"
22162 (struct string_fragment): New struct.
22163 (struct edit_distance_traits<const string_fragment &>): New
22165 (get_closest_sanitizer_option): New function.
22166 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22168 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22170 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22172 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22173 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22174 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22175 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22176 for initial die_offset if dwarf_split_debug_info.
22177 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22178 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22180 (output_skeleton_debug_sections): Formatting fix. Use
22181 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22182 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22184 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22186 * config/arm/cortex-a53.md: Add bypasses for
22187 cortex_a53_r2f_cvt.
22188 (cortex_a53_r2f): Only use for transfers.
22189 (cortex_a53_f2r): Likewise.
22190 (cortex_a53_r2f_cvt): Add reservation for conversions.
22191 (cortex_a53_f2r_cvt): Likewise.
22193 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22195 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22196 to all inlined functions, change static to extern.
22198 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22201 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22203 (arm_assemble_integer): Likewise.
22205 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22207 * config.gcc: Use new awk script to check CPU, FPU and architecture
22208 parameters for --with-... options.
22209 * config/arm/parsecpu.awk: New file
22210 * config/arm/arm-cpus.in: New file.
22211 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22213 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22215 * config/arm/t-arm: Update dependency rules.
22216 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22217 of processing .def files.
22218 * config/arm/genopt.sh: Deleted.
22219 * config/arm/gentune.sh: Deleted.
22220 * config/arm/arm-cores.def: Deleted.
22221 * config/arm/arm-arches.def: Deleted.
22222 * config/arm/arm-fpus.def: Deleted.
22223 * config/arm/arm-tune.md: Regenerated.
22224 * config/arm/arm-tables.opt: Regenerated.
22225 * config/arm/arm-cpu.h: New generated file.
22226 * config/arm/arm-cpu-data.h: New generated file.
22227 * config/arm/arm-cpu-cdata.h: New generated file.
22229 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22232 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22234 (input_varpool_node): Unpack dynamically_initialized bit.
22236 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22238 PR rtl-optimization/79032
22239 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22240 the alignment of the adjusted memory reference against that of MODE,
22241 instead of the alignment of the original memory reference.
22243 2017-01-11 Martin Jambor <mjambor@suse.cz>
22245 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22247 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22248 decorated functions.
22250 2017-01-11 Richard Biener <rguenther@suse.de>
22252 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22253 set range/nonnull info for PHI results. Do not set it on
22254 stmts marked for removal.
22256 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22258 * expr.c (store_field): In the bitfield case, fetch the return value
22259 from the registers before applying a single big-endian adjustment.
22260 Always do a final load for a BLKmode value not larger than a word.
22262 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22265 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22266 that we correctly handle column numbers greater than
22267 LINE_MAP_MAX_COLUMN_NUMBER.
22269 2017-01-10 Martin Sebor <msebor@redhat.com>
22271 PR middle-end/78245
22272 * gimple-ssa-sprintf.c (get_destination_size): Call
22273 {init,fini}object_sizes.
22274 * tree-object-size.c (addr_object_size): Adjust.
22275 (pass_through_call): Adjust.
22276 (pass_object_sizes::execute): Adjust.
22277 * tree-object-size.h (fini_object_sizes): Declare.
22279 2017-01-10 Martin Sebor <msebor@redhat.com>
22281 PR tree-optimization/78775
22282 * builtins.c (get_size_range): Move...
22283 * calls.c: ...to here.
22284 (alloc_max_size): Accept zero argument.
22285 (operand_signed_p): Remove.
22286 (maybe_warn_alloc_args_overflow): Call get_size_range.
22287 * calls.h (get_size_range): Declare.
22289 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22291 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22292 from TI's devices.csv file as of September 2016.
22293 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22295 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22297 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22298 * doc/invoke.texi: Likewise.
22299 * doc/md.texi: Likewise.
22300 * doc/objc.texi: Likewise.
22302 2017-01-10 Joshua Conner <joshconner@google.com>
22304 * config/arm/fuchsia-elf.h: New file.
22305 * config/fuchsia.h: New file.
22306 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22307 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22309 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22311 2016-01-10 Richard Biener <rguenther@suse.de>
22313 PR tree-optimization/79034
22314 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22315 Propagate out degenerate PHIs in the joiner.
22317 2017-01-10 Martin Liska <mliska@suse.cz>
22319 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22320 (sort_congruence_classes_by_decl_uid): Likewise.
22321 (sort_congruence_class_groups_by_decl_uid): Likewise.
22322 (sem_item_optimizer::merge_classes): Sort class, groups in these
22323 classes and members in the groups by DECL_UID of declarations.
22324 This would make merge operations stable.
22326 2017-01-10 Martin Liska <mliska@suse.cz>
22328 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22329 usage of m_classes_vec.
22330 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22331 (sem_item_optimizer::get_group_by_hash): Likewise.
22332 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22333 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22334 (sem_item_optimizer::verify_classes): Likewise.
22335 (sem_item_optimizer::process_cong_reduction): Likewise.
22336 (sem_item_optimizer::dump_cong_classes): Likewise.
22337 (sem_item_optimizer::merge_classes): Likewise.
22338 * ipa-icf.h (congruence_class_hash): Rename from
22339 congruence_class_group_hash. Remove declaration of m_classes_vec.
22341 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22343 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22344 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22345 * config.gcc: Add avx512vpopcntdqintrin.h.
22346 * config/i386/avx512vpopcntdqintrin.h: New.
22347 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22348 * config/i386/i386-builtin-types.def: Add new types.
22349 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22350 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22351 __builtin_ia32_vpopcountq_v8di_mask): New.
22352 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22353 __AVX512VPOPCNTDQ__.
22354 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22355 (PTA_AVX512VPOPCNTDQ): Define.
22356 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22357 TARGET_AVX512VPOPCNTDQ_P): Define.
22358 * config/i386/i386.opt: Add mavx512vpopcntdq.
22359 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22360 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22362 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22364 PR middle-end/77484
22365 * predict.def (PRED_CALL): Set to 67.
22367 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22369 * expr.c (store_field): In the bitfield case, if the value comes from
22370 a function call and is of an aggregate type returned in registers, do
22371 not modify the field mode; extract the value in all cases if the mode
22372 is BLKmode and the size is not larger than a word.
22374 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22377 * config/i386/cpuid.h: Fix undefined behavior.
22379 2017-01-04 Jeff Law <law@redhat.com>
22381 PR tree-optimization/79007
22382 PR tree-optimization/67955
22383 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22384 conservative for pt.null when flag_non_call_exceptions is on.
22386 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22388 PR translation/79019
22389 PR translation/79020
22390 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22391 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22392 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22394 * config/avr/avr.opt (maccumulate-args): Likewise.
22395 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22396 * common.opt (freport-bug): Likewise.
22397 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22398 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22399 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22400 translatable string.
22401 * config/i386/i386.c (function_value_32): Likewise.
22402 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22403 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22405 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22406 * common/config/msp430/msp430-common.c (msp430_handle_option):
22408 * symtab.c (symtab_node::verify_base): Likewise.
22409 * opts.c (set_debug_level): Likewise.
22410 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22411 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22412 missing whitespace to translatable strings.
22413 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22414 * config/sh/superh.h: Likewise.
22415 * config/i386/xopintrin.h: Likewise.
22416 * config/i386/znver1.md: Likewise.
22417 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22418 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22419 * double-int.h (struct double_int): Likewise.
22420 * double-int.c (div_and_round_double): Likewise.
22421 * wide-int.cc: Likewise.
22422 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22423 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22424 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22425 (crossjumps_occurred): ... this.
22426 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22429 PR tree-optimization/78899
22430 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22431 returning bool return struct loop *, NULL for failure and the new
22433 (versionable_outer_loop_p): Don't version outer loop if it has
22434 dont_vectorized bit set.
22435 (tree_if_conversion): When versioning outer loop, ensure
22436 tree_if_conversion is performed also on the inner loop of the
22437 non-vectorizable outer loop copy.
22438 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22439 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22440 prevent vectorization of it.
22441 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22442 the outer loop vectorization of the non-scalar version is attempted
22443 before vectorization of the inner loop in scalar version. If
22444 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22445 vectorization of its inner loop.
22446 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22447 has 2 inner loops, rename also on edges from bb whose single pred
22448 is outer_loop->header. Fix typo in function comment.
22450 2017-01-09 Martin Sebor <msebor@redhat.com>
22453 * asan.c (asan_emit_stack_protection): Increase local buffer size
22454 to avoid snprintf truncation warning.
22456 2017-01-09 Andrew Pinski <apinski@cavium.com>
22458 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22459 to reference thunderx2t99 for the tuning structure
22460 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22462 (thunderx2t99_extra_costs): This.
22463 * config/aarch64/aarch64-tune.md: Regenerate.
22464 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22465 (vulcan_addrcost_table): This.
22466 (vulcan_regmove_cost): Rename to ...
22467 (thunderx2t99_regmove_cost): This.
22468 (vulcan_vector_cost): Rename to ...
22469 (thunderx2t99_vector_cost): this.
22470 (vulcan_branch_cost): Rename to ...
22471 (thunderx2t99_branch_cost): This.
22472 (vulcan_tunings): Rename to ...
22473 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22474 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22476 2017-01-09 Martin Jambor <mjambor@suse.cz>
22480 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22481 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22482 (propagate_vr_accross_jump_function): Use the above function for all
22483 value range computations for pass-through jump functions and type
22484 converasion from explicit value range values.
22485 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22486 parameters from TYPE_ARG_TYPES.
22487 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22488 (ipa_write_node_info): Stream type of the actual argument.
22489 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22491 2017-01-09 Martin Liska <mliska@suse.cz>
22494 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22495 (lookup_compiler): Do not show error message with have_E.
22497 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22499 PR tree-optimization/78938
22500 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22501 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22502 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22503 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22506 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22508 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22511 2017-01-09 Richard Biener <rguenther@suse.de>
22513 PR tree-optimization/78997
22514 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22515 name condition properly.
22517 2017-01-09 Richard Biener <rguenther@suse.de>
22520 * dwarf2out.c (is_cxx): New overload with context.
22521 (is_naming_typedef_decl): Use it.
22523 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22525 * invoke.texi (Option Summary): Correct spacing in option lists
22526 and add line breaks to fix over-long lines.
22528 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22530 PR middle-end/17660
22532 * extend.texi (Common Variable Attributes): Add xref to GCC
22533 Internals manual to explain mode attribute keywords.
22535 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22538 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22539 and Preprocessor Options.
22540 (Options for Linking): Document -pthread here....
22541 (RS/6000 and PowerPC Options): ...not here.
22542 (Solaris 2 Options): ...or here.
22543 * doc/cppopts.texi: Document -pthread.
22545 2017-01-08 Martin Sebor <msebor@redhat.com>
22547 PR middle-end/77708
22548 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22549 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22550 New member functions.
22551 (format_directive): Used them.
22553 (pass_sprintf_length::handle_gimple_call): Same.
22554 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22555 to avoid truncation for any argument.
22556 (extract_affine_mul): Same.
22557 * tree.c (get_file_function_name): Same.
22559 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22561 PR middle-end/77484
22562 * predict.def (PRED_INDIR_CALL): Set to 86.
22564 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22566 PR preprocessor/54124
22567 * doc/cppopts.texi: Reformat -d subtable to list the full name
22568 of the options. Add cross-reference to the docs for the general
22569 compiler -d options.
22570 * doc/invoke.texi (Developer Options): Add cross-reference to the
22571 preprocessor-specific -d option documentation.
22573 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22575 PR preprocessor/13498
22576 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22577 redudant material, and reflect new command-line options.
22578 (System Headers): Likewise.
22580 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22582 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22583 -isystem, and -idirafter. Copy-edit.
22584 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22585 default for -ftrack-macro-expansion. Delete obsolete and
22586 badly-formatted implementation details about -fdebug-cpp output.
22587 * doc/cppwarnopts.texi: Copy-edit.
22589 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22592 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22593 that the transition from a max line width >= 1<<10 to narrower
22594 lines works correctly.
22596 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22598 * doc/options.texi (PerFunction): New.
22599 * opt-functions.awk (switch_flags): Map both Optimization and
22600 PerFunction to CL_OPTIMIZATION.
22601 * opth-gen.awk: Test for PerFunction flag along with
22603 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22604 it only when the latter is present. Skip those that don't in
22605 the hash function generator.
22606 * common.opt (fvar-tracking): Mark as PerFunction instead of
22608 (fvar-tracking-assignments): Likewise.
22609 (fvar-tracking-assignments-toggle): Likewise.
22610 (fvar-tracking-uninit): Likewise.
22612 2017-01-07 Jakub Jelinek <jakub@redhat.com>
22614 PR translation/79018
22615 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
22618 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
22621 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
22622 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
22623 TARGET_LONG_JUMP_TABLE_OFFSETS.
22624 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
22625 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
22626 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
22627 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
22628 * config/m68k/m68k.md (tablejump expander): Likewise.
22629 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
22630 TARGET_LONG_JUMP_TABLE_OFFSETS.
22631 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
22632 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
22634 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
22635 David Holsgrove <david.holsgrove@xilinx.com>
22637 * common/config/microblaze/microblaze-common.c
22638 (TARGET_EXCEPT_UNWIND_INFO): Remove.
22639 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
22641 * config/microblaze/microblaze.c (microblaze_must_save_register)
22642 (microblaze_expand_epilogue, microblaze_return_addr): Handle
22644 (microblaze_eh_return): New function.
22645 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
22646 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
22647 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
22648 * config/microblaze/microblaze.md (eh_return): New pattern.
22650 2017-01-06 Jakub Jelinek <jakub@redhat.com>
22652 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
22653 GCC_DIAGNOSTIC_STRINGIFY): Define.
22655 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
22657 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22659 * config/arm/arm.md (<mcrr>): New.
22661 * config/arm/arm.c (arm_arch5te): New.
22662 (arm_option_override): Set arm_arch5te.
22663 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
22665 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
22666 (arm_mcrr_qualifiers): ... this. New.
22667 (MRRC_QUALIFIERS): Define to...
22668 (arm_mrrc_qualifiers): ... this. New.
22669 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
22671 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
22672 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
22673 (MRRCI, mrrc, MRRC): New.
22674 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
22675 VUNSPEC_MRRC2): New.
22677 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22679 * config/arm/arm.md (<mcr>): New.
22681 * config/arm/arm.c (arm_coproc_builtin_available): Add
22682 support for mcr, mrc, mcr2 and mrc2.
22683 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
22684 (arm_mcr_qualifiers): ... this. New.
22685 (MRC_QUALIFIERS): Define to ...
22686 (arm_mrc_qualifiers): ... this. New.
22687 (MCR_QUALIFIERS): Define to ...
22688 (arm_mcr_qualifiers): ... this. New.
22689 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
22691 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
22692 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
22693 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
22694 VUNSPEC_MRC2): New.
22696 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22698 * config/arm/arm.md (*ldc): New.
22702 * config/arm/arm.c (arm_coproc_builtin_available): Add
22703 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
22704 (arm_coproc_ldc_stc_legitimate_address): New.
22705 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
22706 'qualifier_const_pointer'.
22707 (LDC_QUALIFIERS): Define to...
22708 (arm_ldc_qualifiers): ... this. New.
22709 (STC_QUALIFIERS): Define to...
22710 (arm_stc_qualifiers): ... this. New.
22711 * config/arm/arm-protos.h
22712 (arm_coproc_ldc_stc_legitimate_address): New.
22713 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
22714 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
22715 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
22716 stc2, stcl, stc2l): New.
22717 * config/arm/constraints.md (Uz): New.
22718 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
22719 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
22720 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
22721 VUNSPEC_STC2L): New.
22723 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22725 * config/arm/arm.md (<cdp>): New.
22726 * config/arm/arm.c (neon_const_bounds): Rename this ...
22727 (arm_const_bounds): ... this.
22728 (arm_coproc_builtin_available): New.
22729 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
22730 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
22731 (CDP_QUALIFIERS): Define to...
22732 (arm_cdp_qualifiers): ... this. New.
22734 (arm_expand_builtin_args): Add case for 6 arguments.
22735 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
22736 (arm_const_bounds): ... this.
22737 (arm_coproc_builtin_available): New.
22738 * config/arm/arm_acle.h (__arm_cdp): New.
22740 * config/arm/arm_acle_builtins.def (cdp): New.
22742 * config/arm/iterators.md (CDPI,CDP,cdp): New.
22743 * config/arm/neon.md: Rename all 'neon_const_bounds' to
22744 'arm_const_bounds'.
22745 * config/arm/types.md (coproc): New.
22746 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
22747 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
22748 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
22749 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
22751 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22753 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
22754 (UBINOP_QUALIFIERS): New.
22756 (acle_builtin_data): New. Change comment.
22757 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
22758 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
22759 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
22760 arm_acle_builtins.def.
22761 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
22762 (arm_init_acle_builtins): New.
22763 (CRC32_BUILTIN): Remove.
22764 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
22765 crc32cb, crc32ch and crc32cw.
22766 (arm_init_crc32_builtins): Remove.
22767 (arm_init_builtins): Use arm_init_acle_builtins rather
22768 than arm_init_crc32_builtins.
22769 (arm_expand_acle_builtin): New.
22770 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
22771 * config/arm/arm_acle_builtins.def: New.
22773 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22775 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
22776 (arm_builtin_datum): ... this.
22777 (arm_init_neon_builtin): Rename to ...
22778 (arm_init_builtin): ... this. Add a new parameters PREFIX
22779 and USE_SIG_IN_NAME.
22780 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
22781 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
22782 'arm_builtin_datum'.
22783 (arm_init_vfp_builtins): Likewise.
22784 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
22785 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
22786 (arm_expand_neon_args): Rename to ...
22787 (arm_expand_builtin_args): ... this. Rename builtin_arg
22788 enum values and differentiate between ARG_BUILTIN_MEMORY
22789 and ARG_BUILTIN_NEON_MEMORY.
22790 (arm_expand_neon_builtin_1): Rename to ...
22791 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
22792 values, arm_expand_builtin_args and add bool parameter NEON.
22793 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
22794 (arm_expand_vfp_builtin): Likewise.
22795 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
22797 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22799 PR middle-end/77484
22800 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
22801 * predict.c (tree_estimate_probability_bb): Reverse direction of
22802 polymorphic call predictor.
22804 2017-01-06 David Malcolm <dmalcolm@redhat.com>
22806 * passes.c (execute_one_pass): Split out pass-skipping logic into...
22807 (determine_pass_name_match): ...this new function and...
22808 (should_skip_pass_p): ...this new function.
22810 2017-01-06 Nathan Sidwell <nathan@acm.org>
22812 * ipa-visibility.c (function_and_variable_visibility): Reformat
22813 comments and long lines. Remove extrneous if.
22814 * symtab.c (symtab_node::make_decl_local): Fix code format.
22815 (symtab_node::set_section_for_node): Fix comment typo.
22817 2017-01-06 Martin Liska <mliska@suse.cz>
22820 * lra-constraints.c: Rename invariant to lra_invariant.
22821 * predict.c (set_even_probabilities): Initialize e to NULL.
22823 2017-01-05 Martin Sebor <msebor@redhat.com>
22825 PR tree-optimization/78910
22826 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
22827 (format_integer): Correct off-by-one error in the handling
22828 of precision with negative numbers in signed conversions..
22830 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
22832 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
22834 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22836 PR tree-optimization/71016
22837 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
22838 factor_out_conditional_conversion. Formatting fix.
22839 (factor_out_conditional_conversion): Add cond_stmt argument.
22840 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
22841 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
22844 2017-01-05 David Malcolm <dmalcolm@redhat.com>
22846 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
22847 read-rtl-function.o, and selftest-rtl.o.
22848 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
22849 (selftest::aarch64_test_loading_full_dump): New function.
22850 (selftest::aarch64_run_selftests): New function.
22851 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
22852 selftest::aarch64_run_selftests.
22853 * config/i386/i386.c
22854 (selftest::ix86_test_loading_dump_fragment_1): New function.
22855 (selftest::ix86_test_loading_call_insn): New function.
22856 (selftest::ix86_test_loading_full_dump): New function.
22857 (selftest::ix86_test_loading_unspec): New function.
22858 (selftest::ix86_run_selftests): Call the new functions.
22859 * emit-rtl.c (maybe_set_max_label_num): New function.
22860 * emit-rtl.h (maybe_set_max_label_num): New decl.
22861 * function.c (instantiate_decls): Guard call to
22862 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
22863 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
22865 * gensupport.c (gen_reader::gen_reader): Pass "false"
22866 for new "compact" param of rtx_reader.
22867 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
22868 rather than an empty string for NULL strings.
22869 * read-md.c: Potentially include config.h rather than bconfig.h.
22870 Wrap include of errors.h with #ifdef GENERATOR_FILE.
22871 (have_error): New global, copied from errors.c.
22872 (md_reader::read_name): Rename to...
22873 (md_reader::read_name_1): ...this, adding "out_loc" param,
22874 and converting "missing name or number" to returning false, rather
22876 (md_reader::read_name): Reimplement in terms of read_name_1.
22877 (md_reader::read_name_or_nil): New function.
22878 (md_reader::read_string): Handle "(nil)" by returning NULL.
22879 (md_reader::md_reader): Add new param "compact".
22880 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
22881 (md_reader::read_file): New method.
22882 * read-md.h (md_reader::md_reader): Add new param "compact".
22883 (md_reader::read_file): New method.
22884 (md_reader::is_compact): New accessor.
22885 (md_reader::read_name): Convert return type from void to file_location.
22886 (md_reader::read_name_or_nil): New decl.
22887 (md_reader::read_name_1): New decl.
22888 (md_reader::m_compact): New field.
22889 (noop_reader::noop_reader): Pass "false" for new "compact" param
22891 (rtx_reader::rtx_reader): Add new "compact" param.
22892 (rtx_reader::read_rtx_operand): Make virtual and convert return
22893 type from void to rtx.
22894 (rtx_reader::read_until): New decl.
22895 (rtx_reader::handle_any_trailing_information): New virtual function.
22896 (rtx_reader::postprocess): New virtual function.
22897 (rtx_reader::finalize_string): New virtual function.
22898 (rtx_reader::m_in_call_function_usage): New field.
22899 (rtx_reader::m_reuse_rtx_by_id): New field.
22900 * read-rtl-function.c: New file.
22901 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
22902 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
22903 (selftest::verify_three_block_rtl_cfg): New decl.
22904 * read-rtl-function.h: New file.
22905 * read-rtl.c: Potentially include config.h rather than bconfig.h.
22906 For host, include function.h, memmodel.h, and emit-rtl.h.
22907 (one_time_initialization): New function.
22908 (struct compact_insn_name): New struct.
22909 (compact_insn_names): New array.
22910 (find_code): Handle insn codes in compact dumps.
22911 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
22912 (bind_subst_iter_and_attr): Likewise.
22913 (add_condition_to_string): Likewise.
22914 (add_condition_to_rtx): Likewise.
22915 (apply_attribute_uses): Likewise.
22916 (add_current_iterators): Likewise.
22917 (apply_iterators): Likewise.
22918 (initialize_iterators): Guard usage of apply_subst_iterator with
22919 #ifdef GENERATOR_FILE.
22920 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
22921 (md_reader::read_mapping): Likewise.
22922 (add_define_attr_for_define_subst): Likewise.
22923 (add_define_subst_attr): Likewise.
22924 (read_subst_mapping): Likewise.
22925 (check_code_iterator): Likewise.
22926 (rtx_reader::read_rtx): Likewise. Move one-time initialization
22928 (one_time_initialization): New function.
22929 (rtx_reader::read_until): New method.
22930 (read_flags): New function.
22931 (parse_reg_note_name): New function.
22932 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
22933 Handle reuse_rtx ids.
22934 Wrap iterator lookup within #ifdef GENERATOR_FILE.
22935 Add parsing support for RTL dumps, mirroring the special-cases in
22936 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
22937 values, and calling handle_any_trailing_information.
22938 (rtx_reader::read_rtx_operand): Convert return type from void
22939 to rtx, returning return_rtx. Handle case 'e'. Call
22940 finalize_string on XSTR and XTMPL fields.
22941 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
22942 "(nil)" values were omitted. Call the postprocess vfunc on the
22944 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
22945 class ctor. Initialize m_in_call_function_usage. Call
22946 one_time_initialization.
22947 * rtl-tests.c (selftest::test_uncond_jump): Call
22948 set_new_first_and_last_insn.
22949 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
22950 * selftest-rtl.c: New file.
22951 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
22952 (selftest::get_insn_by_uid): New decl.
22953 * selftest-run-tests.c (selftest::run_tests): Call
22954 read_rtl_function_c_tests.
22955 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
22956 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
22959 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
22961 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
22962 operands in a special way. Assert that pos+len <= mode precision.
22964 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22966 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
22967 3 argument Alias with unlimited for the negative form.
22968 (fno-vect-cost-model): Removed.
22970 2017-01-05 Martin Liska <mliska@suse.cz>
22972 * hsa-gen.c (gen_hsa_divmod): New function.
22973 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
22975 2017-01-05 Martin Liska <mliska@suse.cz>
22978 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
22981 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22983 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
22984 small constant length operands.
22986 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22988 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
22989 between loop iterations.
22991 2017-01-05 Martin Liska <mliska@suse.cz>
22994 * gimplify.c (gimplify_decl_expr): Compare to
22995 asan_poisoned_variables instread of checking flags.
22996 (gimplify_target_expr): Likewise.
22997 (gimplify_expr): Likewise.
22998 (gimplify_function_tree): Conditionally initialize
22999 asan_poisoned_variables.
23001 2017-01-04 Jeff Law <law@redhat.com>
23003 PR tree-optimizatin/78812
23004 * rtl.h (contains_mem_rtx_p): Prototype.
23005 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23006 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23007 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23008 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23009 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23011 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23013 * input.c (assert_char_at_range): Default-initialize actual_range.
23015 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23017 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23018 to match the caller.
23020 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23022 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23023 insns after final jump in test to emit dummy move.
23025 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23027 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23028 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23030 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23032 * multiple_target.c (create_dispatcher_calls): Init e_next.
23033 * tree-ssa-loop-split.c (split_loop): Init border.
23034 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23037 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23042 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23043 (altivec_register_operand): Do not return true if the operand
23044 contains a SUBREG mixing SImode and SFmode.
23045 (vsx_register_operand): Likewise.
23046 (vsx_reg_sfsubreg_ok): New predicate.
23047 (vfloat_operand): Do not return true if the operand contains a
23048 SUBREG mixing SImode and SFmode.
23049 (vint_operand): Likewise.
23050 (vlogical_operand): Likewise.
23051 (gpc_reg_operand): Likewise.
23052 (int_reg_operand): Likewise.
23053 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23054 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23055 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23057 (rs6000_emit_move_si_sf_subreg): New helper function.
23058 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23059 fixup SUBREGs involving SImode and SFmode.
23060 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23061 numbers for the new peephole2 optimization.
23062 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23063 the GLIBC math library that do AND/IOR/XOR operations on single
23064 precision floating point.
23065 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23066 target macros to say whether we need to avoid SUBREGs mixing
23068 (TARGET_ALLOW_SF_SUBREG): Likewise.
23069 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23070 (UNSPEC_SI_FROM_SF): Likewise.
23071 (iorxor): Change spacing.
23072 (and_ior_xor): New iterator for AND, IOR, and XOR.
23073 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23074 (movdi_from_sf_zero_ext): Likewise.
23075 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23076 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23077 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23078 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23079 (fms<mode>4): Likewise.
23080 (fnma<mode>4): Likewise.
23081 (fnms<mode>4): Likewise.
23082 (nfma<mode>4): Likewise.
23083 (nfms<mode>4): Likewise.
23085 2017-01-04 Marek Polacek <polacek@redhat.com>
23088 * doc/invoke.texi: Document -Wpointer-compare.
23090 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23092 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23095 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23096 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23099 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23102 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23103 documentation of the powerpc_popcntb_ok attribute.
23104 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23105 code to issue warning messages if a requested CPU configuration is
23106 not supported by the binary (assembler and loader) toolchain.
23107 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23108 made to define a built-in function that has been disabled.
23109 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23110 made to define a built-in function that has been disabled.
23111 (altivec_init_builtins): Add comment explaining why definition
23112 of the DST built-in functions is not preceded by an assertion
23113 check. Add assertions to prevent ICE if attempts are made to
23114 define an altivec predicate or an abs* built-in function that has
23116 (htm_init_builtins): Add comment explaining why definition of the
23117 htm built-in functions is not preceded by an assertion check.
23119 2017-01-04 Jeff Law <law@redhat.com>
23121 PR tree-optimizatin/67955
23122 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23123 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23124 the points-to solution does not include pt_null. Use DECL_PT_UID
23127 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23129 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23130 Use gen_int_mode instead of gen_lopwart for const_int operands.
23132 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23134 PR tree-optimization/71563
23135 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23136 out of range value - has low bits known to be zero.
23138 2017-01-04 Alan Modra <amodra@gmail.com>
23140 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23141 * configure: Regenerate.
23142 * config.in: Regenerate.
23144 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23147 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23148 a substring of the message, but strcmp with the whole message. Ifdef
23149 ENABLE_NLS, translate the message first using dgettext.
23151 2017-01-03 Jeff Law <law@redhat.com>
23153 PR tree-optimizatin/78856
23154 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23155 (mark_threaded_blocks): Remove code to truncate thread paths that
23156 cross multiple loop headers. Instead invalidate the cached loop
23157 iteration information and handle case of a thread path walking
23158 into an irreducible region.
23160 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23163 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23164 assertions. Add support for doing the signbit if the IEEE 128-bit
23165 floating point value is in a GPR.
23166 * config/rs6000/rs6000.md (Fsignbit): Delete.
23167 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23168 Update the length attribute if the value is in a GPR.
23169 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23170 the sign or zero extension instruction, since the value is always 0/1.
23171 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23174 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23175 extracting SImode to a GPR register so that we can generate a
23176 store, limit the vector to be in a traditional Altivec register
23177 for the vextuwrx instruction.
23179 2017-01-03 Ian Lance Taylor <iant@google.com>
23181 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23183 2017-01-03 Martin Sebor <msebor@redhat.com>
23185 PR tree-optimization/78696
23186 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23187 precision. Use MPFR for %f for greater fidelity. Correct handling
23189 (pass_sprintf_length::compute_format_length): Set width and precision
23190 specified by asrerisk to void_node for vararg functions.
23191 (try_substitute_return_value): Adjust dump output.
23193 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23195 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23197 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23199 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23200 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23201 -mlra/-mno-lra was passed to the compiler.
23203 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23205 PR rtl-optimization/65618
23206 * emit-rtl.c (try_split): Move initialization of "before" and
23207 "after" to just before the call to emit_insn_after_setloc.
23209 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23211 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23213 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23215 * dwarf2out.c (gen_enumeration_type_die): When
23216 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23218 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23220 PR tree-optimization/78965
23221 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23222 Change first argument from const call_info & to call_info &. For %n
23223 set info.nowrite to false.
23225 PR middle-end/78901
23226 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23227 possibly throwing calls.
23229 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23230 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23231 and fns handling, rather than in a separate case SSA_NAME.
23233 2017-01-02 Jeff Law <law@redhat.com>
23235 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23236 fixes for first_period and second_period variables.
23238 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23241 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23242 (*insvqi_1): New insn pattern.
23243 (*insvqi_1_mem_rex64): Ditto.
23244 (*insvqi_2): Ditto.
23245 (*insvqi_3): Rename from *insvqi.
23247 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23249 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23251 * doc/cfg.texi (Edges): Remove reference to Java.
23252 (Maintaining the CFG): Ditto.
23254 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23256 PR middle-end/77674
23257 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23258 transparent aliases.
23260 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23262 PR middle-end/77484
23263 * predict.def (PRED_CALL): Update hitrate.
23264 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23265 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23266 into direct/indirect/polymorphic variants.
23268 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23270 Update copyright years.
23272 * gcc.c (process_command): Update copyright notice dates.
23273 * gcov-dump.c (print_version): Ditto.
23274 * gcov.c (print_version): Ditto.
23275 * gcov-tool.c (print_version): Ditto.
23276 * gengtype.c (create_file): Ditto.
23277 * doc/cpp.texi: Bump @copying's copyright year.
23278 * doc/cppinternals.texi: Ditto.
23279 * doc/gcc.texi: Ditto.
23280 * doc/gccint.texi: Ditto.
23281 * doc/gcov.texi: Ditto.
23282 * doc/install.texi: Ditto.
23283 * doc/invoke.texi: Ditto.
23285 Copyright (C) 2017 Free Software Foundation, Inc.
23287 Copying and distribution of this file, with or without modification,
23288 are permitted in any medium without royalty provided the copyright
23289 notice and this notice are preserved.